mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-26 04:32:44 +01:00
c4a674e21e
the compilation time and there is no case for which we see any improvement in performance. This patch removes this pass and its associated test cases from the tree. Differential Revision: https://reviews.llvm.org/D101313 Change-Id: I0599169a7609c19a887f8d847a71e664030cc141
3528 lines
163 KiB
LLVM
3528 lines
163 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx700 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX7 %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx803 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX8 %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX9 %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx906 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX9-DL %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1011 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10-DL-XNACK %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1012 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10-DL-XNACK %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1030 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10-DL-NOXNACK %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1031 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10-DL-NOXNACK %s
|
|
|
|
define amdgpu_kernel void @idot8_acc32(<8 x i4> addrspace(1)* %src1,
|
|
; GFX7-LABEL: idot8_acc32:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; GFX7-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX7-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX7-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x9
|
|
; GFX7-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xd
|
|
; GFX7-NEXT: s_mov_b32 s14, -1
|
|
; GFX7-NEXT: s_mov_b32 s15, 0xe8f000
|
|
; GFX7-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX7-NEXT: s_mov_b32 s3, 0xf000
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX7-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX7-NEXT: buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[6:7]
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_load_dword s4, s[0:1], 0x0
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_i32 v1, v2, 0, 4
|
|
; GFX7-NEXT: v_bfe_i32 v3, v2, 4, 4
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_bfe_i32 v9, v0, 0, 4
|
|
; GFX7-NEXT: v_bfe_i32 v10, v0, 4, 4
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v1, v9, s4
|
|
; GFX7-NEXT: v_bfe_i32 v4, v2, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v11, v0, 8, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v3, v10, v1
|
|
; GFX7-NEXT: v_bfe_i32 v5, v2, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v12, v0, 12, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v4, v11, v1
|
|
; GFX7-NEXT: v_bfe_i32 v6, v2, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v13, v0, 16, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v5, v12, v1
|
|
; GFX7-NEXT: v_bfe_i32 v7, v2, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v14, v0, 20, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v6, v13, v1
|
|
; GFX7-NEXT: v_bfe_i32 v8, v2, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v15, v0, 24, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v7, v14, v1
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v2, 28, v2
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v0, 28, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v8, v15, v1
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: idot8_acc32:
|
|
; GFX8: ; %bb.0: ; %entry
|
|
; GFX8-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX8-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
; GFX8-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX8-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s5
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s4, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v3, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX8-NEXT: s_mov_b32 s10, -1
|
|
; GFX8-NEXT: s_mov_b32 s11, 0xe80000
|
|
; GFX8-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX8-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_bfe_i32 v1, v3, 0, 4
|
|
; GFX8-NEXT: v_bfe_i32 v4, v3, 4, 4
|
|
; GFX8-NEXT: v_bfe_i32 v6, v3, 8, 4
|
|
; GFX8-NEXT: v_bfe_i32 v8, v3, 12, 4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_bfe_i32 v2, v0, 0, 4
|
|
; GFX8-NEXT: v_bfe_i32 v5, v0, 4, 4
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v1, v2, s2
|
|
; GFX8-NEXT: v_bfe_i32 v7, v0, 8, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v4, v5, v1
|
|
; GFX8-NEXT: v_bfe_i32 v9, v0, 12, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v6, v7, v1
|
|
; GFX8-NEXT: v_bfe_i32 v10, v3, 16, 4
|
|
; GFX8-NEXT: v_bfe_i32 v11, v0, 16, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v8, v9, v1
|
|
; GFX8-NEXT: v_bfe_i32 v12, v3, 20, 4
|
|
; GFX8-NEXT: v_bfe_i32 v13, v0, 20, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v10, v11, v1
|
|
; GFX8-NEXT: v_bfe_i32 v14, v3, 24, 4
|
|
; GFX8-NEXT: v_bfe_i32 v15, v0, 24, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v12, v13, v1
|
|
; GFX8-NEXT: v_ashrrev_i32_e32 v3, 28, v3
|
|
; GFX8-NEXT: v_ashrrev_i32_e32 v0, 28, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v14, v15, v1
|
|
; GFX8-NEXT: v_mad_i32_i24 v2, v3, v0, v1
|
|
; GFX8-NEXT: v_mov_b32_e32 v0, s0
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s1
|
|
; GFX8-NEXT: flat_store_dword v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-LABEL: idot8_acc32:
|
|
; GFX9: ; %bb.0: ; %entry
|
|
; GFX9-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NEXT: v_bfe_i32 v3, v1, 0, 4
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_bfe_i32 v4, v2, 0, 4
|
|
; GFX9-NEXT: v_bfe_i32 v5, v1, 4, 4
|
|
; GFX9-NEXT: v_bfe_i32 v6, v2, 4, 4
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v3, v3, v4
|
|
; GFX9-NEXT: v_bfe_i32 v7, v1, 8, 4
|
|
; GFX9-NEXT: v_bfe_i32 v8, v2, 8, 4
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v4, v5, v6
|
|
; GFX9-NEXT: v_bfe_i32 v9, v1, 12, 4
|
|
; GFX9-NEXT: v_bfe_i32 v10, v2, 12, 4
|
|
; GFX9-NEXT: v_bfe_i32 v11, v1, 16, 4
|
|
; GFX9-NEXT: v_bfe_i32 v12, v2, 16, 4
|
|
; GFX9-NEXT: v_bfe_i32 v13, v1, 20, 4
|
|
; GFX9-NEXT: v_bfe_i32 v15, v1, 24, 4
|
|
; GFX9-NEXT: v_bfe_i32 v14, v2, 20, 4
|
|
; GFX9-NEXT: v_bfe_i32 v16, v2, 24, 4
|
|
; GFX9-NEXT: v_ashrrev_i32_e32 v1, 28, v1
|
|
; GFX9-NEXT: v_ashrrev_i32_e32 v2, 28, v2
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v5, v7, v8
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v6, v9, v10
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v1, v1, v2
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: v_add3_u32 v2, v3, s0, v4
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v7, v11, v12
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v8, v13, v14
|
|
; GFX9-NEXT: v_add3_u32 v2, v2, v5, v6
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v9, v15, v16
|
|
; GFX9-NEXT: v_add3_u32 v2, v2, v7, v8
|
|
; GFX9-NEXT: v_add3_u32 v1, v2, v9, v1
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: idot8_acc32:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; GFX9-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-DL-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-DL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-DL-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-DL-NEXT: v_dot8_i32_i4 v0, v2, v3, s0
|
|
; GFX9-DL-NEXT: global_store_dword v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-XNACK-LABEL: idot8_acc32:
|
|
; GFX10-DL-XNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-XNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-XNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_dot8_i32_i4 v1, v1, v2, s2
|
|
; GFX10-DL-XNACK-NEXT: global_store_dword v0, v1, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-NOXNACK-LABEL: idot8_acc32:
|
|
; GFX10-DL-NOXNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NOXNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NOXNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v0, v0, s[6:7]
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_dot8_i32_i4 v0, v1, v0, s2
|
|
; GFX10-DL-NOXNACK-NEXT: global_store_dword v2, v0, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_endpgm
|
|
; GFX10-DL-LABEL: idot8_acc32:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX10-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_load_dword s6, s[4:5], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s0, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, s6
|
|
; GFX10-DL-NEXT: v_dot8_i32_i4 v0, s0, s1, v0
|
|
; GFX10-DL-NEXT: global_store_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<8 x i4> addrspace(1)* %src2,
|
|
i32 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <8 x i4> %vec1, i64 0
|
|
%cv1e0 = sext i4 %v1e0 to i32
|
|
%v2e0 = extractelement <8 x i4> %vec2, i64 0
|
|
%cv2e0 = sext i4 %v2e0 to i32
|
|
%mul0 = mul nuw nsw i32 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <8 x i4> %vec1, i64 1
|
|
%cv1e1 = sext i4 %v1e1 to i32
|
|
%v2e1 = extractelement <8 x i4> %vec2, i64 1
|
|
%cv2e1 = sext i4 %v2e1 to i32
|
|
%mul1 = mul nuw nsw i32 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <8 x i4> %vec1, i64 2
|
|
%cv1e2 = sext i4 %v1e2 to i32
|
|
%v2e2 = extractelement <8 x i4> %vec2, i64 2
|
|
%cv2e2 = sext i4 %v2e2 to i32
|
|
%mul2 = mul nuw nsw i32 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <8 x i4> %vec1, i64 3
|
|
%cv1e3 = sext i4 %v1e3 to i32
|
|
%v2e3 = extractelement <8 x i4> %vec2, i64 3
|
|
%cv2e3 = sext i4 %v2e3 to i32
|
|
%mul3 = mul nuw nsw i32 %cv1e3, %cv2e3
|
|
|
|
%v1e4 = extractelement <8 x i4> %vec1, i64 4
|
|
%cv1e4 = sext i4 %v1e4 to i32
|
|
%v2e4 = extractelement <8 x i4> %vec2, i64 4
|
|
%cv2e4 = sext i4 %v2e4 to i32
|
|
%mul4 = mul nuw nsw i32 %cv1e4, %cv2e4
|
|
|
|
%v1e5 = extractelement <8 x i4> %vec1, i64 5
|
|
%cv1e5 = sext i4 %v1e5 to i32
|
|
%v2e5 = extractelement <8 x i4> %vec2, i64 5
|
|
%cv2e5 = sext i4 %v2e5 to i32
|
|
%mul5 = mul nuw nsw i32 %cv1e5, %cv2e5
|
|
|
|
%v1e6 = extractelement <8 x i4> %vec1, i64 6
|
|
%cv1e6 = sext i4 %v1e6 to i32
|
|
%v2e6 = extractelement <8 x i4> %vec2, i64 6
|
|
%cv2e6 = sext i4 %v2e6 to i32
|
|
%mul6 = mul nuw nsw i32 %cv1e6, %cv2e6
|
|
|
|
%v1e7 = extractelement <8 x i4> %vec1, i64 7
|
|
%cv1e7 = sext i4 %v1e7 to i32
|
|
%v2e7 = extractelement <8 x i4> %vec2, i64 7
|
|
%cv2e7 = sext i4 %v2e7 to i32
|
|
%mul7 = mul nuw nsw i32 %cv1e7, %cv2e7
|
|
|
|
%acc = load i32, i32 addrspace(1)* %dst, align 4
|
|
%add1 = add i32 %mul0, %acc
|
|
%add2 = add i32 %add1, %mul1
|
|
%add3 = add i32 %add2, %mul2
|
|
%add4 = add i32 %add3, %mul3
|
|
%add5 = add i32 %add4, %mul4
|
|
%add6 = add i32 %add5, %mul5
|
|
%add7 = add i32 %add6, %mul6
|
|
%add8 = add i32 %add7, %mul7
|
|
|
|
store i32 %add8, i32 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; TODO: Once the unnecessary zero extentions of the elements are removed;
|
|
; pattern recognizer will kick in.
|
|
define amdgpu_kernel void @idot8_acc16(<8 x i4> addrspace(1)* %src1,
|
|
; GFX7-LABEL: idot8_acc16:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; GFX7-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX7-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX7-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x9
|
|
; GFX7-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xd
|
|
; GFX7-NEXT: s_mov_b32 s14, -1
|
|
; GFX7-NEXT: s_mov_b32 s15, 0xe8f000
|
|
; GFX7-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX7-NEXT: s_mov_b32 s3, 0xf000
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX7-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX7-NEXT: buffer_load_dword v3, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[6:7]
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: v_mov_b32_e32 v2, 0xffff
|
|
; GFX7-NEXT: s_mov_b32 s4, 0xffff
|
|
; GFX7-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_i32 v1, v3, 0, 4
|
|
; GFX7-NEXT: v_bfe_i32 v4, v3, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v5, v3, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v6, v3, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v7, v3, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v8, v3, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v9, v3, 24, 4
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v3, 28, v3
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_bfe_i32 v10, v0, 0, 4
|
|
; GFX7-NEXT: v_bfe_i32 v11, v0, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v12, v0, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v13, v0, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v14, v0, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v15, v0, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v16, v0, 24, 4
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v0, 28, v0
|
|
; GFX7-NEXT: v_and_b32_e32 v9, v2, v9
|
|
; GFX7-NEXT: v_and_b32_e32 v3, v2, v3
|
|
; GFX7-NEXT: v_and_b32_e32 v15, v2, v15
|
|
; GFX7-NEXT: v_and_b32_e32 v16, v2, v16
|
|
; GFX7-NEXT: v_and_b32_e32 v0, v2, v0
|
|
; GFX7-NEXT: buffer_load_ushort v2, off, s[0:3], 0
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v1
|
|
; GFX7-NEXT: v_and_b32_e32 v10, s4, v10
|
|
; GFX7-NEXT: v_and_b32_e32 v4, s4, v4
|
|
; GFX7-NEXT: v_and_b32_e32 v11, s4, v11
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v5
|
|
; GFX7-NEXT: v_and_b32_e32 v12, s4, v12
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s4, v6
|
|
; GFX7-NEXT: v_and_b32_e32 v13, s4, v13
|
|
; GFX7-NEXT: v_and_b32_e32 v7, s4, v7
|
|
; GFX7-NEXT: v_and_b32_e32 v14, s4, v14
|
|
; GFX7-NEXT: v_and_b32_e32 v8, s4, v8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v10, v2
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v11, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v5, v12, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v6, v13, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v7, v14, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v8, v15, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v9, v16, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v3, v0, v1
|
|
; GFX7-NEXT: buffer_store_short v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: idot8_acc16:
|
|
; GFX8: ; %bb.0: ; %entry
|
|
; GFX8-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX8-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
; GFX8-NEXT: v_mov_b32_e32 v5, 12
|
|
; GFX8-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s5
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s4, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v3, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v2, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v0, s0
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s1
|
|
; GFX8-NEXT: flat_load_ushort v4, v[0:1]
|
|
; GFX8-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX8-NEXT: s_mov_b32 s10, -1
|
|
; GFX8-NEXT: s_mov_b32 s11, 0xe80000
|
|
; GFX8-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX8-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v10, 4, v3
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v16, 12, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 28, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 20, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v15, 4, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v17, 12, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 12, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 8, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v11, 28, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v12, 20, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v13, 12, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v14, 8, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v10, 12, v10
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v15, 12, v15
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v18, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v3, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v19, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v2, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v5, 12, v16
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v16, 12, v17
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v10, 12, v10
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v15, 12, v15
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v4, v5, v16, v4
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v14, 12, v14
|
|
; GFX8-NEXT: v_mad_u16 v4, v10, v15, v4
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_mad_u16 v4, v9, v14, v4
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v18, 12, v19
|
|
; GFX8-NEXT: v_mad_u16 v4, v8, v13, v4
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_mad_u16 v4, v17, v18, v4
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v3, 12, v3
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX8-NEXT: v_mad_u16 v4, v7, v12, v4
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX8-NEXT: v_mad_u16 v2, v3, v2, v4
|
|
; GFX8-NEXT: v_mad_u16 v2, v6, v11, v2
|
|
; GFX8-NEXT: flat_store_short v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-LABEL: idot8_acc16:
|
|
; GFX9: ; %bb.0: ; %entry
|
|
; GFX9-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NEXT: v_mov_b32_e32 v4, 12
|
|
; GFX9-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NEXT: global_load_ushort v3, v0, s[2:3]
|
|
; GFX9-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v9, 4, v1
|
|
; GFX9-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v14, 4, v2
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v15, 12, v1
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v16, 12, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v5, 28, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v7, 12, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v8, 8, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v10, 28, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v11, 20, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v12, 12, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v13, 8, v2
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v17, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v1, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v18, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v4, 12, v15
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v15, 12, v16
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v14, 12, v14
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v4, v15, v3
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v9, v14, v3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v8, v13, v3
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v16, 12, v17
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v7, v12, v3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v16, v17, v3
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v5, 12, v5
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v10, 12, v10
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v1, 12, v1
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v6, v11, v3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v10, 12, v10
|
|
; GFX9-NEXT: v_mad_legacy_u16 v1, v1, v2, v3
|
|
; GFX9-NEXT: v_mad_legacy_u16 v1, v5, v10, v1
|
|
; GFX9-NEXT: global_store_short v0, v1, s[2:3]
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: idot8_acc16:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; GFX9-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-DL-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v4, 12
|
|
; GFX9-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-DL-NEXT: global_load_ushort v3, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v9, 4, v1
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v14, 4, v2
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v15, 12, v1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v16, 12, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 28, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 12, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v8, 8, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v10, 28, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v11, 20, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v12, 12, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v13, 8, v2
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v17, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v1, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v18, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v4, 12, v15
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v15, 12, v16
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v14, 12, v14
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v4, v15, v3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v9, v14, v3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v8, v13, v3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v16, 12, v17
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v7, v12, v3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v16, v17, v3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v5, 12, v5
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v10, 12, v10
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v1, 12, v1
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v6, v11, v3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v10, 12, v10
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v1, v1, v2, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v1, v5, v10, v1
|
|
; GFX9-DL-NEXT: global_store_short v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-XNACK-LABEL: idot8_acc16:
|
|
; GFX10-DL-XNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-XNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-XNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-XNACK-NEXT: global_load_ushort v3, v0, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v4, 28, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v5, 24, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v7, 16, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v8, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v10, 4, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v1, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v16, 4, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v17, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v11, 28, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v10, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v12, 24, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v16, 12, v16
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v13, 20, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v14, 16, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v15, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v2, 8, v2
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v1, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v17, 12, v17
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v9, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v10, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v2, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v16, 12, v16
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v1, v17, v3
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v8, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v3, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v9, 12, v15
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v2, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v10, v16, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v7, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v10, 12, v14
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v8, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v9, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v3, v2, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v6, 12, v6
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v2, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v7, 12, v13
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v3, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v8, v9, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v5, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v8, 12, v12
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v6, 12, v6
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v7, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v2, v3, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v4, 12, v4
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v2, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v5, 12, v11
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v3, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v6, v7, v1
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v4, 12, v4
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v5, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v2, v3, v1
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v4, v5, v1
|
|
; GFX10-DL-XNACK-NEXT: global_store_short v0, v1, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-NOXNACK-LABEL: idot8_acc16:
|
|
; GFX10-DL-NOXNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NOXNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NOXNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v0, v0, s[6:7]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_ushort v3, v2, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v4, 28, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v5, 24, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v7, 16, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v8, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v10, 4, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v1, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v16, 4, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v17, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v11, 28, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v10, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v12, 24, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v16, 12, v16
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v13, 20, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v14, 16, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v15, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v0, 8, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v1, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v17, 12, v17
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v9, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v10, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v0, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v16, 12, v16
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v1, v1, v17, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v8, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v3, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v9, 12, v15
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v0, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v1, v10, v16, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v7, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v10, 12, v14
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v8, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v9, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v3, v0, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v6, 12, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v1, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v7, 12, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v3, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v8, v9, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v5, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v8, 12, v12
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v6, 12, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v7, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v1, v3, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v4, 12, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v1, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v5, 12, v11
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v3, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v6, v7, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v4, 12, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v5, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v1, v3, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v4, v5, v0
|
|
; GFX10-DL-NOXNACK-NEXT: global_store_short v2, v0, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_endpgm
|
|
; GFX10-DL-LABEL: idot8_acc16:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NEXT: s_mov_b32 s14, -1
|
|
; GFX10-DL-NEXT: s_mov_b32 s15, 0x31c16000
|
|
; GFX10-DL-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: global_load_ushort v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: s_load_dword s0, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_lshr_b32 s2, s0, 12
|
|
; GFX10-DL-NEXT: s_lshr_b32 s3, s1, 12
|
|
; GFX10-DL-NEXT: s_bfe_i32 s6, s0, 0x40000
|
|
; GFX10-DL-NEXT: s_bfe_i32 s7, s1, 0x40000
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v2, 12, s2
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v3, 12, s3
|
|
; GFX10-DL-NEXT: s_bfe_i32 s8, s0, 0x40004
|
|
; GFX10-DL-NEXT: s_bfe_i32 s9, s0, 0x40008
|
|
; GFX10-DL-NEXT: s_bfe_i32 s10, s1, 0x40008
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s1, 0x40004
|
|
; GFX10-DL-NEXT: v_mul_i32_i24_e64 v4, s9, s10
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v2, 12, v2
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v3, 12, v3
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40010
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s6, s7, v1
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s8, s2, v1
|
|
; GFX10-DL-NEXT: s_mov_b32 s2, 0xffff
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v2, s2, v2
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v3, s2, v3
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40010
|
|
; GFX10-DL-NEXT: v_mad_u32_u24 v1, v2, v3, v1
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40014
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40014
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40018
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40018
|
|
; GFX10-DL-NEXT: s_ashr_i32 s0, s0, 28
|
|
; GFX10-DL-NEXT: s_ashr_i32 s1, s1, 28
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s0, s1, v1
|
|
; GFX10-DL-NEXT: global_store_short v0, v1, s[4:5]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<8 x i4> addrspace(1)* %src2,
|
|
i16 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <8 x i4> %vec1, i64 0
|
|
%cv1e0 = sext i4 %v1e0 to i16
|
|
%v2e0 = extractelement <8 x i4> %vec2, i64 0
|
|
%cv2e0 = sext i4 %v2e0 to i16
|
|
%mul0 = mul nuw nsw i16 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <8 x i4> %vec1, i64 1
|
|
%cv1e1 = sext i4 %v1e1 to i16
|
|
%v2e1 = extractelement <8 x i4> %vec2, i64 1
|
|
%cv2e1 = sext i4 %v2e1 to i16
|
|
%mul1 = mul nuw nsw i16 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <8 x i4> %vec1, i64 2
|
|
%cv1e2 = sext i4 %v1e2 to i16
|
|
%v2e2 = extractelement <8 x i4> %vec2, i64 2
|
|
%cv2e2 = sext i4 %v2e2 to i16
|
|
%mul2 = mul nuw nsw i16 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <8 x i4> %vec1, i64 3
|
|
%cv1e3 = sext i4 %v1e3 to i16
|
|
%v2e3 = extractelement <8 x i4> %vec2, i64 3
|
|
%cv2e3 = sext i4 %v2e3 to i16
|
|
%mul3 = mul nuw nsw i16 %cv1e3, %cv2e3
|
|
|
|
%v1e4 = extractelement <8 x i4> %vec1, i64 4
|
|
%cv1e4 = sext i4 %v1e4 to i16
|
|
%v2e4 = extractelement <8 x i4> %vec2, i64 4
|
|
%cv2e4 = sext i4 %v2e4 to i16
|
|
%mul4 = mul nuw nsw i16 %cv1e4, %cv2e4
|
|
|
|
%v1e5 = extractelement <8 x i4> %vec1, i64 5
|
|
%cv1e5 = sext i4 %v1e5 to i16
|
|
%v2e5 = extractelement <8 x i4> %vec2, i64 5
|
|
%cv2e5 = sext i4 %v2e5 to i16
|
|
%mul5 = mul nuw nsw i16 %cv1e5, %cv2e5
|
|
|
|
%v1e6 = extractelement <8 x i4> %vec1, i64 6
|
|
%cv1e6 = sext i4 %v1e6 to i16
|
|
%v2e6 = extractelement <8 x i4> %vec2, i64 6
|
|
%cv2e6 = sext i4 %v2e6 to i16
|
|
%mul6 = mul nuw nsw i16 %cv1e6, %cv2e6
|
|
|
|
%v1e7 = extractelement <8 x i4> %vec1, i64 7
|
|
%cv1e7 = sext i4 %v1e7 to i16
|
|
%v2e7 = extractelement <8 x i4> %vec2, i64 7
|
|
%cv2e7 = sext i4 %v2e7 to i16
|
|
%mul7 = mul nuw nsw i16 %cv1e7, %cv2e7
|
|
|
|
%acc = load i16, i16 addrspace(1)* %dst, align 4
|
|
%add1 = add i16 %mul0, %acc
|
|
%add2 = add i16 %add1, %mul1
|
|
%add3 = add i16 %add2, %mul2
|
|
%add4 = add i16 %add3, %mul3
|
|
%add5 = add i16 %add4, %mul4
|
|
%add6 = add i16 %add5, %mul5
|
|
%add7 = add i16 %add6, %mul6
|
|
%add8 = add i16 %add7, %mul7
|
|
|
|
store i16 %add8, i16 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; TODO: Support this pattern.
|
|
define amdgpu_kernel void @idot8_acc8(<8 x i4> addrspace(1)* %src1,
|
|
; GFX7-LABEL: idot8_acc8:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; GFX7-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX7-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX7-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x9
|
|
; GFX7-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xd
|
|
; GFX7-NEXT: s_mov_b32 s14, -1
|
|
; GFX7-NEXT: s_mov_b32 s15, 0xe8f000
|
|
; GFX7-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX7-NEXT: s_mov_b32 s3, 0xf000
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX7-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX7-NEXT: buffer_load_dword v3, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[6:7]
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: v_mov_b32_e32 v2, 0xff
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_i32 v1, v3, 0, 4
|
|
; GFX7-NEXT: v_bfe_i32 v4, v3, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v5, v3, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v6, v3, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v7, v3, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v8, v3, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v9, v3, 24, 4
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v3, 28, v3
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_bfe_i32 v10, v0, 0, 4
|
|
; GFX7-NEXT: v_bfe_i32 v11, v0, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v12, v0, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v13, v0, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v14, v0, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v15, v0, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v16, v0, 24, 4
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v0, 28, v0
|
|
; GFX7-NEXT: v_and_b32_e32 v9, v2, v9
|
|
; GFX7-NEXT: v_and_b32_e32 v3, v2, v3
|
|
; GFX7-NEXT: v_and_b32_e32 v15, v2, v15
|
|
; GFX7-NEXT: v_and_b32_e32 v16, v2, v16
|
|
; GFX7-NEXT: v_and_b32_e32 v0, v2, v0
|
|
; GFX7-NEXT: buffer_load_ubyte v2, off, s[0:3], 0
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v1
|
|
; GFX7-NEXT: v_and_b32_e32 v10, s4, v10
|
|
; GFX7-NEXT: v_and_b32_e32 v4, s4, v4
|
|
; GFX7-NEXT: v_and_b32_e32 v11, s4, v11
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v5
|
|
; GFX7-NEXT: v_and_b32_e32 v12, s4, v12
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s4, v6
|
|
; GFX7-NEXT: v_and_b32_e32 v13, s4, v13
|
|
; GFX7-NEXT: v_and_b32_e32 v7, s4, v7
|
|
; GFX7-NEXT: v_and_b32_e32 v14, s4, v14
|
|
; GFX7-NEXT: v_and_b32_e32 v8, s4, v8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v10, v2
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v11, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v5, v12, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v6, v13, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v7, v14, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v8, v15, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v9, v16, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v3, v0, v1
|
|
; GFX7-NEXT: buffer_store_byte v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: idot8_acc8:
|
|
; GFX8: ; %bb.0: ; %entry
|
|
; GFX8-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX8-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
; GFX8-NEXT: v_mov_b32_e32 v5, 12
|
|
; GFX8-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s5
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s4, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v3, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v2, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v0, s0
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s1
|
|
; GFX8-NEXT: flat_load_ubyte v4, v[0:1]
|
|
; GFX8-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX8-NEXT: s_mov_b32 s10, -1
|
|
; GFX8-NEXT: s_mov_b32 s11, 0xe80000
|
|
; GFX8-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX8-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v10, 4, v3
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v16, 12, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 28, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 20, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v15, 4, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v17, 12, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 12, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 8, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v11, 28, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v12, 20, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v13, 12, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v14, 8, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v10, 12, v10
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v15, 12, v15
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v18, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v3, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v19, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v2, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v5, 12, v16
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v16, 12, v17
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v10, 12, v10
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v15, 12, v15
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v4, v5, v16, v4
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v14, 12, v14
|
|
; GFX8-NEXT: v_mad_u16 v4, v10, v15, v4
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_mad_u16 v4, v9, v14, v4
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v18, 12, v19
|
|
; GFX8-NEXT: v_mad_u16 v4, v8, v13, v4
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_mad_u16 v4, v17, v18, v4
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v3, 12, v3
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX8-NEXT: v_mad_u16 v4, v7, v12, v4
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX8-NEXT: v_mad_u16 v2, v3, v2, v4
|
|
; GFX8-NEXT: v_mad_u16 v2, v6, v11, v2
|
|
; GFX8-NEXT: flat_store_byte v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-LABEL: idot8_acc8:
|
|
; GFX9: ; %bb.0: ; %entry
|
|
; GFX9-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NEXT: v_mov_b32_e32 v4, 12
|
|
; GFX9-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NEXT: global_load_ubyte v3, v0, s[2:3]
|
|
; GFX9-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v9, 4, v1
|
|
; GFX9-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v14, 4, v2
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v15, 12, v1
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v16, 12, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v5, 28, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v7, 12, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v8, 8, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v10, 28, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v11, 20, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v12, 12, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v13, 8, v2
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v17, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v1, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v18, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v4, 12, v15
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v15, 12, v16
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v14, 12, v14
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v4, v15, v3
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v9, v14, v3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v8, v13, v3
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v16, 12, v17
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v7, v12, v3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v16, v17, v3
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v5, 12, v5
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v10, 12, v10
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v1, 12, v1
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX9-NEXT: v_mad_legacy_u16 v3, v6, v11, v3
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v10, 12, v10
|
|
; GFX9-NEXT: v_mad_legacy_u16 v1, v1, v2, v3
|
|
; GFX9-NEXT: v_mad_legacy_u16 v1, v5, v10, v1
|
|
; GFX9-NEXT: global_store_byte v0, v1, s[2:3]
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: idot8_acc8:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; GFX9-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-DL-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v4, 12
|
|
; GFX9-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-DL-NEXT: global_load_ubyte v3, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v9, 4, v1
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v14, 4, v2
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v15, 12, v1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v16, 12, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 28, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 12, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v8, 8, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v10, 28, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v11, 20, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v12, 12, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v13, 8, v2
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v17, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v1, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v18, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v4, 12, v15
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v15, 12, v16
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v14, 12, v14
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v4, v15, v3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v9, v14, v3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v8, v13, v3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v16, 12, v17
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v7, v12, v3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v16, v17, v3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v5, 12, v5
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v10, 12, v10
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v1, 12, v1
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v3, v6, v11, v3
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v10, 12, v10
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v1, v1, v2, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v1, v5, v10, v1
|
|
; GFX9-DL-NEXT: global_store_byte v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-XNACK-LABEL: idot8_acc8:
|
|
; GFX10-DL-XNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-XNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-XNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-XNACK-NEXT: global_load_ubyte v3, v0, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v4, 28, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v5, 24, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v7, 16, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v8, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v10, 4, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v1, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v16, 4, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v17, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v11, 28, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v10, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v12, 24, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v16, 12, v16
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v13, 20, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v14, 16, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v15, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v2, 8, v2
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v1, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v17, 12, v17
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v9, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v10, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v2, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v16, 12, v16
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v1, v17, v3
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v8, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v3, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v9, 12, v15
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v2, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v10, v16, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v7, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v10, 12, v14
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v8, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v9, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v3, v2, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v6, 12, v6
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v2, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v7, 12, v13
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v3, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v8, v9, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v5, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v8, 12, v12
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v6, 12, v6
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v7, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v2, v3, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v4, 12, v4
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v2, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v5, 12, v11
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v3, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v6, v7, v1
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v4, 12, v4
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v5, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v2, v3, v1
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v1, v4, v5, v1
|
|
; GFX10-DL-XNACK-NEXT: global_store_byte v0, v1, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-NOXNACK-LABEL: idot8_acc8:
|
|
; GFX10-DL-NOXNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NOXNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NOXNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v0, v0, s[6:7]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_ubyte v3, v2, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v4, 28, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v5, 24, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v6, 20, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v7, 16, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v8, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v10, 4, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v1, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v16, 4, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v17, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v11, 28, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v10, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v12, 24, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v16, 12, v16
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v13, 20, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v14, 16, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v15, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v0, 8, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v1, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v17, 12, v17
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v9, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v10, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v0, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v16, 12, v16
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v1, v1, v17, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v8, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v3, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v9, 12, v15
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v0, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v1, v10, v16, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v7, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v10, 12, v14
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v8, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v9, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v3, v0, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v6, 12, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v1, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v7, 12, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v3, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v8, v9, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v5, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v8, 12, v12
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v6, 12, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v7, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v1, v3, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v4, 12, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v1, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v5, 12, v11
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v3, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v6, v7, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v4, 12, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v5, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v1, v3, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v4, v5, v0
|
|
; GFX10-DL-NOXNACK-NEXT: global_store_byte v2, v0, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_endpgm
|
|
; GFX10-DL-LABEL: idot8_acc8:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NEXT: s_mov_b32 s14, -1
|
|
; GFX10-DL-NEXT: s_mov_b32 s15, 0x31c16000
|
|
; GFX10-DL-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: global_load_ubyte v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: s_load_dword s0, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_lshr_b32 s2, s0, 12
|
|
; GFX10-DL-NEXT: s_lshr_b32 s3, s1, 12
|
|
; GFX10-DL-NEXT: s_bfe_i32 s6, s0, 0x40000
|
|
; GFX10-DL-NEXT: s_bfe_i32 s7, s1, 0x40000
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v2, 12, s2
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v3, 12, s3
|
|
; GFX10-DL-NEXT: s_bfe_i32 s8, s0, 0x40004
|
|
; GFX10-DL-NEXT: s_bfe_i32 s9, s0, 0x40008
|
|
; GFX10-DL-NEXT: s_bfe_i32 s10, s1, 0x40008
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s1, 0x40004
|
|
; GFX10-DL-NEXT: v_mul_i32_i24_e64 v4, s9, s10
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v2, 12, v2
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v3, 12, v3
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40010
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s6, s7, v1
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s8, s2, v1
|
|
; GFX10-DL-NEXT: s_movk_i32 s2, 0xff
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v2, s2, v2
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v3, s2, v3
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40010
|
|
; GFX10-DL-NEXT: v_mad_u32_u24 v1, v2, v3, v1
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40014
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40014
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40018
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40018
|
|
; GFX10-DL-NEXT: s_ashr_i32 s0, s0, 28
|
|
; GFX10-DL-NEXT: s_ashr_i32 s1, s1, 28
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s0, s1, v1
|
|
; GFX10-DL-NEXT: global_store_byte v0, v1, s[4:5]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<8 x i4> addrspace(1)* %src2,
|
|
i8 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <8 x i4> %vec1, i64 0
|
|
%cv1e0 = sext i4 %v1e0 to i8
|
|
%v2e0 = extractelement <8 x i4> %vec2, i64 0
|
|
%cv2e0 = sext i4 %v2e0 to i8
|
|
%mul0 = mul nuw nsw i8 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <8 x i4> %vec1, i64 1
|
|
%cv1e1 = sext i4 %v1e1 to i8
|
|
%v2e1 = extractelement <8 x i4> %vec2, i64 1
|
|
%cv2e1 = sext i4 %v2e1 to i8
|
|
%mul1 = mul nuw nsw i8 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <8 x i4> %vec1, i64 2
|
|
%cv1e2 = sext i4 %v1e2 to i8
|
|
%v2e2 = extractelement <8 x i4> %vec2, i64 2
|
|
%cv2e2 = sext i4 %v2e2 to i8
|
|
%mul2 = mul nuw nsw i8 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <8 x i4> %vec1, i64 3
|
|
%cv1e3 = sext i4 %v1e3 to i8
|
|
%v2e3 = extractelement <8 x i4> %vec2, i64 3
|
|
%cv2e3 = sext i4 %v2e3 to i8
|
|
%mul3 = mul nuw nsw i8 %cv1e3, %cv2e3
|
|
|
|
%v1e4 = extractelement <8 x i4> %vec1, i64 4
|
|
%cv1e4 = sext i4 %v1e4 to i8
|
|
%v2e4 = extractelement <8 x i4> %vec2, i64 4
|
|
%cv2e4 = sext i4 %v2e4 to i8
|
|
%mul4 = mul nuw nsw i8 %cv1e4, %cv2e4
|
|
|
|
%v1e5 = extractelement <8 x i4> %vec1, i64 5
|
|
%cv1e5 = sext i4 %v1e5 to i8
|
|
%v2e5 = extractelement <8 x i4> %vec2, i64 5
|
|
%cv2e5 = sext i4 %v2e5 to i8
|
|
%mul5 = mul nuw nsw i8 %cv1e5, %cv2e5
|
|
|
|
%v1e6 = extractelement <8 x i4> %vec1, i64 6
|
|
%cv1e6 = sext i4 %v1e6 to i8
|
|
%v2e6 = extractelement <8 x i4> %vec2, i64 6
|
|
%cv2e6 = sext i4 %v2e6 to i8
|
|
%mul6 = mul nuw nsw i8 %cv1e6, %cv2e6
|
|
|
|
%v1e7 = extractelement <8 x i4> %vec1, i64 7
|
|
%cv1e7 = sext i4 %v1e7 to i8
|
|
%v2e7 = extractelement <8 x i4> %vec2, i64 7
|
|
%cv2e7 = sext i4 %v2e7 to i8
|
|
%mul7 = mul nuw nsw i8 %cv1e7, %cv2e7
|
|
|
|
%acc = load i8, i8 addrspace(1)* %dst, align 4
|
|
%add1 = add i8 %mul0, %acc
|
|
%add2 = add i8 %add1, %mul1
|
|
%add3 = add i8 %add2, %mul2
|
|
%add4 = add i8 %add3, %mul3
|
|
%add5 = add i8 %add4, %mul4
|
|
%add6 = add i8 %add5, %mul5
|
|
%add7 = add i8 %add6, %mul6
|
|
%add8 = add i8 %add7, %mul7
|
|
|
|
store i8 %add8, i8 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; Make sure the pattern is not recognized if there are multiple uses of the
|
|
; intermediate multiplications.
|
|
define amdgpu_kernel void @idot8_multiuses_mul1(<8 x i4> addrspace(1)* %src1,
|
|
; GFX7-LABEL: idot8_multiuses_mul1:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; GFX7-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX7-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX7-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x9
|
|
; GFX7-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xd
|
|
; GFX7-NEXT: s_mov_b32 s14, -1
|
|
; GFX7-NEXT: s_mov_b32 s15, 0xe8f000
|
|
; GFX7-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX7-NEXT: s_mov_b32 s3, 0xf000
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX7-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX7-NEXT: buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[6:7]
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_load_dword s4, s[0:1], 0x0
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_i32 v1, v2, 0, 4
|
|
; GFX7-NEXT: v_bfe_i32 v3, v2, 4, 4
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_bfe_i32 v9, v0, 0, 4
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: v_mad_i32_i24 v16, v1, v9, s4
|
|
; GFX7-NEXT: v_bfe_i32 v10, v0, 4, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v1, v9, v16
|
|
; GFX7-NEXT: v_bfe_i32 v4, v2, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v11, v0, 8, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v3, v10, v1
|
|
; GFX7-NEXT: v_bfe_i32 v5, v2, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v12, v0, 12, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v4, v11, v1
|
|
; GFX7-NEXT: v_bfe_i32 v6, v2, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v13, v0, 16, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v5, v12, v1
|
|
; GFX7-NEXT: v_bfe_i32 v7, v2, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v14, v0, 20, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v6, v13, v1
|
|
; GFX7-NEXT: v_bfe_i32 v8, v2, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v15, v0, 24, 4
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v7, v14, v1
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v2, 28, v2
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v0, 28, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v1, v8, v15, v1
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v2, v0, v1
|
|
; GFX7-NEXT: v_add_i32_e32 v0, vcc, v16, v0
|
|
; GFX7-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: idot8_multiuses_mul1:
|
|
; GFX8: ; %bb.0: ; %entry
|
|
; GFX8-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX8-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
; GFX8-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX8-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s5
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s4, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v3, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX8-NEXT: s_mov_b32 s10, -1
|
|
; GFX8-NEXT: s_mov_b32 s11, 0xe80000
|
|
; GFX8-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX8-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_bfe_i32 v1, v3, 0, 4
|
|
; GFX8-NEXT: v_bfe_i32 v4, v3, 4, 4
|
|
; GFX8-NEXT: v_bfe_i32 v6, v3, 8, 4
|
|
; GFX8-NEXT: v_bfe_i32 v8, v3, 12, 4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_bfe_i32 v2, v0, 0, 4
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mad_i32_i24 v16, v1, v2, s2
|
|
; GFX8-NEXT: v_bfe_i32 v5, v0, 4, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v1, v2, v16
|
|
; GFX8-NEXT: v_bfe_i32 v7, v0, 8, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v4, v5, v1
|
|
; GFX8-NEXT: v_bfe_i32 v9, v0, 12, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v6, v7, v1
|
|
; GFX8-NEXT: v_bfe_i32 v10, v3, 16, 4
|
|
; GFX8-NEXT: v_bfe_i32 v11, v0, 16, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v8, v9, v1
|
|
; GFX8-NEXT: v_bfe_i32 v12, v3, 20, 4
|
|
; GFX8-NEXT: v_bfe_i32 v13, v0, 20, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v10, v11, v1
|
|
; GFX8-NEXT: v_bfe_i32 v14, v3, 24, 4
|
|
; GFX8-NEXT: v_bfe_i32 v15, v0, 24, 4
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v12, v13, v1
|
|
; GFX8-NEXT: v_ashrrev_i32_e32 v3, 28, v3
|
|
; GFX8-NEXT: v_ashrrev_i32_e32 v0, 28, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v1, v14, v15, v1
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v3, v0, v1
|
|
; GFX8-NEXT: v_add_u32_e32 v2, vcc, v16, v0
|
|
; GFX8-NEXT: v_mov_b32_e32 v0, s0
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s1
|
|
; GFX8-NEXT: flat_store_dword v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-LABEL: idot8_multiuses_mul1:
|
|
; GFX9: ; %bb.0: ; %entry
|
|
; GFX9-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NEXT: v_bfe_i32 v3, v1, 0, 4
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_bfe_i32 v4, v2, 0, 4
|
|
; GFX9-NEXT: v_bfe_i32 v5, v1, 4, 4
|
|
; GFX9-NEXT: v_bfe_i32 v6, v2, 4, 4
|
|
; GFX9-NEXT: v_bfe_i32 v7, v1, 8, 4
|
|
; GFX9-NEXT: v_bfe_i32 v8, v2, 8, 4
|
|
; GFX9-NEXT: v_bfe_i32 v9, v1, 12, 4
|
|
; GFX9-NEXT: v_bfe_i32 v10, v2, 12, 4
|
|
; GFX9-NEXT: v_bfe_i32 v11, v1, 16, 4
|
|
; GFX9-NEXT: v_bfe_i32 v12, v2, 16, 4
|
|
; GFX9-NEXT: v_bfe_i32 v13, v1, 20, 4
|
|
; GFX9-NEXT: v_bfe_i32 v15, v1, 24, 4
|
|
; GFX9-NEXT: v_bfe_i32 v14, v2, 20, 4
|
|
; GFX9-NEXT: v_bfe_i32 v16, v2, 24, 4
|
|
; GFX9-NEXT: v_ashrrev_i32_e32 v1, 28, v1
|
|
; GFX9-NEXT: v_ashrrev_i32_e32 v2, 28, v2
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v1, v1, v2
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: v_mad_i32_i24 v2, v3, v4, s0
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v5, v5, v6
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v6, v7, v8
|
|
; GFX9-NEXT: v_mad_i32_i24 v3, v3, v4, v2
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v7, v9, v10
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v8, v11, v12
|
|
; GFX9-NEXT: v_add3_u32 v3, v3, v5, v6
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v9, v13, v14
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v10, v15, v16
|
|
; GFX9-NEXT: v_add3_u32 v3, v3, v7, v8
|
|
; GFX9-NEXT: v_add3_u32 v3, v3, v9, v10
|
|
; GFX9-NEXT: v_add3_u32 v1, v3, v1, v2
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: idot8_multiuses_mul1:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; GFX9-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-DL-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-DL-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_bfe_i32 v3, v1, 0, 4
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_bfe_i32 v4, v2, 0, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v5, v1, 4, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v6, v2, 4, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v7, v1, 8, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v8, v2, 8, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v9, v1, 12, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v10, v2, 12, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v11, v1, 16, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v12, v2, 16, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v13, v1, 20, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v15, v1, 24, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v14, v2, 20, 4
|
|
; GFX9-DL-NEXT: v_bfe_i32 v16, v2, 24, 4
|
|
; GFX9-DL-NEXT: v_ashrrev_i32_e32 v1, 28, v1
|
|
; GFX9-DL-NEXT: v_ashrrev_i32_e32 v2, 28, v2
|
|
; GFX9-DL-NEXT: v_mul_i32_i24_e32 v1, v1, v2
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_i32_i24 v2, v3, v4, s0
|
|
; GFX9-DL-NEXT: v_mul_i32_i24_e32 v5, v5, v6
|
|
; GFX9-DL-NEXT: v_mul_i32_i24_e32 v6, v7, v8
|
|
; GFX9-DL-NEXT: v_mad_i32_i24 v3, v3, v4, v2
|
|
; GFX9-DL-NEXT: v_mul_i32_i24_e32 v7, v9, v10
|
|
; GFX9-DL-NEXT: v_mul_i32_i24_e32 v8, v11, v12
|
|
; GFX9-DL-NEXT: v_add3_u32 v3, v3, v5, v6
|
|
; GFX9-DL-NEXT: v_mul_i32_i24_e32 v9, v13, v14
|
|
; GFX9-DL-NEXT: v_mul_i32_i24_e32 v10, v15, v16
|
|
; GFX9-DL-NEXT: v_add3_u32 v3, v3, v7, v8
|
|
; GFX9-DL-NEXT: v_add3_u32 v3, v3, v9, v10
|
|
; GFX9-DL-NEXT: v_add3_u32 v1, v3, v1, v2
|
|
; GFX9-DL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-XNACK-LABEL: idot8_multiuses_mul1:
|
|
; GFX10-DL-XNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-XNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-XNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-XNACK-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v0, v1, 0, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v3, v1, 4, 4
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v4, v2, 4, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v5, v1, 8, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v6, v2, 8, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v7, v2, 0, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v8, v1, 12, 4
|
|
; GFX10-DL-XNACK-NEXT: v_mul_i32_i24_e32 v3, v3, v4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v9, v2, 12, 4
|
|
; GFX10-DL-XNACK-NEXT: v_mul_i32_i24_e32 v4, v5, v6
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_mad_i32_i24 v5, v0, v7, s2
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v6, v1, 16, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v10, v2, 16, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v11, v1, 20, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v12, v2, 20, 4
|
|
; GFX10-DL-XNACK-NEXT: v_mad_i32_i24 v0, v0, v7, v5
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v7, v1, 24, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_i32 v13, v2, 24, 4
|
|
; GFX10-DL-XNACK-NEXT: v_mul_i32_i24_e32 v8, v8, v9
|
|
; GFX10-DL-XNACK-NEXT: v_mul_i32_i24_e32 v6, v6, v10
|
|
; GFX10-DL-XNACK-NEXT: v_add3_u32 v0, v0, v3, v4
|
|
; GFX10-DL-XNACK-NEXT: v_mul_i32_i24_e32 v3, v11, v12
|
|
; GFX10-DL-XNACK-NEXT: v_mul_i32_i24_e32 v4, v7, v13
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i32_e32 v1, 28, v1
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i32_e32 v2, 28, v2
|
|
; GFX10-DL-XNACK-NEXT: v_add3_u32 v0, v0, v8, v6
|
|
; GFX10-DL-XNACK-NEXT: v_mul_i32_i24_e32 v1, v1, v2
|
|
; GFX10-DL-XNACK-NEXT: v_add3_u32 v0, v0, v3, v4
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-XNACK-NEXT: v_add3_u32 v0, v0, v1, v5
|
|
; GFX10-DL-XNACK-NEXT: global_store_dword v2, v0, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-NOXNACK-LABEL: idot8_multiuses_mul1:
|
|
; GFX10-DL-NOXNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NOXNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NOXNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v0, v0, s[6:7]
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v2, v1, 0, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v3, v1, 4, 4
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v4, v0, 4, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v5, v1, 8, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v6, v0, 8, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v7, v0, 0, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v8, v1, 12, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_i32_i24_e32 v3, v3, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v9, v0, 12, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_i32_i24_e32 v4, v5, v6
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_i32_i24 v5, v2, v7, s2
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v6, v1, 16, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v10, v0, 16, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v11, v1, 20, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v12, v0, 20, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_i32_i24 v2, v2, v7, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v7, v1, 24, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_i32 v13, v0, 24, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_i32_i24_e32 v8, v8, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_i32_i24_e32 v6, v6, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_add3_u32 v2, v2, v3, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_i32_i24_e32 v3, v11, v12
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_i32_i24_e32 v4, v7, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i32_e32 v1, 28, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i32_e32 v0, 28, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_add3_u32 v2, v2, v8, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_i32_i24_e32 v0, v1, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_add3_u32 v1, v2, v3, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NOXNACK-NEXT: v_add3_u32 v0, v1, v0, v5
|
|
; GFX10-DL-NOXNACK-NEXT: global_store_dword v2, v0, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_endpgm
|
|
; GFX10-DL-LABEL: idot8_multiuses_mul1:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_load_dword s6, s[4:5], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s0, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, s6
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40000
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40000
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v0, s2, s3, v0
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v0
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40004
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40004
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40008
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40008
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x4000c
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x4000c
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40010
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40010
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40014
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40014
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: s_bfe_i32 s2, s0, 0x40018
|
|
; GFX10-DL-NEXT: s_bfe_i32 s3, s1, 0x40018
|
|
; GFX10-DL-NEXT: s_ashr_i32 s0, s0, 28
|
|
; GFX10-DL-NEXT: s_ashr_i32 s1, s1, 28
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s2, s3, v1
|
|
; GFX10-DL-NEXT: v_mad_i32_i24 v1, s0, s1, v1
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v0, v0, v1
|
|
; GFX10-DL-NEXT: global_store_dword v2, v0, s[4:5]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<8 x i4> addrspace(1)* %src2,
|
|
i32 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <8 x i4> %vec1, i64 0
|
|
%cv1e0 = sext i4 %v1e0 to i32
|
|
%v2e0 = extractelement <8 x i4> %vec2, i64 0
|
|
%cv2e0 = sext i4 %v2e0 to i32
|
|
%mul0 = mul nuw nsw i32 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <8 x i4> %vec1, i64 1
|
|
%cv1e1 = sext i4 %v1e1 to i32
|
|
%v2e1 = extractelement <8 x i4> %vec2, i64 1
|
|
%cv2e1 = sext i4 %v2e1 to i32
|
|
%mul1 = mul nuw nsw i32 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <8 x i4> %vec1, i64 2
|
|
%cv1e2 = sext i4 %v1e2 to i32
|
|
%v2e2 = extractelement <8 x i4> %vec2, i64 2
|
|
%cv2e2 = sext i4 %v2e2 to i32
|
|
%mul2 = mul nuw nsw i32 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <8 x i4> %vec1, i64 3
|
|
%cv1e3 = sext i4 %v1e3 to i32
|
|
%v2e3 = extractelement <8 x i4> %vec2, i64 3
|
|
%cv2e3 = sext i4 %v2e3 to i32
|
|
%mul3 = mul nuw nsw i32 %cv1e3, %cv2e3
|
|
|
|
%v1e4 = extractelement <8 x i4> %vec1, i64 4
|
|
%cv1e4 = sext i4 %v1e4 to i32
|
|
%v2e4 = extractelement <8 x i4> %vec2, i64 4
|
|
%cv2e4 = sext i4 %v2e4 to i32
|
|
%mul4 = mul nuw nsw i32 %cv1e4, %cv2e4
|
|
|
|
%v1e5 = extractelement <8 x i4> %vec1, i64 5
|
|
%cv1e5 = sext i4 %v1e5 to i32
|
|
%v2e5 = extractelement <8 x i4> %vec2, i64 5
|
|
%cv2e5 = sext i4 %v2e5 to i32
|
|
%mul5 = mul nuw nsw i32 %cv1e5, %cv2e5
|
|
|
|
%v1e6 = extractelement <8 x i4> %vec1, i64 6
|
|
%cv1e6 = sext i4 %v1e6 to i32
|
|
%v2e6 = extractelement <8 x i4> %vec2, i64 6
|
|
%cv2e6 = sext i4 %v2e6 to i32
|
|
%mul6 = mul nuw nsw i32 %cv1e6, %cv2e6
|
|
|
|
%v1e7 = extractelement <8 x i4> %vec1, i64 7
|
|
%cv1e7 = sext i4 %v1e7 to i32
|
|
%v2e7 = extractelement <8 x i4> %vec2, i64 7
|
|
%cv2e7 = sext i4 %v2e7 to i32
|
|
%mul7 = mul nuw nsw i32 %cv1e7, %cv2e7
|
|
|
|
%acc = load i32, i32 addrspace(1)* %dst, align 4
|
|
%add = add i32 %mul0, %acc
|
|
%add1 = add i32 %mul0, %add
|
|
%add2 = add i32 %add1, %mul1
|
|
%add3 = add i32 %add2, %mul2
|
|
%add4 = add i32 %add3, %mul3
|
|
%add5 = add i32 %add4, %mul4
|
|
%add6 = add i32 %add5, %mul5
|
|
%add7 = add i32 %add6, %mul6
|
|
%add8 = add i32 %add7, %mul7
|
|
|
|
%res = add i32 %add, %add8
|
|
store i32 %res, i32 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; TODO: Support this pattern.
|
|
define amdgpu_kernel void @idot8_acc32_vecMul(<8 x i4> addrspace(1)* %src1,
|
|
; GFX7-LABEL: idot8_acc32_vecMul:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; GFX7-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX7-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX7-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x9
|
|
; GFX7-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xd
|
|
; GFX7-NEXT: s_mov_b32 s14, -1
|
|
; GFX7-NEXT: s_mov_b32 s15, 0xe8f000
|
|
; GFX7-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX7-NEXT: s_mov_b32 s3, 0xf000
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX7-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX7-NEXT: buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[6:7]
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_load_dword s4, s[0:1], 0x0
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v1, 28, v2
|
|
; GFX7-NEXT: v_bfe_i32 v3, v2, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v4, v2, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v5, v2, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v6, v2, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v7, v2, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v8, v2, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v2, v2, 0, 4
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v9, 28, v0
|
|
; GFX7-NEXT: v_bfe_i32 v10, v0, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v11, v0, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v12, v0, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v13, v0, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v14, v0, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v15, v0, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v0, v0, 0, 4
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v2, v0, s4
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v8, v15, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v7, v14, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v6, v13, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v5, v12, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v4, v11, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v3, v10, v0
|
|
; GFX7-NEXT: v_mad_i32_i24 v0, v1, v9, v0
|
|
; GFX7-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: idot8_acc32_vecMul:
|
|
; GFX8: ; %bb.0: ; %entry
|
|
; GFX8-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX8-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
; GFX8-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX8-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s5
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s4, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s7
|
|
; GFX8-NEXT: v_add_u32_e32 v2, vcc, s6, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v3, vcc, 0, v3, vcc
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: flat_load_dword v1, v[2:3]
|
|
; GFX8-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX8-NEXT: s_mov_b32 s10, -1
|
|
; GFX8-NEXT: s_mov_b32 s11, 0xe80000
|
|
; GFX8-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX8-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_ashrrev_i32_e32 v2, 28, v0
|
|
; GFX8-NEXT: v_bfe_i32 v3, v0, 24, 4
|
|
; GFX8-NEXT: v_bfe_i32 v4, v0, 20, 4
|
|
; GFX8-NEXT: v_bfe_i32 v5, v0, 16, 4
|
|
; GFX8-NEXT: v_bfe_i32 v6, v0, 12, 4
|
|
; GFX8-NEXT: v_bfe_i32 v7, v0, 8, 4
|
|
; GFX8-NEXT: v_bfe_i32 v8, v0, 4, 4
|
|
; GFX8-NEXT: v_bfe_i32 v0, v0, 0, 4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_ashrrev_i32_e32 v9, 28, v1
|
|
; GFX8-NEXT: v_bfe_i32 v10, v1, 24, 4
|
|
; GFX8-NEXT: v_bfe_i32 v11, v1, 20, 4
|
|
; GFX8-NEXT: v_bfe_i32 v12, v1, 16, 4
|
|
; GFX8-NEXT: v_bfe_i32 v13, v1, 12, 4
|
|
; GFX8-NEXT: v_bfe_i32 v14, v1, 8, 4
|
|
; GFX8-NEXT: v_bfe_i32 v15, v1, 4, 4
|
|
; GFX8-NEXT: v_bfe_i32 v1, v1, 0, 4
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v0, v1, s2
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v8, v15, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v7, v14, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v6, v13, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v5, v12, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v4, v11, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v0, v3, v10, v0
|
|
; GFX8-NEXT: v_mad_i32_i24 v2, v2, v9, v0
|
|
; GFX8-NEXT: v_mov_b32_e32 v0, s0
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s1
|
|
; GFX8-NEXT: flat_store_dword v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-LABEL: idot8_acc32_vecMul:
|
|
; GFX9: ; %bb.0: ; %entry
|
|
; GFX9-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NEXT: v_ashrrev_i32_e32 v3, 28, v1
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_ashrrev_i32_e32 v10, 28, v2
|
|
; GFX9-NEXT: v_bfe_i32 v4, v1, 24, 4
|
|
; GFX9-NEXT: v_bfe_i32 v11, v2, 24, 4
|
|
; GFX9-NEXT: v_bfe_i32 v5, v1, 20, 4
|
|
; GFX9-NEXT: v_bfe_i32 v12, v2, 20, 4
|
|
; GFX9-NEXT: v_bfe_i32 v6, v1, 16, 4
|
|
; GFX9-NEXT: v_bfe_i32 v13, v2, 16, 4
|
|
; GFX9-NEXT: v_bfe_i32 v7, v1, 12, 4
|
|
; GFX9-NEXT: v_bfe_i32 v14, v2, 12, 4
|
|
; GFX9-NEXT: v_bfe_i32 v8, v1, 8, 4
|
|
; GFX9-NEXT: v_bfe_i32 v9, v1, 4, 4
|
|
; GFX9-NEXT: v_bfe_i32 v15, v2, 8, 4
|
|
; GFX9-NEXT: v_bfe_i32 v16, v2, 4, 4
|
|
; GFX9-NEXT: v_bfe_i32 v1, v1, 0, 4
|
|
; GFX9-NEXT: v_bfe_i32 v2, v2, 0, 4
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v1, v1, v2
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v2, v9, v16
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v8, v8, v15
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v7, v7, v14
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: v_add3_u32 v1, v1, s0, v2
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v6, v6, v13
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v5, v5, v12
|
|
; GFX9-NEXT: v_add3_u32 v1, v1, v8, v7
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v4, v4, v11
|
|
; GFX9-NEXT: v_mul_i32_i24_e32 v3, v3, v10
|
|
; GFX9-NEXT: v_add3_u32 v1, v1, v6, v5
|
|
; GFX9-NEXT: v_add3_u32 v1, v1, v4, v3
|
|
; GFX9-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: idot8_acc32_vecMul:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; GFX9-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-DL-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-DL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-DL-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX9-DL-NEXT: v_dot8_i32_i4 v0, v2, v3, s0
|
|
; GFX9-DL-NEXT: global_store_dword v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-XNACK-LABEL: idot8_acc32_vecMul:
|
|
; GFX10-DL-XNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-XNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-XNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_dot8_i32_i4 v1, v1, v2, s2
|
|
; GFX10-DL-XNACK-NEXT: global_store_dword v0, v1, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-NOXNACK-LABEL: idot8_acc32_vecMul:
|
|
; GFX10-DL-NOXNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NOXNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NOXNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v0, v0, s[6:7]
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_dot8_i32_i4 v0, v1, v0, s2
|
|
; GFX10-DL-NOXNACK-NEXT: global_store_dword v2, v0, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_endpgm
|
|
; GFX10-DL-LABEL: idot8_acc32_vecMul:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX10-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_load_dword s6, s[4:5], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s0, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, s6
|
|
; GFX10-DL-NEXT: v_dot8_i32_i4 v0, s0, s1, v0
|
|
; GFX10-DL-NEXT: global_store_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<8 x i4> addrspace(1)* %src2,
|
|
i32 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %gep2
|
|
|
|
%cvec1 = sext <8 x i4> %vec1 to <8 x i32>
|
|
%cvec2 = sext <8 x i4> %vec2 to <8 x i32>
|
|
|
|
%mul = mul <8 x i32> %cvec1, %cvec2
|
|
%mul0 = extractelement <8 x i32> %mul, i64 0
|
|
%mul1 = extractelement <8 x i32> %mul, i64 1
|
|
%mul2 = extractelement <8 x i32> %mul, i64 2
|
|
%mul3 = extractelement <8 x i32> %mul, i64 3
|
|
%mul4 = extractelement <8 x i32> %mul, i64 4
|
|
%mul5 = extractelement <8 x i32> %mul, i64 5
|
|
%mul6 = extractelement <8 x i32> %mul, i64 6
|
|
%mul7 = extractelement <8 x i32> %mul, i64 7
|
|
|
|
%acc = load i32, i32 addrspace(1)* %dst, align 4
|
|
%add1 = add i32 %mul0, %acc
|
|
%add2 = add i32 %add1, %mul1
|
|
%add3 = add i32 %add2, %mul2
|
|
%add4 = add i32 %add3, %mul3
|
|
%add5 = add i32 %add4, %mul4
|
|
%add6 = add i32 %add5, %mul5
|
|
%add7 = add i32 %add6, %mul6
|
|
%add8 = add i32 %add7, %mul7
|
|
|
|
store i32 %add8, i32 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; TODO: Support this pattern.
|
|
define amdgpu_kernel void @idot8_acc16_vecMul(<8 x i4> addrspace(1)* %src1,
|
|
; GFX7-LABEL: idot8_acc16_vecMul:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; GFX7-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX7-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX7-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x9
|
|
; GFX7-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xd
|
|
; GFX7-NEXT: s_mov_b32 s14, -1
|
|
; GFX7-NEXT: s_mov_b32 s15, 0xe8f000
|
|
; GFX7-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX7-NEXT: s_mov_b32 s3, 0xf000
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX7-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX7-NEXT: buffer_load_dword v3, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[6:7]
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b32 s4, 0xffff
|
|
; GFX7-NEXT: v_mov_b32_e32 v2, 0xffff
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_i32 v1, v3, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v4, v3, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v5, v3, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v6, v3, 0, 4
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v1, 16, v1
|
|
; GFX7-NEXT: v_and_b32_e32 v4, s4, v4
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v5, 16, v5
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s4, v6
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_bfe_i32 v10, v0, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v11, v0, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v12, v0, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v13, v0, 0, 4
|
|
; GFX7-NEXT: v_or_b32_e32 v1, v4, v1
|
|
; GFX7-NEXT: v_or_b32_e32 v4, v6, v5
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v5, 16, v10
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s4, v11
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v10, 16, v12
|
|
; GFX7-NEXT: v_and_b32_e32 v11, v2, v13
|
|
; GFX7-NEXT: v_bfe_i32 v7, v3, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v8, v3, 8, 4
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v9, 28, v3
|
|
; GFX7-NEXT: v_bfe_i32 v3, v3, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v14, v0, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v15, v0, 8, 4
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v16, 28, v0
|
|
; GFX7-NEXT: v_bfe_i32 v0, v0, 12, 4
|
|
; GFX7-NEXT: v_or_b32_e32 v5, v6, v5
|
|
; GFX7-NEXT: v_or_b32_e32 v6, v11, v10
|
|
; GFX7-NEXT: v_and_b32_e32 v12, v2, v14
|
|
; GFX7-NEXT: v_and_b32_e32 v13, v2, v15
|
|
; GFX7-NEXT: v_and_b32_e32 v14, v2, v16
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v15, 16, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v16, 16, v4
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v11, 16, v6
|
|
; GFX7-NEXT: v_and_b32_e32 v3, v2, v3
|
|
; GFX7-NEXT: v_and_b32_e32 v9, v2, v9
|
|
; GFX7-NEXT: v_and_b32_e32 v0, v2, v0
|
|
; GFX7-NEXT: v_and_b32_e32 v4, v2, v4
|
|
; GFX7-NEXT: v_and_b32_e32 v1, v2, v1
|
|
; GFX7-NEXT: v_and_b32_e32 v6, v2, v6
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v10, 16, v5
|
|
; GFX7-NEXT: v_and_b32_e32 v2, v2, v5
|
|
; GFX7-NEXT: buffer_load_ushort v5, off, s[0:3], 0
|
|
; GFX7-NEXT: v_and_b32_e32 v8, s4, v8
|
|
; GFX7-NEXT: v_and_b32_e32 v7, s4, v7
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v4, v4, v6, v5
|
|
; GFX7-NEXT: v_mad_u32_u24 v4, v16, v11, v4
|
|
; GFX7-NEXT: v_mad_u32_u24 v4, v8, v13, v4
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v3, v0, v4
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v1, v2, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v15, v10, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v7, v12, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v9, v14, v0
|
|
; GFX7-NEXT: buffer_store_short v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: idot8_acc16_vecMul:
|
|
; GFX8: ; %bb.0: ; %entry
|
|
; GFX8-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX8-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
; GFX8-NEXT: v_mov_b32_e32 v5, 12
|
|
; GFX8-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s5
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s4, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v3, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v2, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v0, s0
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s1
|
|
; GFX8-NEXT: flat_load_ushort v4, v[0:1]
|
|
; GFX8-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX8-NEXT: s_mov_b32 s10, -1
|
|
; GFX8-NEXT: s_mov_b32 s11, 0xe80000
|
|
; GFX8-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX8-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 4, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 8, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 12, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 20, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v10, 28, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v11, 4, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v12, 8, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v13, 12, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v14, 20, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v15, 28, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v16, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v17, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v18, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v5, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v3, 12, v3
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v2, 12, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v3, 12, v3
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v2, v3, v2, v4
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_mad_u16 v2, v6, v11, v2
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_mad_u16 v2, v7, v12, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v17, 12, v17
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX8-NEXT: v_mad_u16 v2, v8, v13, v2
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v14, 12, v14
|
|
; GFX8-NEXT: v_mad_u16 v2, v17, v5, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v10, 12, v10
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v15, 12, v15
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v16, 12, v16
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v18, 12, v18
|
|
; GFX8-NEXT: v_mad_u16 v2, v9, v14, v2
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v10, 12, v10
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v15, 12, v15
|
|
; GFX8-NEXT: v_mad_u16 v2, v16, v18, v2
|
|
; GFX8-NEXT: v_mad_u16 v2, v10, v15, v2
|
|
; GFX8-NEXT: flat_store_short v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-LABEL: idot8_acc16_vecMul:
|
|
; GFX9: ; %bb.0: ; %entry
|
|
; GFX9-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NEXT: v_mov_b32_e32 v2, 0xffff
|
|
; GFX9-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: global_load_dword v3, v0, s[4:5]
|
|
; GFX9-NEXT: global_load_dword v4, v0, s[6:7]
|
|
; GFX9-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NEXT: v_and_b32_e32 v10, 15, v3
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_and_b32_e32 v17, 15, v4
|
|
; GFX9-NEXT: v_bfe_u32 v0, v3, 24, 4
|
|
; GFX9-NEXT: v_bfe_u32 v6, v3, 16, 4
|
|
; GFX9-NEXT: v_bfe_u32 v8, v3, 8, 4
|
|
; GFX9-NEXT: v_bfe_u32 v13, v4, 16, 4
|
|
; GFX9-NEXT: v_bfe_u32 v15, v4, 8, 4
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v5, 28, v3
|
|
; GFX9-NEXT: v_bfe_u32 v7, v3, 20, 4
|
|
; GFX9-NEXT: v_bfe_u32 v9, v3, 12, 4
|
|
; GFX9-NEXT: v_bfe_u32 v3, v3, 4, 4
|
|
; GFX9-NEXT: v_and_b32_e32 v10, v2, v10
|
|
; GFX9-NEXT: v_bfe_u32 v11, v4, 24, 4
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v12, 28, v4
|
|
; GFX9-NEXT: v_bfe_u32 v14, v4, 20, 4
|
|
; GFX9-NEXT: v_bfe_u32 v16, v4, 12, 4
|
|
; GFX9-NEXT: v_bfe_u32 v4, v4, 4, 4
|
|
; GFX9-NEXT: v_and_b32_e32 v17, v2, v17
|
|
; GFX9-NEXT: v_lshl_or_b32 v3, v3, 16, v10
|
|
; GFX9-NEXT: v_lshl_or_b32 v4, v4, 16, v17
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v3, 12, v3 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v4, 12, v4 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v3, 12, v3 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v4, 12, v4 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_mul_lo_u16 v3, v3, v4
|
|
; GFX9-NEXT: global_load_ushort v4, v1, s[2:3]
|
|
; GFX9-NEXT: v_and_b32_e32 v8, v2, v8
|
|
; GFX9-NEXT: v_and_b32_e32 v0, v2, v0
|
|
; GFX9-NEXT: v_and_b32_e32 v15, v2, v15
|
|
; GFX9-NEXT: v_lshl_or_b32 v0, v5, 16, v0
|
|
; GFX9-NEXT: v_lshl_or_b32 v8, v9, 16, v8
|
|
; GFX9-NEXT: v_lshl_or_b32 v5, v16, 16, v15
|
|
; GFX9-NEXT: v_and_b32_e32 v6, v2, v6
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v8, 12, v8 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v5, 12, v5 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_and_b32_e32 v13, v2, v13
|
|
; GFX9-NEXT: v_and_b32_e32 v2, v2, v11
|
|
; GFX9-NEXT: v_lshl_or_b32 v6, v7, 16, v6
|
|
; GFX9-NEXT: v_lshl_or_b32 v7, v14, 16, v13
|
|
; GFX9-NEXT: v_lshl_or_b32 v2, v12, 16, v2
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v8, 12, v8 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v5, 12, v5 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v6, 12, v6 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v0, 12, v0 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_lshlrev_b16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_mul_lo_u16 v5, v8, v5
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v6, 12, v6 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v0, 12, v0 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_ashrrev_i16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX9-NEXT: v_pk_mul_lo_u16 v0, v0, v2
|
|
; GFX9-NEXT: v_pk_mul_lo_u16 v2, v6, v7
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_add_u16_e32 v4, v3, v4
|
|
; GFX9-NEXT: v_add_u16_sdwa v3, v4, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_add_u16_e32 v3, v3, v5
|
|
; GFX9-NEXT: v_add_u16_sdwa v3, v3, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_add_u16_e32 v3, v3, v2
|
|
; GFX9-NEXT: v_add_u16_sdwa v2, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_add_u16_e32 v2, v2, v0
|
|
; GFX9-NEXT: v_add_u16_sdwa v0, v2, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: global_store_short v1, v0, s[2:3]
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: idot8_acc16_vecMul:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; GFX9-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-DL-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v2, 0xffff
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: global_load_dword v3, v0, s[4:5]
|
|
; GFX9-DL-NEXT: global_load_dword v4, v0, s[6:7]
|
|
; GFX9-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v10, 15, v3
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v17, 15, v4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v0, v3, 24, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v6, v3, 16, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v8, v3, 8, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v13, v4, 16, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v15, v4, 8, 4
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 28, v3
|
|
; GFX9-DL-NEXT: v_bfe_u32 v7, v3, 20, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v9, v3, 12, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v3, v3, 4, 4
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v10, v2, v10
|
|
; GFX9-DL-NEXT: v_bfe_u32 v11, v4, 24, 4
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v12, 28, v4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v14, v4, 20, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v16, v4, 12, 4
|
|
; GFX9-DL-NEXT: v_bfe_u32 v4, v4, 4, 4
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v17, v2, v17
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v3, v3, 16, v10
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v4, v4, 16, v17
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v3, 12, v3 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v4, 12, v4 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v3, 12, v3 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v4, 12, v4 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_mul_lo_u16 v3, v3, v4
|
|
; GFX9-DL-NEXT: global_load_ushort v4, v1, s[2:3]
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v8, v2, v8
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v0, v2, v0
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v15, v2, v15
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v0, v5, 16, v0
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v8, v9, 16, v8
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v5, v16, 16, v15
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v6, v2, v6
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v8, 12, v8 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v5, 12, v5 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v13, v2, v13
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v2, v2, v11
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v6, v7, 16, v6
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v7, v14, 16, v13
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v2, v12, 16, v2
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v8, 12, v8 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v5, 12, v5 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v6, 12, v6 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v0, 12, v0 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_lshlrev_b16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_mul_lo_u16 v5, v8, v5
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v6, 12, v6 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v0, 12, v0 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_ashrrev_i16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX9-DL-NEXT: v_pk_mul_lo_u16 v0, v0, v2
|
|
; GFX9-DL-NEXT: v_pk_mul_lo_u16 v2, v6, v7
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v4, v3, v4
|
|
; GFX9-DL-NEXT: v_add_u16_sdwa v3, v4, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v3, v3, v5
|
|
; GFX9-DL-NEXT: v_add_u16_sdwa v3, v3, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v3, v3, v2
|
|
; GFX9-DL-NEXT: v_add_u16_sdwa v2, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v2, v2, v0
|
|
; GFX9-DL-NEXT: v_add_u16_sdwa v0, v2, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: global_store_short v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-XNACK-LABEL: idot8_acc16_vecMul:
|
|
; GFX10-DL-XNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v4, 0xffff
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-XNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-XNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-XNACK-NEXT: global_load_ushort v3, v0, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v11, 15, v1
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v13, 15, v2
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v7, v1, 16, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v9, v1, 8, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v5, v1, 24, 4
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v6, 28, v1
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v8, v1, 20, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v10, v1, 12, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v1, v1, 4, 4
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v11, v4, v11
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v16, v2, 4, 4
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v13, v4, v13
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v18, v2, 8, 4
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v9, v4, v9
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v1, v1, 16, v11
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v7, v4, v7
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v11, v16, 16, v13
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v12, v2, 24, 4
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v14, 28, v2
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v1, 12, v1 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v15, v2, 16, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v17, v2, 20, 4
|
|
; GFX10-DL-XNACK-NEXT: v_bfe_u32 v2, v2, 12, 4
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v13, v4, v18
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v11, 12, v11 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v9, v10, 16, v9
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v7, v8, 16, v7
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v1, 12, v1 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v2, v2, 16, v13
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v8, 12, v11 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v9, 12, v9 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v10, v4, v15
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_pk_mul_lo_u16 v1, v1, v8
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v9, 12, v9 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v8, v17, 16, v10
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v10, 16, v1
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v3
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v3, v4, v5
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v5, 12, v8 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_and_b32_e32 v4, v4, v12
|
|
; GFX10-DL-XNACK-NEXT: v_pk_mul_lo_u16 v2, v9, v2
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v10
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v3, v6, 16, v3
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v5, 12, v5 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_lshl_or_b32 v4, v14, 16, v4
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v6, 16, v2
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v2
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v2, 12, v3 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_pk_lshlrev_b16 v3, 12, v4 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_pk_mul_lo_u16 v4, v7, v5
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v6
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_pk_ashrrev_i16 v3, 12, v3 op_sel_hi:[0,1]
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v5, 16, v4
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v4
|
|
; GFX10-DL-XNACK-NEXT: v_pk_mul_lo_u16 v2, v2, v3
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v5
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v3, 16, v2
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v2
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v1, v1, v3
|
|
; GFX10-DL-XNACK-NEXT: global_store_short v0, v1, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-NOXNACK-LABEL: idot8_acc16_vecMul:
|
|
; GFX10-DL-NOXNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v4, 0xffff
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NOXNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NOXNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v0, v0, s[6:7]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_ushort v3, v2, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v11, 15, v1
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v13, 15, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v7, v1, 16, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v9, v1, 8, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v5, v1, 24, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v6, 28, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v8, v1, 20, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v10, v1, 12, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v1, v1, 4, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v11, v4, v11
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v16, v0, 4, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v13, v4, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v18, v0, 8, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v9, v4, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v1, v1, 16, v11
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v7, v4, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v11, v16, 16, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v12, v0, 24, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v14, 28, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v1, 12, v1 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v15, v0, 16, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v17, v0, 20, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_bfe_u32 v0, v0, 12, 4
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v13, v4, v18
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v11, 12, v11 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v9, v10, 16, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v7, v8, 16, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v1, 12, v1 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v0, v0, 16, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v8, 12, v11 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v9, 12, v9 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v10, v4, v15
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v0, 12, v0 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_mul_lo_u16 v1, v1, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v9, 12, v9 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v8, v17, 16, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v7, 12, v7 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v0, 12, v0 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v10, 16, v1
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v1, v1, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v3, v4, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v5, 12, v8 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_and_b32_e32 v4, v4, v12
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_mul_lo_u16 v0, v9, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v1, v1, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v3, v6, 16, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v5, 12, v5 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshl_or_b32 v4, v14, 16, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v6, 16, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v1, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v1, 12, v3 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_lshlrev_b16 v3, 12, v4 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_mul_lo_u16 v4, v7, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v1, 12, v1 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_ashrrev_i16 v3, 12, v3 op_sel_hi:[0,1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v5, 16, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v4
|
|
; GFX10-DL-NOXNACK-NEXT: v_pk_mul_lo_u16 v1, v1, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v3, 16, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v3
|
|
; GFX10-DL-NOXNACK-NEXT: global_store_short v2, v0, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_endpgm
|
|
; GFX10-DL-LABEL: idot8_acc16_vecMul:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NEXT: s_mov_b32 s14, -1
|
|
; GFX10-DL-NEXT: s_mov_b32 s15, 0x31c16000
|
|
; GFX10-DL-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: global_load_ushort v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: s_load_dword s0, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_bfe_u32 s2, s0, 0x40018
|
|
; GFX10-DL-NEXT: s_lshr_b32 s3, s0, 28
|
|
; GFX10-DL-NEXT: s_bfe_u32 s6, s0, 0x40010
|
|
; GFX10-DL-NEXT: s_bfe_u32 s7, s0, 0x40014
|
|
; GFX10-DL-NEXT: s_bfe_u32 s8, s0, 0x40008
|
|
; GFX10-DL-NEXT: s_bfe_u32 s9, s0, 0x4000c
|
|
; GFX10-DL-NEXT: s_and_b32 s10, s0, 15
|
|
; GFX10-DL-NEXT: s_bfe_u32 s0, s0, 0x40004
|
|
; GFX10-DL-NEXT: s_and_b32 s11, s1, 15
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s0, s10, s0
|
|
; GFX10-DL-NEXT: s_bfe_u32 s10, s1, 0x40004
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v2, 12, s0 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s0, s11, s10
|
|
; GFX10-DL-NEXT: s_bfe_u32 s11, s1, 0x4000c
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v3, 12, s0 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: s_bfe_u32 s0, s1, 0x40008
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v2, 12, v2 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s8, s8, s9
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s0, s0, s11
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v3, 12, v3 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v4, 12, s8 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v5, 12, s0 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: s_bfe_u32 s8, s1, 0x40010
|
|
; GFX10-DL-NEXT: s_bfe_u32 s0, s1, 0x40014
|
|
; GFX10-DL-NEXT: v_pk_mul_lo_u16 v2, v2, v3
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v3, 12, v4 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v4, 12, v5 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s6, s6, s7
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s0, s8, s0
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v5, 12, s6 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v6, 12, s0 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_pk_mul_lo_u16 v3, v3, v4
|
|
; GFX10-DL-NEXT: s_bfe_u32 s10, s1, 0x40018
|
|
; GFX10-DL-NEXT: s_lshr_b32 s0, s1, 28
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s1, s2, s3
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v4, 12, v6 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: s_pack_ll_b32_b16 s0, s10, s0
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v6, 12, s0 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v1, v2, v1
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v2, 12, v5 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_pk_lshlrev_b16 v5, 12, s1 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
|
|
; GFX10-DL-NEXT: v_pk_mul_lo_u16 v2, v2, v4
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v4, 12, v6 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-DL-NEXT: v_pk_ashrrev_i16 v3, 12, v5 op_sel_hi:[0,1]
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v1, v1, v2
|
|
; GFX10-DL-NEXT: v_pk_mul_lo_u16 v3, v3, v4
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v1, v1, v3
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-DL-NEXT: global_store_short v0, v1, s[4:5]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<8 x i4> addrspace(1)* %src2,
|
|
i16 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %gep2
|
|
|
|
%cvec1 = sext <8 x i4> %vec1 to <8 x i16>
|
|
%cvec2 = sext <8 x i4> %vec2 to <8 x i16>
|
|
|
|
%mul = mul <8 x i16> %cvec1, %cvec2
|
|
%mul0 = extractelement <8 x i16> %mul, i64 0
|
|
%mul1 = extractelement <8 x i16> %mul, i64 1
|
|
%mul2 = extractelement <8 x i16> %mul, i64 2
|
|
%mul3 = extractelement <8 x i16> %mul, i64 3
|
|
%mul4 = extractelement <8 x i16> %mul, i64 4
|
|
%mul5 = extractelement <8 x i16> %mul, i64 5
|
|
%mul6 = extractelement <8 x i16> %mul, i64 6
|
|
%mul7 = extractelement <8 x i16> %mul, i64 7
|
|
|
|
%acc = load i16, i16 addrspace(1)* %dst, align 4
|
|
%add1 = add i16 %mul0, %acc
|
|
%add2 = add i16 %add1, %mul1
|
|
%add3 = add i16 %add2, %mul2
|
|
%add4 = add i16 %add3, %mul3
|
|
%add5 = add i16 %add4, %mul4
|
|
%add6 = add i16 %add5, %mul5
|
|
%add7 = add i16 %add6, %mul6
|
|
%add8 = add i16 %add7, %mul7
|
|
|
|
store i16 %add8, i16 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; TODO: Support this pattern.
|
|
define amdgpu_kernel void @idot8_acc8_vecMul(<8 x i4> addrspace(1)* %src1,
|
|
; GFX7-LABEL: idot8_acc8_vecMul:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; GFX7-NEXT: s_mov_b32 s12, SCRATCH_RSRC_DWORD0
|
|
; GFX7-NEXT: s_mov_b32 s13, SCRATCH_RSRC_DWORD1
|
|
; GFX7-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x9
|
|
; GFX7-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0xd
|
|
; GFX7-NEXT: s_mov_b32 s14, -1
|
|
; GFX7-NEXT: s_mov_b32 s15, 0xe8f000
|
|
; GFX7-NEXT: s_add_u32 s12, s12, s3
|
|
; GFX7-NEXT: s_mov_b32 s3, 0xf000
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX7-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX7-NEXT: buffer_load_dword v4, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[6:7]
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: v_mov_b32_e32 v2, 0xff
|
|
; GFX7-NEXT: s_mov_b32 s5, 0xffff
|
|
; GFX7-NEXT: v_mov_b32_e32 v3, 0xffff
|
|
; GFX7-NEXT: s_addc_u32 s13, s13, 0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_i32 v6, v4, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v7, v4, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v8, v4, 12, 4
|
|
; GFX7-NEXT: v_bfe_i32 v9, v4, 8, 4
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v6, 8, v6
|
|
; GFX7-NEXT: v_and_b32_e32 v7, s4, v7
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v8, 8, v8
|
|
; GFX7-NEXT: v_and_b32_e32 v9, s4, v9
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_bfe_i32 v13, v0, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v16, v0, 12, 4
|
|
; GFX7-NEXT: v_or_b32_e32 v6, v7, v6
|
|
; GFX7-NEXT: v_or_b32_e32 v7, v9, v8
|
|
; GFX7-NEXT: v_and_b32_e32 v9, v2, v13
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v13, 8, v16
|
|
; GFX7-NEXT: buffer_load_ubyte v16, off, s[0:3], 0
|
|
; GFX7-NEXT: v_bfe_i32 v5, v4, 24, 4
|
|
; GFX7-NEXT: v_bfe_i32 v10, v4, 4, 4
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v1, 28, v4
|
|
; GFX7-NEXT: v_bfe_i32 v4, v4, 0, 4
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v11, 8, v1
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v5
|
|
; GFX7-NEXT: v_ashrrev_i32_e32 v12, 28, v0
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v10, 8, v10
|
|
; GFX7-NEXT: v_and_b32_e32 v4, v2, v4
|
|
; GFX7-NEXT: v_bfe_i32 v14, v0, 20, 4
|
|
; GFX7-NEXT: v_bfe_i32 v15, v0, 16, 4
|
|
; GFX7-NEXT: v_bfe_i32 v17, v0, 8, 4
|
|
; GFX7-NEXT: v_bfe_i32 v18, v0, 4, 4
|
|
; GFX7-NEXT: v_bfe_i32 v0, v0, 0, 4
|
|
; GFX7-NEXT: v_or_b32_e32 v5, v5, v11
|
|
; GFX7-NEXT: v_or_b32_e32 v4, v4, v10
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v8, 8, v12
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v10, 8, v14
|
|
; GFX7-NEXT: v_and_b32_e32 v11, v2, v15
|
|
; GFX7-NEXT: v_and_b32_e32 v14, v2, v17
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v15, 8, v18
|
|
; GFX7-NEXT: v_and_b32_e32 v0, v2, v0
|
|
; GFX7-NEXT: v_or_b32_e32 v0, v0, v15
|
|
; GFX7-NEXT: v_or_b32_e32 v8, v9, v8
|
|
; GFX7-NEXT: v_or_b32_e32 v9, v11, v10
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v5, 16, v5
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s5, v6
|
|
; GFX7-NEXT: v_or_b32_e32 v10, v14, v13
|
|
; GFX7-NEXT: v_or_b32_e32 v5, v6, v5
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v6, 16, v8
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v8, 16, v10
|
|
; GFX7-NEXT: v_and_b32_e32 v0, v3, v0
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v7, 16, v7
|
|
; GFX7-NEXT: v_and_b32_e32 v4, s5, v4
|
|
; GFX7-NEXT: v_or_b32_e32 v0, v0, v8
|
|
; GFX7-NEXT: v_or_b32_e32 v4, v4, v7
|
|
; GFX7-NEXT: v_and_b32_e32 v7, v3, v9
|
|
; GFX7-NEXT: v_or_b32_e32 v3, v7, v6
|
|
; GFX7-NEXT: v_and_b32_e32 v7, v2, v4
|
|
; GFX7-NEXT: v_and_b32_e32 v13, v2, v0
|
|
; GFX7-NEXT: v_bfe_u32 v8, v4, 8, 8
|
|
; GFX7-NEXT: v_bfe_u32 v14, v0, 8, 8
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v6, 24, v4
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v11, 24, v0
|
|
; GFX7-NEXT: v_bfe_u32 v4, v4, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v0, v0, 16, 8
|
|
; GFX7-NEXT: v_and_b32_e32 v1, v2, v1
|
|
; GFX7-NEXT: v_and_b32_e32 v12, v2, v12
|
|
; GFX7-NEXT: v_and_b32_e32 v9, v2, v5
|
|
; GFX7-NEXT: v_and_b32_e32 v2, v2, v3
|
|
; GFX7-NEXT: v_bfe_u32 v10, v5, 8, 8
|
|
; GFX7-NEXT: v_bfe_u32 v15, v3, 8, 8
|
|
; GFX7-NEXT: v_bfe_u32 v5, v5, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v3, v3, 16, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v7, v7, v13, v16
|
|
; GFX7-NEXT: v_mad_u32_u24 v7, v8, v14, v7
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v4, v0, v7
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v6, v11, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v9, v2, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v10, v15, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v5, v3, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v1, v12, v0
|
|
; GFX7-NEXT: buffer_store_byte v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: idot8_acc8_vecMul:
|
|
; GFX8: ; %bb.0: ; %entry
|
|
; GFX8-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX8-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 2, v0
|
|
; GFX8-NEXT: v_mov_b32_e32 v5, 12
|
|
; GFX8-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s5
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s4, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v3, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v2, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v0, s0
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s1
|
|
; GFX8-NEXT: flat_load_ubyte v4, v[0:1]
|
|
; GFX8-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX8-NEXT: s_mov_b32 s10, -1
|
|
; GFX8-NEXT: s_mov_b32 s11, 0xe80000
|
|
; GFX8-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX8-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 28, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 12, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 8, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 20, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v12, 28, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v13, 12, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v14, 8, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v10, 4, v3
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v11, 20, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v15, 4, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v16, 12, v3
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v17, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v3, v5, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v18, 12, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v19, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_lshlrev_b16_sdwa v2, v5, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v5, 12, v10
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v10, 12, v16
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v16, 12, v17
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v17, 12, v3
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v3, 12, v6
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v6, 12, v15
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v15, 12, v18
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v18, 12, v19
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v19, 12, v2
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v2, 12, v11
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_lshlrev_b16_e32 v14, 12, v14
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v3, 12, v3
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v11, 12, v14
|
|
; GFX8-NEXT: v_mul_lo_u16_e32 v10, v10, v15
|
|
; GFX8-NEXT: v_mul_lo_u16_sdwa v2, v3, v2 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v9, 12, v9
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v13, 12, v13
|
|
; GFX8-NEXT: v_mul_lo_u16_sdwa v3, v7, v12 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX8-NEXT: v_mul_lo_u16_e32 v15, v16, v18
|
|
; GFX8-NEXT: v_mul_lo_u16_sdwa v7, v8, v13 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX8-NEXT: v_mul_lo_u16_e32 v8, v9, v11
|
|
; GFX8-NEXT: v_or_b32_sdwa v3, v15, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX8-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX8-NEXT: v_mul_lo_u16_e32 v14, v17, v19
|
|
; GFX8-NEXT: v_mul_lo_u16_sdwa v5, v5, v6 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX8-NEXT: v_or_b32_sdwa v7, v8, v7 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX8-NEXT: v_or_b32_sdwa v6, v14, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v9, 16, v3
|
|
; GFX8-NEXT: v_or_b32_sdwa v8, v10, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v2, 16, v7
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v10, 8, v3
|
|
; GFX8-NEXT: v_or_b32_sdwa v3, v6, v9 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX8-NEXT: v_or_b32_e32 v5, v5, v2
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 8, v3
|
|
; GFX8-NEXT: v_lshrrev_b64 v[2:3], 24, v[2:3]
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v5, 8, v5
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_add_u16_e32 v3, v8, v4
|
|
; GFX8-NEXT: v_add_u16_e32 v3, v3, v5
|
|
; GFX8-NEXT: v_add_u16_e32 v3, v3, v7
|
|
; GFX8-NEXT: v_add_u16_e32 v2, v3, v2
|
|
; GFX8-NEXT: v_mad_u16 v2, v17, v19, v2
|
|
; GFX8-NEXT: v_add_u16_e32 v2, v2, v6
|
|
; GFX8-NEXT: v_mad_u16 v2, v16, v18, v2
|
|
; GFX8-NEXT: v_add_u16_e32 v2, v2, v10
|
|
; GFX8-NEXT: flat_store_byte v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-LABEL: idot8_acc8_vecMul:
|
|
; GFX9: ; %bb.0: ; %entry
|
|
; GFX9-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NEXT: v_mov_b32_e32 v3, 0
|
|
; GFX9-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NEXT: global_load_ubyte v4, v3, s[2:3]
|
|
; GFX9-NEXT: v_mov_b32_e32 v0, 12
|
|
; GFX9-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v6, 28, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v7, 12, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v8, 8, v1
|
|
; GFX9-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v11, 28, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v5, 20, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v9, 4, v1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v10, 20, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v14, 4, v2
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v15, 12, v1
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v16, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v1, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v18, v0, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v17, 12, v2
|
|
; GFX9-NEXT: v_lshlrev_b16_sdwa v0, v0, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v12, 12, v2
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v13, 8, v2
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v2, 12, v9
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v9, 12, v15
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v15, 12, v16
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v16, 12, v1
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v1, 12, v5
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v5, 12, v14
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v14, 12, v17
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v18, 12, v0
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v0, 12, v10
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX9-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v1, 12, v1
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v0, 12, v0
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v10, 12, v13
|
|
; GFX9-NEXT: v_mul_lo_u16_sdwa v0, v1, v0 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX9-NEXT: v_mul_lo_u16_sdwa v1, v6, v11 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-NEXT: v_mul_lo_u16_e32 v19, v15, v17
|
|
; GFX9-NEXT: v_mul_lo_u16_sdwa v6, v7, v12 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-NEXT: v_mul_lo_u16_e32 v7, v8, v10
|
|
; GFX9-NEXT: v_or_b32_sdwa v1, v19, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX9-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX9-NEXT: v_mul_lo_u16_e32 v13, v16, v18
|
|
; GFX9-NEXT: v_mul_lo_u16_sdwa v2, v2, v5 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-NEXT: v_mul_lo_u16_e32 v9, v9, v14
|
|
; GFX9-NEXT: v_or_b32_sdwa v6, v7, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-NEXT: v_or_b32_sdwa v5, v13, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v8, 16, v1
|
|
; GFX9-NEXT: v_or_b32_sdwa v7, v9, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 16, v6
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX9-NEXT: v_or_b32_sdwa v1, v5, v8 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX9-NEXT: v_or_b32_e32 v2, v2, v0
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v5, 8, v1
|
|
; GFX9-NEXT: v_lshrrev_b64 v[0:1], 24, v[0:1]
|
|
; GFX9-NEXT: v_lshrrev_b32_e32 v2, 8, v2
|
|
; GFX9-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NEXT: v_add_u16_e32 v1, v7, v4
|
|
; GFX9-NEXT: v_add_u16_e32 v1, v1, v2
|
|
; GFX9-NEXT: v_add_u16_e32 v1, v1, v6
|
|
; GFX9-NEXT: v_add_u16_e32 v0, v1, v0
|
|
; GFX9-NEXT: v_mad_legacy_u16 v0, v16, v18, v0
|
|
; GFX9-NEXT: v_add_u16_e32 v0, v0, v5
|
|
; GFX9-NEXT: v_mad_legacy_u16 v0, v15, v17, v0
|
|
; GFX9-NEXT: v_add_u16_e32 v0, v0, v9
|
|
; GFX9-NEXT: global_store_byte v3, v0, s[2:3]
|
|
; GFX9-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: idot8_acc8_vecMul:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; GFX9-DL-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX9-DL-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX9-DL-NEXT: s_mov_b32 s10, -1
|
|
; GFX9-DL-NEXT: s_mov_b32 s11, 0xe00000
|
|
; GFX9-DL-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX9-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v3, 0
|
|
; GFX9-DL-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-DL-NEXT: global_load_ubyte v4, v3, s[2:3]
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v0, 12
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 28, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 12, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v8, 8, v1
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v11, 28, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 20, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v9, 4, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v10, 20, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v14, 4, v2
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v15, 12, v1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v16, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v1, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v18, v0, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v17, 12, v2
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_sdwa v0, v0, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v12, 12, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v13, 8, v2
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v2, 12, v9
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v6, 12, v6
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v11, 12, v11
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v9, 12, v15
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v15, 12, v16
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v16, 12, v1
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v1, 12, v5
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v5, 12, v14
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v14, 12, v17
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v17, 12, v18
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v18, 12, v0
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v0, 12, v10
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v8, 12, v8
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v7, 12, v7
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v12, 12, v12
|
|
; GFX9-DL-NEXT: v_lshlrev_b16_e32 v13, 12, v13
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v6, 12, v6
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v1, 12, v1
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v0, 12, v0
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v11, 12, v11
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v10, 12, v13
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_sdwa v0, v1, v0 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v8, 12, v8
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v7, 12, v7
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v12, 12, v12
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_sdwa v1, v6, v11 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_e32 v19, v15, v17
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_sdwa v6, v7, v12 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_e32 v7, v8, v10
|
|
; GFX9-DL-NEXT: v_or_b32_sdwa v1, v19, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v2, 12, v2
|
|
; GFX9-DL-NEXT: v_ashrrev_i16_e32 v5, 12, v5
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_e32 v13, v16, v18
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_sdwa v2, v2, v5 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_e32 v9, v9, v14
|
|
; GFX9-DL-NEXT: v_or_b32_sdwa v6, v7, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_or_b32_sdwa v5, v13, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v8, 16, v1
|
|
; GFX9-DL-NEXT: v_or_b32_sdwa v7, v9, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v0, 16, v6
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX9-DL-NEXT: v_or_b32_sdwa v1, v5, v8 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_or_b32_e32 v2, v2, v0
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v1
|
|
; GFX9-DL-NEXT: v_lshrrev_b64 v[0:1], 24, v[0:1]
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v2, 8, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v1, v7, v4
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v1, v1, v2
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v1, v1, v6
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v0, v1, v0
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v16, v18, v0
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v0, v0, v5
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v15, v17, v0
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v0, v0, v9
|
|
; GFX9-DL-NEXT: global_store_byte v3, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-XNACK-LABEL: idot8_acc8_vecMul:
|
|
; GFX10-DL-XNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-XNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-XNACK-NEXT: v_mov_b32_e32 v4, 0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-XNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-XNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-XNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-XNACK-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-XNACK-NEXT: global_load_ubyte v3, v4, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v8, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v15, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v16, 8, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v6, 28, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v8, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v15, 12, v15
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v13, 28, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v10, 4, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v17, 4, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v9, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v16, 12, v16
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v8, 12, v8
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v15, 12, v15
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v0, 20, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v7, 24, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v11, 20, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v14, 24, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v6, 12, v6
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v13, 12, v13
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v10, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v8, v8, v15
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v17, 12, v17
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v9, 12, v9
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v16, 12, v16
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v12, 16, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v7, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v0, 12, v0
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v14, 12, v14
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v11, 12, v11
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v6, 12, v6
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v13, 12, v13
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v1, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v2, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v10, 12, v10
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v15, 12, v17
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v9, v9, v16
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v8, 8, v8
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v5, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v12, 12, v12
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v6, v6, v13
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v7, 12, v7
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v0, 12, v0
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v14, 12, v14
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v11, 12, v11
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v10, v10, v15
|
|
; GFX10-DL-XNACK-NEXT: v_or_b32_sdwa v8, v9, v8 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v1, 12, v1
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v2, 12, v2
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v9, v0, v11
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v5, 12, v5
|
|
; GFX10-DL-XNACK-NEXT: v_ashrrev_i16 v12, 12, v12
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v11, v7, v14
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v6, 8, v6
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v1, v1, v2
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v10, 8, v10
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v0, 16, v8
|
|
; GFX10-DL-XNACK-NEXT: v_mul_lo_u16 v2, v5, v12
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b16 v9, 8, v9
|
|
; GFX10-DL-XNACK-NEXT: v_or_b32_sdwa v6, v11, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-XNACK-NEXT: v_or_b32_sdwa v1, v1, v10 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-XNACK-NEXT: v_or_b32_sdwa v11, v10, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX10-DL-XNACK-NEXT: v_or_b32_sdwa v2, v2, v9 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-XNACK-NEXT: v_lshlrev_b32_e32 v9, 16, v6
|
|
; GFX10-DL-XNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v3, v1, v3
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v10, 8, v11
|
|
; GFX10-DL-XNACK-NEXT: v_or_b32_sdwa v1, v2, v9 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v9, v3, v10
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b64 v[2:3], 24, v[0:1]
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v1, 8, v1
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v0, v9, v8
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v0, v0, v2
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v0, v5, v12, v0
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-DL-XNACK-NEXT: v_lshrrev_b32_e32 v1, 8, v6
|
|
; GFX10-DL-XNACK-NEXT: v_mad_u16 v0, v7, v14, v0
|
|
; GFX10-DL-XNACK-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-DL-XNACK-NEXT: global_store_byte v4, v0, s[0:1]
|
|
; GFX10-DL-XNACK-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-NOXNACK-LABEL: idot8_acc8_vecMul:
|
|
; GFX10-DL-NOXNACK: ; %bb.0: ; %entry
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NOXNACK-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NOXNACK-NEXT: v_mov_b32_e32 v4, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s9, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s10, -1
|
|
; GFX10-DL-NOXNACK-NEXT: s_mov_b32 s11, 0x31c16000
|
|
; GFX10-DL-NOXNACK-NEXT: s_add_u32 s8, s8, s3
|
|
; GFX10-DL-NOXNACK-NEXT: s_addc_u32 s9, s9, 0
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: s_clause 0x1
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_dword v0, v0, s[6:7]
|
|
; GFX10-DL-NOXNACK-NEXT: global_load_ubyte v2, v4, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v8, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v15, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v9, 8, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v11, 20, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v12, 16, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v13, 28, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v14, 24, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v17, 4, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v8, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v15, 12, v15
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v16, 8, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v0, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v6, 28, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v10, 4, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v9, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v8, 12, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v18, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v0, 12, v16
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v15, 12, v15
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v3, 20, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v7, 24, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v6, 12, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v13, 12, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v10, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v8, v8, v15
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v17, 12, v17
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v9, 12, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v0, 12, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v7, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v3, 12, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v14, 12, v14
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v11, 12, v11
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v6, 12, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v13, 12, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v1, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v0, v9, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v8, 8, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v10, 12, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v15, 12, v17
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v5, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v6, v6, v13
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v12, 12, v12
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v9, 12, v11
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v3, 12, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v7, 12, v7
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v14, 12, v14
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v1, 12, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v10, v10, v15
|
|
; GFX10-DL-NOXNACK-NEXT: v_or_b32_sdwa v8, v0, v8 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v3, v3, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v5, 12, v5
|
|
; GFX10-DL-NOXNACK-NEXT: v_ashrrev_i16 v11, 12, v12
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v9, v7, v14
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v6, 8, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v10, 8, v10
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v1, v1, v18
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v0, 16, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_mul_lo_u16 v12, v5, v11
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b16 v3, 8, v3
|
|
; GFX10-DL-NOXNACK-NEXT: v_or_b32_sdwa v6, v9, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NOXNACK-NEXT: v_or_b32_sdwa v1, v1, v10 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NOXNACK-NEXT: v_or_b32_sdwa v9, v10, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX10-DL-NOXNACK-NEXT: v_or_b32_sdwa v3, v12, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshlrev_b32_e32 v10, 16, v6
|
|
; GFX10-DL-NOXNACK-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v2, v1, v2
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v9, 8, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_or_b32_sdwa v1, v3, v10 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v9, v2, v9
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b64 v[2:3], 24, v[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v1, 8, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v9, v8
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v2
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v5, v11, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-DL-NOXNACK-NEXT: v_lshrrev_b32_e32 v1, 8, v6
|
|
; GFX10-DL-NOXNACK-NEXT: v_mad_u16 v0, v7, v14, v0
|
|
; GFX10-DL-NOXNACK-NEXT: v_add_nc_u16 v0, v0, v1
|
|
; GFX10-DL-NOXNACK-NEXT: global_store_byte v4, v0, s[0:1]
|
|
; GFX10-DL-NOXNACK-NEXT: s_endpgm
|
|
; GFX10-DL-LABEL: idot8_acc8_vecMul:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[4:5], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: s_mov_b32 s20, SCRATCH_RSRC_DWORD0
|
|
; GFX10-DL-NEXT: s_mov_b32 s21, SCRATCH_RSRC_DWORD1
|
|
; GFX10-DL-NEXT: s_mov_b32 s22, -1
|
|
; GFX10-DL-NEXT: s_mov_b32 s23, 0x31c16000
|
|
; GFX10-DL-NEXT: s_add_u32 s20, s20, s3
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[0:3], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_addc_u32 s21, s21, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: global_load_ubyte v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: s_load_dword s0, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX10-DL-NEXT: s_mov_b32 s2, 0xffff
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_lshr_b32 s9, s0, 4
|
|
; GFX10-DL-NEXT: s_lshr_b32 s16, s1, 4
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v6, 12, s9
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v12, 12, s16
|
|
; GFX10-DL-NEXT: s_lshr_b32 s10, s0, 12
|
|
; GFX10-DL-NEXT: s_lshr_b32 s17, s1, 12
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v2, 12, s0
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v3, 12, s1
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v13, 12, s17
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v5, 12, s10
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v6, 12, v6
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v12, 12, v12
|
|
; GFX10-DL-NEXT: s_lshr_b32 s11, s0, 8
|
|
; GFX10-DL-NEXT: s_lshr_b32 s18, s1, 8
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v4, 12, s11
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v11, 12, s18
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v2, 12, v2
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v3, 12, v3
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v19, 12, v5
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v6, v6, v12
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v13, 12, v13
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v4, 12, v4
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v11, 12, v11
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v2, v2, v3
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v5, 8, v6
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v3, v19, v13
|
|
; GFX10-DL-NEXT: s_lshr_b32 s3, s0, 20
|
|
; GFX10-DL-NEXT: s_lshr_b32 s6, s0, 16
|
|
; GFX10-DL-NEXT: s_lshr_b32 s7, s0, 28
|
|
; GFX10-DL-NEXT: s_lshr_b32 s8, s0, 24
|
|
; GFX10-DL-NEXT: s_lshr_b32 s12, s1, 20
|
|
; GFX10-DL-NEXT: v_or_b32_sdwa v2, v2, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v7, 12, s8
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v8, 12, s7
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v9, 12, s6
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v10, 12, s3
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v12, 12, s12
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v4, v4, v11
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v3, 8, v3
|
|
; GFX10-DL-NEXT: s_lshr_b32 s13, s1, 16
|
|
; GFX10-DL-NEXT: s_lshr_b32 s14, s1, 28
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v6, 12, s13
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v5, 12, v7
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v7, 12, v8
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v8, 12, v9
|
|
; GFX10-DL-NEXT: v_or_b32_sdwa v3, v4, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v2, s2, v2
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v15, 12, s14
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v4, 12, v10
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v9, 12, v12
|
|
; GFX10-DL-NEXT: s_lshr_b32 s15, s1, 24
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v6, 12, v6
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v10, 12, v15
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v14, 12, s15
|
|
; GFX10-DL-NEXT: v_or_b32_e32 v3, v2, v3
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v4, v4, v9
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v15, v8, v6
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v7, v7, v10
|
|
; GFX10-DL-NEXT: v_ashrrev_i16_e64 v11, 12, v14
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v8, 8, v3
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v1, v2, v1
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v2, 8, v4
|
|
; GFX10-DL-NEXT: v_mul_lo_u16_e64 v4, v5, v11
|
|
; GFX10-DL-NEXT: v_lshlrev_b16_e64 v5, 8, v7
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v1, v1, v8
|
|
; GFX10-DL-NEXT: v_or_b32_sdwa v2, v15, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_or_b32_sdwa v4, v4, v5 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_2
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v2, s2, v2
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX10-DL-NEXT: v_or_b32_e32 v3, v2, v4
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v1, v1, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 8, v3
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v1, v1, v2
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX10-DL-NEXT: v_add_nc_u32_sdwa v1, v1, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
|
|
; GFX10-DL-NEXT: global_store_byte v0, v1, s[4:5]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<8 x i4> addrspace(1)* %src2,
|
|
i8 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <8 x i4>, <8 x i4> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %gep2
|
|
|
|
%cvec1 = sext <8 x i4> %vec1 to <8 x i8>
|
|
%cvec2 = sext <8 x i4> %vec2 to <8 x i8>
|
|
|
|
%mul = mul <8 x i8> %cvec1, %cvec2
|
|
%mul0 = extractelement <8 x i8> %mul, i64 0
|
|
%mul1 = extractelement <8 x i8> %mul, i64 1
|
|
%mul2 = extractelement <8 x i8> %mul, i64 2
|
|
%mul3 = extractelement <8 x i8> %mul, i64 3
|
|
%mul4 = extractelement <8 x i8> %mul, i64 4
|
|
%mul5 = extractelement <8 x i8> %mul, i64 5
|
|
%mul6 = extractelement <8 x i8> %mul, i64 6
|
|
%mul7 = extractelement <8 x i8> %mul, i64 7
|
|
|
|
%acc = load i8, i8 addrspace(1)* %dst, align 4
|
|
%add1 = add i8 %mul0, %acc
|
|
%add2 = add i8 %add1, %mul1
|
|
%add3 = add i8 %add2, %mul2
|
|
%add4 = add i8 %add3, %mul3
|
|
%add5 = add i8 %add4, %mul4
|
|
%add6 = add i8 %add5, %mul5
|
|
%add7 = add i8 %add6, %mul6
|
|
%add8 = add i8 %add7, %mul7
|
|
|
|
store i8 %add8, i8 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
declare i32 @llvm.amdgcn.workitem.id.x()
|