mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-22 18:54:02 +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
2273 lines
98 KiB
LLVM
2273 lines
98 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-NODL %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 %s
|
|
; RUN: llc -mtriple=amdgcn -mcpu=gfx1012 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX10-DL %s
|
|
|
|
define amdgpu_kernel void @udot4_acc32(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_acc32:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: s_load_dword s4, s[0:1], 0x0
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v5, s4
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v3, v6, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v7, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_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_movk_i32 s2, 0xff
|
|
; 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 s3, s[0:1], 0x0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_and_b32_e32 v1, s2, v3
|
|
; GFX8-NEXT: v_bfe_u32 v4, v3, 8, 8
|
|
; GFX8-NEXT: v_bfe_u32 v6, v3, 16, 8
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_and_b32_e32 v2, s2, v0
|
|
; GFX8-NEXT: v_bfe_u32 v5, v0, 8, 8
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mad_u32_u24 v1, v1, v2, s3
|
|
; GFX8-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX8-NEXT: v_mad_u32_u24 v1, v4, v5, v1
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX8-NEXT: v_mad_u32_u24 v1, v6, v7, v1
|
|
; GFX8-NEXT: v_mad_u32_u24 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-NODL-LABEL: udot4_acc32:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v4, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v5, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: v_add3_u32 v2, v3, s0, v4
|
|
; GFX9-NODL-NEXT: v_add3_u32 v1, v2, v5, v1
|
|
; GFX9-NODL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_acc32:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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_dot4_u32_u8 v0, v2, v3, s0
|
|
; GFX9-DL-NEXT: global_store_dword v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_acc32:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0) lgkmcnt(0)
|
|
; GFX10-DL-NEXT: v_dot4_u32_u8 v1, v1, v2, s2
|
|
; GFX10-DL-NEXT: global_store_dword v0, v1, s[0:1]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i32 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%cv1e0 = zext i8 %v1e0 to i32
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%cv2e0 = zext i8 %v2e0 to i32
|
|
%mul1 = mul nuw nsw i32 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%cv1e1 = zext i8 %v1e1 to i32
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%cv2e1 = zext i8 %v2e1 to i32
|
|
%mul2 = mul nuw nsw i32 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%cv1e2 = zext i8 %v1e2 to i32
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%cv2e2 = zext i8 %v2e2 to i32
|
|
%mul3 = mul nuw nsw i32 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%cv1e3 = zext i8 %v1e3 to i32
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%cv2e3 = zext i8 %v2e3 to i32
|
|
%mul4 = mul nuw nsw i32 %cv1e3, %cv2e3
|
|
|
|
%acc = load i32, i32 addrspace(1)* %dst, align 4
|
|
%mad1 = add i32 %mul1, %acc
|
|
%mad2 = add i32 %mad1, %mul2
|
|
%mad3 = add i32 %mad2, %mul3
|
|
%mad4 = add i32 %mad3, %mul4
|
|
|
|
store i32 %mad4, i32 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @udot4_acc16(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_acc16:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_ushort v8, off, s[0:3], 0
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v5, v8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v3, v6, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v7, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_short v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_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: 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 v4, v[0:1]
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s1
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: v_mov_b32_e32 v2, s0
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: flat_load_ushort v1, v[2:3]
|
|
; GFX8-NEXT: s_movk_i32 s0, 0xff
|
|
; GFX8-NEXT: v_mov_b32_e32 v5, s0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 8, v4
|
|
; GFX8-NEXT: v_and_b32_e32 v6, s0, v4
|
|
; GFX8-NEXT: v_and_b32_e32 v8, s0, v8
|
|
; GFX8-NEXT: v_and_b32_sdwa v10, v4, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v4, 24, v4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 8, v0
|
|
; GFX8-NEXT: v_and_b32_e32 v7, s0, v0
|
|
; GFX8-NEXT: v_and_b32_e32 v9, s0, v9
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v1, v6, v7, v1
|
|
; GFX8-NEXT: v_and_b32_sdwa v5, v0, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX8-NEXT: v_mad_u16 v1, v8, v9, v1
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX8-NEXT: v_mad_u16 v1, v10, v5, v1
|
|
; GFX8-NEXT: v_mad_u16 v0, v4, v0, v1
|
|
; GFX8-NEXT: flat_store_short v[2:3], v0
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: udot4_acc16:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-NODL-NEXT: s_movk_i32 s0, 0xff
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: global_load_ushort v4, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v6, 8, v2
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v7, 8, v3
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v0, s0, v2
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v5, s0, v3
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v6, s0, v6
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v7, s0, v7
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v0, v5, v4
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v8, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v9, v3, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v6, v7, v0
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v8, v9, v0
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v2, v3, v0
|
|
; GFX9-NODL-NEXT: global_store_short v1, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_acc16:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_movk_i32 s0, 0xff
|
|
; 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: global_load_ushort v4, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 8, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 8, v3
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v0, s0, v2
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v5, s0, v3
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v6, s0, v6
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v7, s0, v7
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v0, v5, v4
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v8, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v9, v3, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v6, v7, v0
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v8, v9, v0
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v2, v3, v0
|
|
; GFX9-DL-NEXT: global_store_short v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_acc16:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_movk_i32 s0, 0xff
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: global_load_ushort v3, v0, s[2:3]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v1
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v6, 8, v2
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v4, s0, v1
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v7, s0, v2
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v5, s0, v5
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v6, s0, v6
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u16 v3, v4, v7, v3
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v4, v1, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v7, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v1, 24, v1
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX10-DL-NEXT: v_mad_u16 v3, v5, v6, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v3, v4, v7, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v1, v1, v2, v3
|
|
; GFX10-DL-NEXT: global_store_short v0, v1, s[2:3]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i16 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%cv1e0 = zext i8 %v1e0 to i16
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%cv2e0 = zext i8 %v2e0 to i16
|
|
%mul1 = mul nuw nsw i16 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%cv1e1 = zext i8 %v1e1 to i16
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%cv2e1 = zext i8 %v2e1 to i16
|
|
%mul2 = mul nuw nsw i16 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%cv1e2 = zext i8 %v1e2 to i16
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%cv2e2 = zext i8 %v2e2 to i16
|
|
%mul3 = mul nuw nsw i16 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%cv1e3 = zext i8 %v1e3 to i16
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%cv2e3 = zext i8 %v2e3 to i16
|
|
%mul4 = mul nuw nsw i16 %cv1e3, %cv2e3
|
|
|
|
%acc = load i16, i16 addrspace(1)* %dst, align 2
|
|
%mad1 = add i16 %mul1, %acc
|
|
%mad2 = add i16 %mad1, %mul2
|
|
%mad3 = add i16 %mad2, %mul3
|
|
%mad4 = add i16 %mad3, %mul4
|
|
|
|
store i16 %mad4, i16 addrspace(1)* %dst, align 2
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @udot4_acc8(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_acc8:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_ubyte v8, off, s[0:3], 0
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v5, v8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v3, v6, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v7, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_byte v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_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: 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 v4, v[0:1]
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s1
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_mov_b32_e32 v2, s0
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_ubyte v5, v[2:3]
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 8, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v1, 16, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 24, v4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 16, v0
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 8, v0
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v10, 24, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v4, v0, v5
|
|
; GFX8-NEXT: v_mad_u16 v0, v7, v8, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v1, v6, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v9, v10, v0
|
|
; GFX8-NEXT: flat_store_byte v[2:3], v0
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: udot4_acc8:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v0, 16, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v6, 8, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v8, 24, v2
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v7, 8, v3
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v2, v2, v3, v4
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v5, 16, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v2, v6, v7, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v9, 24, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v0, v5, v2
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v8, v9, v0
|
|
; GFX9-NODL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_acc8:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v0, 16, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 8, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v8, 24, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 8, v3
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v2, v2, v3, v4
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 16, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v2, v6, v7, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v9, 24, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v0, v5, v2
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v8, v9, v0
|
|
; GFX9-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_acc8:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX10-DL-NEXT: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v0, 8, v2
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v3
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u16 v4, v2, v3, v4
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v6, 16, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v7, 16, v3
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v0, v5, v4
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v6, v7, v0
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v2, v3, v0
|
|
; GFX10-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i8 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%mul1 = mul nuw nsw i8 %v1e0, %v2e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%mul2 = mul nuw nsw i8 %v1e1, %v2e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%mul3 = mul nuw nsw i8 %v1e2, %v2e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%mul4 = mul nuw nsw i8 %v1e3, %v2e3
|
|
|
|
%acc = load i8, i8 addrspace(1)* %dst, align 2
|
|
%mad1 = add i8 %mul1, %acc
|
|
%mad2 = add i8 %mad1, %mul2
|
|
%mad3 = add i8 %mad2, %mul3
|
|
%mad4 = add i8 %mad3, %mul4
|
|
|
|
store i8 %mad4, i8 addrspace(1)* %dst, align 2
|
|
ret void
|
|
}
|
|
|
|
; TODO: Generate udot4?
|
|
define amdgpu_kernel void @udot2_8(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot2_8:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: buffer_load_ubyte v4, off, s[0:3], 0
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: v_bfe_u32 v2, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v3, s4, v0
|
|
; GFX7-NEXT: v_bfe_u32 v0, v0, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v3, v4
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_byte v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot2_8:
|
|
; 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_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 v4, v[0:1]
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s1
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_mov_b32_e32 v2, s0
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_ubyte v5, v[2:3]
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v1, 8, v4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 8, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v4, v0, v5
|
|
; GFX8-NEXT: v_mad_u16 v0, v1, v6, v0
|
|
; GFX8-NEXT: flat_store_byte v[2:3], v0
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: udot2_8:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v0, 8, v2
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v5, 8, v3
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v2, v2, v3, v4
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v0, v5, v2
|
|
; GFX9-NODL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot2_8:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v0, 8, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v3
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v2, v2, v3, v4
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v0, v5, v2
|
|
; GFX9-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot2_8:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX10-DL-NEXT: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v0, 8, v2
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v3
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u16 v2, v2, v3, v4
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v0, v5, v2
|
|
; GFX10-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i8 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%mul1 = mul nuw nsw i8 %v1e0, %v2e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%mul2 = mul nuw nsw i8 %v1e1, %v2e1
|
|
|
|
%acc = load i8, i8 addrspace(1)* %dst, align 2
|
|
%mad1 = add i8 %mul1, %acc
|
|
%mad2 = add i8 %mad1, %mul2
|
|
store i8 %mad2, i8 addrspace(1)* %dst, align 2
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @udot4_CommutationInsideMAD(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_CommutationInsideMAD:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_ubyte v8, off, s[0:3], 0
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v5, v1, v8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v6, v3, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v7, v4, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v0, v2, v1
|
|
; GFX7-NEXT: buffer_store_byte v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_CommutationInsideMAD:
|
|
; 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_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 v4, v[0:1]
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s1
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_mov_b32_e32 v2, s0
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_ubyte v5, v[2:3]
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 8, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v1, 16, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 24, v4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 16, v0
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 8, v0
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v10, 24, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v0, v4, v5
|
|
; GFX8-NEXT: v_mad_u16 v0, v8, v7, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v6, v1, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v10, v9, v0
|
|
; GFX8-NEXT: flat_store_byte v[2:3], v0
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: udot4_CommutationInsideMAD:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v0, 16, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v6, 8, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v8, 24, v2
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v7, 8, v3
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v2, v3, v2, v4
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v5, 16, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v2, v7, v6, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v9, 24, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v5, v0, v2
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v9, v8, v0
|
|
; GFX9-NODL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_CommutationInsideMAD:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v0, 16, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 8, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v8, 24, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 8, v3
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v2, v3, v2, v4
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 16, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v2, v7, v6, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v9, 24, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v5, v0, v2
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v9, v8, v0
|
|
; GFX9-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_CommutationInsideMAD:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX10-DL-NEXT: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v0, 8, v2
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v3
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u16 v4, v3, v2, v4
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v6, 16, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v7, 16, v3
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v5, v0, v4
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v7, v6, v0
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v3, v2, v0
|
|
; GFX10-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i8 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%mul1 = mul nuw nsw i8 %v2e0, %v1e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%mul2 = mul nuw nsw i8 %v2e1, %v1e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%mul3 = mul nuw nsw i8 %v2e2, %v1e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%mul4 = mul nuw nsw i8 %v2e3, %v1e3
|
|
|
|
%acc = load i8, i8 addrspace(1)* %dst, align 2
|
|
%mad1 = add i8 %acc, %mul1
|
|
%mad2 = add i8 %mul2, %mad1
|
|
%mad3 = add i8 %mul3, %mad2
|
|
%mad4 = add i8 %mul4, %mad3
|
|
|
|
store i8 %mad4, i8 addrspace(1)* %dst, align 2
|
|
ret void
|
|
}
|
|
|
|
; TODO: Support commutation accross the adds.
|
|
define amdgpu_kernel void @udot4_CommutationAccrossMADs(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_CommutationAccrossMADs:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_ubyte v8, off, s[0:3], 0
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v3, v6, v3, v8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v5, v1, v3
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v7, v4, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v0, v2, v1
|
|
; GFX7-NEXT: buffer_store_byte v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_CommutationAccrossMADs:
|
|
; 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_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 v4, v[0:1]
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s1
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v2, s0
|
|
; GFX8-NEXT: flat_load_ubyte v10, v[2:3]
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 8, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v1, 16, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 24, v4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 8, v0
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v5, 16, v0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v6, v7, v6, v10
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v9, 24, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v0, v4, v6
|
|
; GFX8-NEXT: v_mad_u16 v0, v5, v1, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v9, v8, v0
|
|
; GFX8-NEXT: flat_store_byte v[2:3], v0
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: udot4_CommutationAccrossMADs:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: global_load_ubyte v9, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v5, 8, v2
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v6, 8, v3
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v5, v6, v5, v9
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v0, 16, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v7, 24, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v4, 16, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v2, v3, v2, v5
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v8, 24, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v4, v0, v2
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v8, v7, v0
|
|
; GFX9-NODL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_CommutationAccrossMADs:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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: global_load_ubyte v9, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 8, v3
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v5, v6, v5, v9
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v0, 16, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 24, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v4, 16, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v2, v3, v2, v5
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v8, 24, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v4, v0, v2
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v8, v7, v0
|
|
; GFX9-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_CommutationAccrossMADs:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX10-DL-NEXT: global_load_ubyte v4, v1, s[2:3]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v0, 8, v2
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v3
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v5, v0, v4
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v4, 16, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 16, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v3, v2, v0
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v5, v4, v0
|
|
; GFX10-DL-NEXT: v_mad_u16 v0, v3, v2, v0
|
|
; GFX10-DL-NEXT: global_store_byte v1, v0, s[2:3]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i8 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%mul1 = mul nuw nsw i8 %v2e0, %v1e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%mul2 = mul nuw nsw i8 %v2e1, %v1e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%mul3 = mul nuw nsw i8 %v2e2, %v1e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%mul4 = mul nuw nsw i8 %v2e3, %v1e3
|
|
|
|
%acc = load i8, i8 addrspace(1)* %dst, align 2
|
|
%mad1 = add i8 %acc, %mul2
|
|
%mad2 = add i8 %mad1, %mul1
|
|
%mad3 = add i8 %mad2, %mul3
|
|
%mad4 = add i8 %mad3, %mul4
|
|
|
|
store i8 %mad4, i8 addrspace(1)* %dst, align 2
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @udot4_multiuse_mul1(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_multiuse_mul1:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: s_load_dword s4, s[0:1], 0x0
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v8, v1, v5, s4
|
|
; GFX7-NEXT: v_mad_u32_u24 v3, v3, v6, v8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v5, v3
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v7, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_multiuse_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_movk_i32 s2, 0xff
|
|
; 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 s3, s[0:1], 0x0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_and_b32_e32 v1, s2, v3
|
|
; GFX8-NEXT: v_bfe_u32 v4, v3, 8, 8
|
|
; GFX8-NEXT: v_bfe_u32 v6, v3, 16, 8
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_and_b32_e32 v2, s2, v0
|
|
; GFX8-NEXT: v_bfe_u32 v5, v0, 8, 8
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mad_u32_u24 v8, v1, v2, s3
|
|
; GFX8-NEXT: v_mad_u32_u24 v4, v4, v5, v8
|
|
; GFX8-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX8-NEXT: v_mad_u32_u24 v1, v1, v2, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX8-NEXT: v_mad_u32_u24 v1, v6, v7, v1
|
|
; GFX8-NEXT: v_mad_u32_u24 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-NODL-LABEL: udot4_multiuse_mul1:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: s_movk_i32 s0, 0xff
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: s_load_dword s1, s[2:3], 0x0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v3, s0, v1
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v4, s0, v2
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v5, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v6, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_e32 v2, v3, v4
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_u32_u24 v3, v3, v4, s1
|
|
; GFX9-NODL-NEXT: v_add3_u32 v2, v5, v3, v2
|
|
; GFX9-NODL-NEXT: v_add3_u32 v1, v2, v6, v1
|
|
; GFX9-NODL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_multiuse_mul1:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_movk_i32 s0, 0xff
|
|
; 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 s1, s[2:3], 0x0
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v3, s0, v1
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v4, s0, v2
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v5, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v6, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_e32 v2, v3, v4
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_u32_u24 v3, v3, v4, s1
|
|
; GFX9-DL-NEXT: v_add3_u32 v2, v5, v3, v2
|
|
; GFX9-DL-NEXT: v_add3_u32 v1, v2, v6, v1
|
|
; GFX9-DL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_multiuse_mul1:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_movk_i32 s3, 0xff
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v0, s3, v1
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v3, s3, v2
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v4, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_e32 v5, v0, v3
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u32_u24 v0, v0, v3, s2
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NEXT: v_add3_u32 v0, v4, v0, v5
|
|
; GFX10-DL-NEXT: v_add3_u32 v0, v0, v3, v1
|
|
; GFX10-DL-NEXT: global_store_dword v2, v0, s[0:1]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i32 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%cv1e0 = zext i8 %v1e0 to i32
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%cv2e0 = zext i8 %v2e0 to i32
|
|
%mul1 = mul nuw nsw i32 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%cv1e1 = zext i8 %v1e1 to i32
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%cv2e1 = zext i8 %v2e1 to i32
|
|
%mul2 = mul nuw nsw i32 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%cv1e2 = zext i8 %v1e2 to i32
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%cv2e2 = zext i8 %v2e2 to i32
|
|
%mul3 = mul nuw nsw i32 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%cv1e3 = zext i8 %v1e3 to i32
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%cv2e3 = zext i8 %v2e3 to i32
|
|
%mul4 = mul nuw nsw i32 %cv1e3, %cv2e3
|
|
|
|
%acc = load i32, i32 addrspace(1)* %dst, align 4
|
|
%add = add i32 %mul1, %acc
|
|
%add1 = add i32 %mul2, %add
|
|
%add2 = add i32 %add1, %mul1
|
|
%add3 = add i32 %add2, %mul3
|
|
%add4 = add i32 %add3, %mul4
|
|
|
|
store i32 %add4, i32 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @udot4_multiuse_add1(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_multiuse_add1:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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 s5, s[0:1], 0x0
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v3, v3, v6, s5
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v5, v3
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v7, v1
|
|
; GFX7-NEXT: v_add_i32_e32 v6, vcc, s5, v3
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: v_add_i32_e32 v0, vcc, v0, v6
|
|
; GFX7-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_multiuse_add1:
|
|
; 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_movk_i32 s2, 0xff
|
|
; 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 s3, s[0:1], 0x0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_bfe_u32 v4, v3, 8, 8
|
|
; GFX8-NEXT: v_and_b32_e32 v1, s2, v3
|
|
; GFX8-NEXT: v_bfe_u32 v6, v3, 16, 8
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_bfe_u32 v5, v0, 8, 8
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mad_u32_u24 v4, v4, v5, s3
|
|
; GFX8-NEXT: v_and_b32_e32 v2, s2, v0
|
|
; GFX8-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX8-NEXT: v_mad_u32_u24 v1, v1, v2, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX8-NEXT: v_mad_u32_u24 v1, v6, v7, v1
|
|
; GFX8-NEXT: v_add_u32_e32 v5, vcc, s3, v4
|
|
; GFX8-NEXT: v_mad_u32_u24 v0, v3, v0, v1
|
|
; GFX8-NEXT: v_add_u32_e32 v2, vcc, v0, v5
|
|
; 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-NODL-LABEL: udot4_multiuse_add1:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_bfe_u32 v4, v1, 8, 8
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_bfe_u32 v5, v2, 8, 8
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v6, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_u32_u24 v2, v4, v5, s0
|
|
; GFX9-NODL-NEXT: v_add_u32_e32 v4, s0, v2
|
|
; GFX9-NODL-NEXT: v_add3_u32 v2, v2, v3, v6
|
|
; GFX9-NODL-NEXT: v_add3_u32 v1, v2, v1, v4
|
|
; GFX9-NODL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_multiuse_add1:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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_u32 v4, v1, 8, 8
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_bfe_u32 v5, v2, 8, 8
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v6, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_u32_u24 v2, v4, v5, s0
|
|
; GFX9-DL-NEXT: v_add_u32_e32 v4, s0, v2
|
|
; GFX9-DL-NEXT: v_add3_u32 v2, v2, v3, v6
|
|
; GFX9-DL-NEXT: v_add3_u32 v1, v2, v1, v4
|
|
; GFX9-DL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_multiuse_add1:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_bfe_u32 v0, v1, 8, 8
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v4, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u32_u24 v0, v0, v3, s2
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX10-DL-NEXT: v_add_nc_u32_e32 v2, s2, v0
|
|
; GFX10-DL-NEXT: v_add3_u32 v0, v0, v4, v3
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v3, 0
|
|
; GFX10-DL-NEXT: v_add3_u32 v0, v0, v1, v2
|
|
; GFX10-DL-NEXT: global_store_dword v3, v0, s[0:1]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i32 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%cv1e0 = zext i8 %v1e0 to i32
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%cv2e0 = zext i8 %v2e0 to i32
|
|
%mul1 = mul nuw nsw i32 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%cv1e1 = zext i8 %v1e1 to i32
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%cv2e1 = zext i8 %v2e1 to i32
|
|
%mul2 = mul nuw nsw i32 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%cv1e2 = zext i8 %v1e2 to i32
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%cv2e2 = zext i8 %v2e2 to i32
|
|
%mul3 = mul nuw nsw i32 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%cv1e3 = zext i8 %v1e3 to i32
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%cv2e3 = zext i8 %v2e3 to i32
|
|
%mul4 = mul nuw nsw i32 %cv1e3, %cv2e3
|
|
|
|
%acc = load i32, i32 addrspace(1)* %dst, align 4
|
|
%add1 = add i32 %mul2, %acc
|
|
%add = add i32 %add1, %acc
|
|
%add2 = add i32 %add1, %mul1
|
|
%add3 = add i32 %add2, %mul3
|
|
%add4 = add i32 %add3, %mul4
|
|
%res = add i32 %add4, %add
|
|
store i32 %res, i32 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @notdot4_mixedtypes(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: notdot4_mixedtypes:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
|
|
; GFX7-NEXT: buffer_load_ushort v1, off, s[0:3], 0
|
|
; GFX7-NEXT: s_mov_b32 s4, 0xffff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_bfe_i32 v3, v2, 0, 8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_bfe_i32 v6, v0, 0, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 8, 8
|
|
; GFX7-NEXT: v_and_b32_e32 v3, s4, v3
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s4, v6
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v7, v1
|
|
; GFX7-NEXT: v_bfe_u32 v5, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v8, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v3, v6, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v5, v8, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_short v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: notdot4_mixedtypes:
|
|
; 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_movk_i32 s2, 0xff
|
|
; GFX8-NEXT: v_mov_b32_e32 v5, s2
|
|
; 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 v4, v[0:1]
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s1
|
|
; GFX8-NEXT: v_mov_b32_e32 v2, s0
|
|
; GFX8-NEXT: flat_load_ushort v10, v[2:3]
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 8, v4
|
|
; GFX8-NEXT: v_and_b32_e32 v7, s2, v7
|
|
; GFX8-NEXT: v_bfe_i32 v1, v4, 0, 8
|
|
; GFX8-NEXT: v_and_b32_sdwa v9, v4, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 8, v0
|
|
; GFX8-NEXT: v_and_b32_e32 v8, s2, v8
|
|
; GFX8-NEXT: v_bfe_i32 v6, v0, 0, 8
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v7, v7, v8, v10
|
|
; GFX8-NEXT: v_and_b32_sdwa v5, v0, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX8-NEXT: v_mad_u16 v1, v1, v6, v7
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v4, 24, v4
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX8-NEXT: v_mad_u16 v1, v9, v5, v1
|
|
; GFX8-NEXT: v_mad_u16 v0, v4, v0, v1
|
|
; GFX8-NEXT: flat_store_short v[2:3], v0
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: notdot4_mixedtypes:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v1, 0
|
|
; GFX9-NODL-NEXT: s_movk_i32 s0, 0xff
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v3, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: global_load_ushort v9, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v5, 8, v2
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v6, 8, v3
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v5, s0, v5
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v6, s0, v6
|
|
; GFX9-NODL-NEXT: v_bfe_i32 v0, v2, 0, 8
|
|
; GFX9-NODL-NEXT: v_bfe_i32 v4, v3, 0, 8
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v5, v5, v6, v9
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v7, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v8, v3, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v0, v4, v5
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v7, v8, v0
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v0, v2, v3, v0
|
|
; GFX9-NODL-NEXT: global_store_short v1, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: notdot4_mixedtypes:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_movk_i32 s0, 0xff
|
|
; 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: global_load_ushort v9, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 8, v3
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v5, s0, v5
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v6, s0, v6
|
|
; GFX9-DL-NEXT: v_bfe_i32 v0, v2, 0, 8
|
|
; GFX9-DL-NEXT: v_bfe_i32 v4, v3, 0, 8
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v5, v5, v6, v9
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v7, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v8, v3, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v0, v4, v5
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v3, 24, v3
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v7, v8, v0
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v0, v2, v3, v0
|
|
; GFX9-DL-NEXT: global_store_short v1, v0, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: notdot4_mixedtypes:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_movk_i32 s0, 0xff
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: global_load_ushort v3, v0, s[2:3]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v4, 8, v1
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v2
|
|
; GFX10-DL-NEXT: v_bfe_i32 v6, v1, 0, 8
|
|
; GFX10-DL-NEXT: v_bfe_i32 v7, v2, 0, 8
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v4, s0, v4
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v5, s0, v5
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u16 v3, v4, v5, v3
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v4, v1, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v5, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v1, 24, v1
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX10-DL-NEXT: v_mad_u16 v3, v6, v7, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v3, v4, v5, v3
|
|
; GFX10-DL-NEXT: v_mad_u16 v1, v1, v2, v3
|
|
; GFX10-DL-NEXT: global_store_short v0, v1, s[2:3]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i16 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%v1e0 = extractelement <4 x i8> %vec1, i64 0
|
|
%cv1e0 = sext i8 %v1e0 to i16
|
|
%v2e0 = extractelement <4 x i8> %vec2, i64 0
|
|
%cv2e0 = sext i8 %v2e0 to i16
|
|
%mul1 = mul nuw nsw i16 %cv1e0, %cv2e0
|
|
|
|
%v1e1 = extractelement <4 x i8> %vec1, i64 1
|
|
%cv1e1 = zext i8 %v1e1 to i16
|
|
%v2e1 = extractelement <4 x i8> %vec2, i64 1
|
|
%cv2e1 = zext i8 %v2e1 to i16
|
|
%mul2 = mul nuw nsw i16 %cv1e1, %cv2e1
|
|
|
|
%v1e2 = extractelement <4 x i8> %vec1, i64 2
|
|
%cv1e2 = zext i8 %v1e2 to i16
|
|
%v2e2 = extractelement <4 x i8> %vec2, i64 2
|
|
%cv2e2 = zext i8 %v2e2 to i16
|
|
%mul3 = mul nuw nsw i16 %cv1e2, %cv2e2
|
|
|
|
%v1e3 = extractelement <4 x i8> %vec1, i64 3
|
|
%cv1e3 = zext i8 %v1e3 to i16
|
|
%v2e3 = extractelement <4 x i8> %vec2, i64 3
|
|
%cv2e3 = zext i8 %v2e3 to i16
|
|
%mul4 = mul nuw nsw i16 %cv1e3, %cv2e3
|
|
|
|
%acc = load i16, i16 addrspace(1)* %dst, align 2
|
|
%add1 = add i16 %mul2, %acc
|
|
%add2 = add i16 %add1, %mul1
|
|
%add3 = add i16 %add2, %mul3
|
|
%add4 = add i16 %add3, %mul4
|
|
|
|
store i16 %add4, i16 addrspace(1)* %dst, align 2
|
|
ret void
|
|
}
|
|
|
|
; TODO: cleanup s_lshr_b32 and support this pattern.
|
|
define amdgpu_kernel void @udot4_acc32_vecMul(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_acc32_vecMul:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_mov_b32 s2, -1
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v1, 24, v2
|
|
; GFX7-NEXT: v_bfe_u32 v3, v2, 8, 8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_and_b32_e32 v2, s4, v2
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v5, 24, v0
|
|
; GFX7-NEXT: v_bfe_u32 v6, v0, 8, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_and_b32_e32 v0, s4, v0
|
|
; GFX7-NEXT: s_load_dword s4, s[0:1], 0x0
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, s4
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v3, v6, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v4, v7, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v1, v5, v0
|
|
; GFX7-NEXT: buffer_store_dword v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_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_movk_i32 s2, 0xff
|
|
; 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 s3, s[0:1], 0x0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v1, 24, v3
|
|
; GFX8-NEXT: v_bfe_u32 v4, v3, 16, 8
|
|
; GFX8-NEXT: v_lshrrev_b16_e32 v5, 8, v3
|
|
; GFX8-NEXT: v_and_b32_e32 v3, s2, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v2, 24, v0
|
|
; GFX8-NEXT: v_bfe_u32 v6, v0, 16, 8
|
|
; GFX8-NEXT: v_lshrrev_b16_e32 v7, 8, v0
|
|
; GFX8-NEXT: v_and_b32_e32 v0, s2, v0
|
|
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX8-NEXT: v_mad_u32_u24 v0, v3, v0, s3
|
|
; GFX8-NEXT: v_mad_u32_u24 v0, v5, v7, v0
|
|
; GFX8-NEXT: v_mad_u32_u24 v0, v4, v6, v0
|
|
; GFX8-NEXT: v_mad_u32_u24 v2, v1, v2, 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-NODL-LABEL: udot4_acc32_vecMul:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: s_load_dword s0, s[2:3], 0x0
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v4, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v5, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX9-NODL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: v_add3_u32 v2, v3, s0, v4
|
|
; GFX9-NODL-NEXT: v_add3_u32 v1, v2, v5, v1
|
|
; GFX9-NODL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_acc32_vecMul:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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(0)
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v4, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v5, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX9-DL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-DL-NEXT: v_add3_u32 v2, v3, s0, v4
|
|
; GFX9-DL-NEXT: v_add3_u32 v1, v2, v5, v1
|
|
; GFX9-DL-NEXT: global_store_dword v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_acc32_vecMul:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_mov_b32 s3, 0xffff
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: s_load_dword s2, s[0:1], 0x0
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v0, s3, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_1
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v3, s3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_1
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v4, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_e32 v0, v0, v3
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v3, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_2
|
|
; GFX10-DL-NEXT: v_mul_u32_u24_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v2, 0
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: v_add3_u32 v0, v4, s2, v0
|
|
; GFX10-DL-NEXT: v_add3_u32 v0, v0, v3, v1
|
|
; GFX10-DL-NEXT: global_store_dword v2, v0, s[0:1]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i32 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%cvec1 = zext <4 x i8> %vec1 to <4 x i32>
|
|
%cvec2 = zext <4 x i8> %vec2 to <4 x i32>
|
|
|
|
%mul = mul <4 x i32> %cvec1, %cvec2
|
|
%mul0 = extractelement <4 x i32> %mul, i64 0
|
|
%mul1 = extractelement <4 x i32> %mul, i64 1
|
|
%mul2 = extractelement <4 x i32> %mul, i64 2
|
|
%mul3 = extractelement <4 x i32> %mul, i64 3
|
|
|
|
%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
|
|
|
|
store i32 %add4, i32 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; TODO: This pattern should be recognized.
|
|
define amdgpu_kernel void @udot4_acc16_vecMul(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_acc16_vecMul:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_ushort v8, off, s[0:3], 0
|
|
; GFX7-NEXT: s_mov_b32 s4, 0xff00
|
|
; GFX7-NEXT: s_movk_i32 s5, 0xff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s4, v2
|
|
; GFX7-NEXT: v_and_b32_e32 v3, s5, v2
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v1, 8, v1
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v5, s4, v0
|
|
; GFX7-NEXT: v_or_b32_e32 v1, v3, v1
|
|
; GFX7-NEXT: v_lshlrev_b32_e32 v3, 8, v5
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s5, v0
|
|
; GFX7-NEXT: v_or_b32_e32 v3, v6, v3
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v5, 16, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v6, 16, v3
|
|
; GFX7-NEXT: v_and_b32_e32 v1, s5, v1
|
|
; GFX7-NEXT: v_and_b32_e32 v3, s5, v3
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v1, v3, v8
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v5, v6, v1
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v0, 24, v0
|
|
; GFX7-NEXT: v_mad_u32_u24 v1, v4, v7, v1
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v1
|
|
; GFX7-NEXT: buffer_store_short v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_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: s_movk_i32 s2, 0xff
|
|
; GFX8-NEXT: v_mov_b32_e32 v5, s2
|
|
; 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 v4, v[0:1]
|
|
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s6, v2
|
|
; GFX8-NEXT: v_mov_b32_e32 v1, s7
|
|
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
|
|
; GFX8-NEXT: v_mov_b32_e32 v3, s1
|
|
; GFX8-NEXT: flat_load_dword v0, v[0:1]
|
|
; GFX8-NEXT: v_mov_b32_e32 v2, s0
|
|
; GFX8-NEXT: flat_load_ushort v10, v[2:3]
|
|
; GFX8-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_and_b32_sdwa v9, v4, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v1, 24, v4
|
|
; GFX8-NEXT: v_lshrrev_b16_e32 v6, 8, v4
|
|
; GFX8-NEXT: v_and_b32_e32 v4, s2, v4
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 24, v0
|
|
; GFX8-NEXT: v_lshrrev_b16_e32 v8, 8, v0
|
|
; GFX8-NEXT: v_and_b32_sdwa v5, v0, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX8-NEXT: v_and_b32_e32 v0, s2, v0
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v0, v4, v0, v10
|
|
; GFX8-NEXT: v_mad_u16 v0, v6, v8, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v9, v5, v0
|
|
; GFX8-NEXT: v_mad_u16 v0, v1, v7, v0
|
|
; GFX8-NEXT: flat_store_short v[2:3], v0
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: udot4_acc16_vecMul:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: s_movk_i32 s0, 0xff
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v4, 0xffff
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NODL-NEXT: global_load_ushort v3, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b16_e32 v5, 8, v1
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v10, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v6, 24, v1
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v9, v1, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v1, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
|
|
; GFX9-NODL-NEXT: v_lshrrev_b16_e32 v7, 8, v2
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v8, 24, v2
|
|
; GFX9-NODL-NEXT: v_and_b32_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
|
|
; GFX9-NODL-NEXT: v_lshl_or_b32 v2, v7, 16, v2
|
|
; GFX9-NODL-NEXT: v_lshl_or_b32 v1, v5, 16, v1
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v10, v4, v10
|
|
; GFX9-NODL-NEXT: v_and_b32_e32 v4, v4, v9
|
|
; GFX9-NODL-NEXT: v_pk_mul_lo_u16 v1, v1, v2
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_add_u16_e32 v3, v1, v3
|
|
; GFX9-NODL-NEXT: v_lshl_or_b32 v5, v8, 16, v10
|
|
; GFX9-NODL-NEXT: v_lshl_or_b32 v4, v6, 16, v4
|
|
; GFX9-NODL-NEXT: v_pk_mul_lo_u16 v2, v4, v5
|
|
; GFX9-NODL-NEXT: v_add_u16_sdwa v1, v3, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NODL-NEXT: v_add_u16_e32 v1, v1, v2
|
|
; GFX9-NODL-NEXT: v_add_u16_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-NODL-NEXT: global_store_short v0, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_acc16_vecMul:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_movk_i32 s0, 0xff
|
|
; GFX9-DL-NEXT: v_mov_b32_e32 v4, 0xffff
|
|
; 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_b16_e32 v5, 8, v1
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v10, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 24, v1
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v9, v1, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v1, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
|
|
; GFX9-DL-NEXT: v_lshrrev_b16_e32 v7, 8, v2
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v8, 24, v2
|
|
; GFX9-DL-NEXT: v_and_b32_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v2, v7, 16, v2
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v1, v5, 16, v1
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v10, v4, v10
|
|
; GFX9-DL-NEXT: v_and_b32_e32 v4, v4, v9
|
|
; GFX9-DL-NEXT: v_pk_mul_lo_u16 v1, v1, v2
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v3, v1, v3
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v5, v8, 16, v10
|
|
; GFX9-DL-NEXT: v_lshl_or_b32 v4, v6, 16, v4
|
|
; GFX9-DL-NEXT: v_pk_mul_lo_u16 v2, v4, v5
|
|
; GFX9-DL-NEXT: v_add_u16_sdwa v1, v3, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v1, v1, v2
|
|
; GFX9-DL-NEXT: v_add_u16_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
|
|
; GFX9-DL-NEXT: global_store_short v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_acc16_vecMul:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v4, 0xffff
|
|
; GFX10-DL-NEXT: s_movk_i32 s2, 0xff
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: global_load_ushort v3, v0, s[0:1]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b16 v5, 8, v1
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v8, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b16 v6, 8, v2
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v7, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v9, v1, s2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_and_b32_sdwa v10, v2, s2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_lshl_or_b32 v5, v5, 16, v8
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v1, 24, v1
|
|
; GFX10-DL-NEXT: v_lshl_or_b32 v6, v6, 16, v7
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 24, v2
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v7, v4, v10
|
|
; GFX10-DL-NEXT: v_and_b32_e32 v4, v4, v9
|
|
; GFX10-DL-NEXT: v_pk_mul_lo_u16 v5, v5, v6
|
|
; GFX10-DL-NEXT: v_lshl_or_b32 v2, v2, 16, v7
|
|
; GFX10-DL-NEXT: v_lshl_or_b32 v1, v1, 16, v4
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v4, 16, v5
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_add_nc_u16 v3, v5, v3
|
|
; GFX10-DL-NEXT: v_pk_mul_lo_u16 v1, v1, v2
|
|
; GFX10-DL-NEXT: v_add_nc_u16 v2, v3, v4
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v3, 16, v1
|
|
; GFX10-DL-NEXT: v_add_nc_u16 v1, v2, v1
|
|
; GFX10-DL-NEXT: v_add_nc_u16 v1, v1, v3
|
|
; GFX10-DL-NEXT: global_store_short v0, v1, s[0:1]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i16 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%cvec1 = zext <4 x i8> %vec1 to <4 x i16>
|
|
%cvec2 = zext <4 x i8> %vec2 to <4 x i16>
|
|
|
|
%mul = mul <4 x i16> %cvec1, %cvec2
|
|
%mul0 = extractelement <4 x i16> %mul, i64 0
|
|
%mul1 = extractelement <4 x i16> %mul, i64 1
|
|
%mul2 = extractelement <4 x i16> %mul, i64 2
|
|
%mul3 = extractelement <4 x i16> %mul, i64 3
|
|
|
|
%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
|
|
|
|
store i16 %add4, i16 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
; TODO: Support this pattern.
|
|
define amdgpu_kernel void @udot4_acc8_vecMul(<4 x i8> addrspace(1)* %src1,
|
|
; GFX7-LABEL: udot4_acc8_vecMul:
|
|
; GFX7: ; %bb.0: ; %entry
|
|
; 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 s3, 0xf000
|
|
; GFX7-NEXT: s_mov_b32 s10, 0
|
|
; GFX7-NEXT: s_mov_b32 s11, s3
|
|
; GFX7-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX7-NEXT: s_mov_b64 s[8:9], s[4:5]
|
|
; 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_mov_b32 s2, -1
|
|
; GFX7-NEXT: buffer_load_ubyte v8, off, s[0:3], 0
|
|
; GFX7-NEXT: s_movk_i32 s4, 0xff
|
|
; GFX7-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX7-NEXT: v_and_b32_e32 v3, s4, v2
|
|
; GFX7-NEXT: v_bfe_u32 v4, v2, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX7-NEXT: v_and_b32_e32 v6, s4, v0
|
|
; GFX7-NEXT: v_bfe_u32 v7, v0, 8, 8
|
|
; GFX7-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX7-NEXT: v_mad_u32_u24 v3, v3, v6, v8
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v1, 24, v2
|
|
; GFX7-NEXT: v_lshrrev_b32_e32 v5, 24, v0
|
|
; GFX7-NEXT: v_bfe_u32 v2, v2, 16, 8
|
|
; GFX7-NEXT: v_bfe_u32 v0, v0, 16, 8
|
|
; GFX7-NEXT: v_mad_u32_u24 v3, v4, v7, v3
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v2, v0, v3
|
|
; GFX7-NEXT: v_mad_u32_u24 v0, v1, v5, v0
|
|
; GFX7-NEXT: buffer_store_byte v0, off, s[0:3], 0
|
|
; GFX7-NEXT: s_endpgm
|
|
;
|
|
; GFX8-LABEL: udot4_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: 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_waitcnt vmcnt(2)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v5, 16, v3
|
|
; GFX8-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v6, 16, v2
|
|
; GFX8-NEXT: v_mul_lo_u16_sdwa v7, v3, v2 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX8-NEXT: v_mul_lo_u16_e32 v9, v5, v6
|
|
; GFX8-NEXT: v_or_b32_sdwa v7, v9, v7 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX8-NEXT: v_mul_lo_u16_sdwa v8, v3, v2 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX8-NEXT: v_lshlrev_b32_e32 v9, 16, v7
|
|
; GFX8-NEXT: v_or_b32_e32 v8, v8, v9
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v8, 8, v8
|
|
; GFX8-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX8-NEXT: v_mad_u16 v2, v3, v2, v4
|
|
; GFX8-NEXT: v_add_u16_e32 v2, v2, v8
|
|
; GFX8-NEXT: v_lshrrev_b32_e32 v7, 8, v7
|
|
; GFX8-NEXT: v_mad_u16 v2, v5, v6, v2
|
|
; GFX8-NEXT: v_add_u16_e32 v2, v2, v7
|
|
; GFX8-NEXT: flat_store_byte v[0:1], v2
|
|
; GFX8-NEXT: s_endpgm
|
|
;
|
|
; GFX9-NODL-LABEL: udot4_acc8_vecMul:
|
|
; GFX9-NODL: ; %bb.0: ; %entry
|
|
; GFX9-NODL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX9-NODL-NEXT: s_load_dwordx2 s[2:3], s[0:1], 0x34
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX9-NODL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX9-NODL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX9-NODL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX9-NODL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX9-NODL-NEXT: global_load_ubyte v3, v0, s[2:3]
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v4, 16, v1
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v5, 16, v2
|
|
; GFX9-NODL-NEXT: v_mul_lo_u16_sdwa v6, v1, v2 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-NODL-NEXT: v_mul_lo_u16_e32 v8, v4, v5
|
|
; GFX9-NODL-NEXT: v_or_b32_sdwa v6, v8, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-NODL-NEXT: v_mul_lo_u16_sdwa v7, v1, v2 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX9-NODL-NEXT: v_lshlrev_b32_e32 v8, 16, v6
|
|
; GFX9-NODL-NEXT: v_or_b32_e32 v7, v7, v8
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v7, 8, v7
|
|
; GFX9-NODL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v1, v1, v2, v3
|
|
; GFX9-NODL-NEXT: v_add_u16_e32 v1, v1, v7
|
|
; GFX9-NODL-NEXT: v_lshrrev_b32_e32 v6, 8, v6
|
|
; GFX9-NODL-NEXT: v_mad_legacy_u16 v1, v4, v5, v1
|
|
; GFX9-NODL-NEXT: v_add_u16_e32 v1, v1, v6
|
|
; GFX9-NODL-NEXT: global_store_byte v0, v1, s[2:3]
|
|
; GFX9-NODL-NEXT: s_endpgm
|
|
;
|
|
; GFX9-DL-LABEL: udot4_acc8_vecMul:
|
|
; GFX9-DL: ; %bb.0: ; %entry
|
|
; 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_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 v4, 16, v1
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v5, 16, v2
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_sdwa v6, v1, v2 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:BYTE_3
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_e32 v8, v4, v5
|
|
; GFX9-DL-NEXT: v_or_b32_sdwa v6, v8, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX9-DL-NEXT: v_mul_lo_u16_sdwa v7, v1, v2 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
|
|
; GFX9-DL-NEXT: v_lshlrev_b32_e32 v8, 16, v6
|
|
; GFX9-DL-NEXT: v_or_b32_e32 v7, v7, v8
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v7, 8, v7
|
|
; GFX9-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v1, v1, v2, v3
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v1, v1, v7
|
|
; GFX9-DL-NEXT: v_lshrrev_b32_e32 v6, 8, v6
|
|
; GFX9-DL-NEXT: v_mad_legacy_u16 v1, v4, v5, v1
|
|
; GFX9-DL-NEXT: v_add_u16_e32 v1, v1, v6
|
|
; GFX9-DL-NEXT: global_store_byte v0, v1, s[2:3]
|
|
; GFX9-DL-NEXT: s_endpgm
|
|
;
|
|
; GFX10-DL-LABEL: udot4_acc8_vecMul:
|
|
; GFX10-DL: ; %bb.0: ; %entry
|
|
; GFX10-DL-NEXT: s_load_dwordx4 s[4:7], s[0:1], 0x24
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v0, 2, v0
|
|
; GFX10-DL-NEXT: s_load_dwordx2 s[0:1], s[0:1], 0x34
|
|
; GFX10-DL-NEXT: s_waitcnt lgkmcnt(0)
|
|
; GFX10-DL-NEXT: s_clause 0x1
|
|
; GFX10-DL-NEXT: global_load_dword v1, v0, s[4:5]
|
|
; GFX10-DL-NEXT: global_load_dword v2, v0, s[6:7]
|
|
; GFX10-DL-NEXT: v_mov_b32_e32 v0, 0
|
|
; GFX10-DL-NEXT: global_load_ubyte v3, v0, s[0:1]
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(2)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v4, 24, v1
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(1)
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 24, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v6, 16, v1
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v7, 16, v2
|
|
; GFX10-DL-NEXT: v_lshrrev_b16 v8, 8, v2
|
|
; GFX10-DL-NEXT: v_mul_lo_u16 v4, v4, v5
|
|
; GFX10-DL-NEXT: v_lshrrev_b16 v5, 8, v1
|
|
; GFX10-DL-NEXT: v_mul_lo_u16 v9, v6, v7
|
|
; GFX10-DL-NEXT: s_waitcnt vmcnt(0)
|
|
; GFX10-DL-NEXT: v_mad_u16 v1, v1, v2, v3
|
|
; GFX10-DL-NEXT: v_lshlrev_b16 v4, 8, v4
|
|
; GFX10-DL-NEXT: v_mul_lo_u16 v5, v5, v8
|
|
; GFX10-DL-NEXT: v_or_b32_sdwa v4, v9, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_lshlrev_b16 v5, 8, v5
|
|
; GFX10-DL-NEXT: v_lshlrev_b32_e32 v8, 16, v4
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v2, 8, v4
|
|
; GFX10-DL-NEXT: v_or_b32_sdwa v5, v5, v8 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
|
|
; GFX10-DL-NEXT: v_lshrrev_b32_e32 v5, 8, v5
|
|
; GFX10-DL-NEXT: v_add_nc_u16 v1, v1, v5
|
|
; GFX10-DL-NEXT: v_mad_u16 v1, v6, v7, v1
|
|
; GFX10-DL-NEXT: v_add_nc_u16 v1, v1, v2
|
|
; GFX10-DL-NEXT: global_store_byte v0, v1, s[0:1]
|
|
; GFX10-DL-NEXT: s_endpgm
|
|
<4 x i8> addrspace(1)* %src2,
|
|
i8 addrspace(1)* nocapture %dst) {
|
|
entry:
|
|
%idx = call i32 @llvm.amdgcn.workitem.id.x()
|
|
%gep1 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src1, i32 %idx
|
|
%vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %gep1
|
|
%gep2 = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %src2, i32 %idx
|
|
%vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %gep2
|
|
|
|
%mul = mul <4 x i8> %vec1, %vec2
|
|
%mul0 = extractelement <4 x i8> %mul, i64 0
|
|
%mul1 = extractelement <4 x i8> %mul, i64 1
|
|
%mul2 = extractelement <4 x i8> %mul, i64 2
|
|
%mul3 = extractelement <4 x i8> %mul, i64 3
|
|
|
|
%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
|
|
|
|
store i8 %add4, i8 addrspace(1)* %dst, align 4
|
|
ret void
|
|
}
|
|
|
|
declare i32 @llvm.amdgcn.workitem.id.x()
|