1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-22 18:54:02 +01:00
llvm-mirror/test/CodeGen/AMDGPU/setcc-limit-load-shrink.ll
Austin Kerbow d1f23a1772 [AMDGPU] Update subtarget features for new target ID support
Support for XNACK and SRAMECC is not static on some GPUs. We must be able
to differentiate between different scenarios for these dynamic subtarget
features.

The possible settings are:

- Unsupported: The GPU has no support for XNACK/SRAMECC.
- Any: Preference is unspecified. Use conservative settings that can run anywhere.
- Off: Request support for XNACK/SRAMECC Off
- On: Request support for XNACK/SRAMECC On

GCNSubtarget will track the four options based on the following criteria. If
the subtarget does not support XNACK/SRAMECC we say the setting is
"Unsupported". If no subtarget features for XNACK/SRAMECC are requested we
must support "Any" mode. If the subtarget features XNACK/SRAMECC exist in the
feature string when initializing the subtarget, the settings are "On/Off".

The defaults are updated to be conservatively correct, meaning if no setting
for XNACK or SRAMECC is explicitly requested, defaults will be used which
generate code that can be run anywhere. This corresponds to the "Any" setting.

Differential Revision: https://reviews.llvm.org/D85882
2021-01-26 11:25:51 -08:00

70 lines
2.8 KiB
LLVM

; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
; GCN-LABEL: {{^}}const_load_no_shrink_dword_to_unaligned_byte:
; GCN: s_load_dword s{{[0-9]+}}
; GCN: s_load_dword [[LD:s[0-9]+]],
; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10013
define amdgpu_kernel void @const_load_no_shrink_dword_to_unaligned_byte(i32 addrspace(1)* %out, i32 addrspace(4)* %in, i32 %x) {
%ptr = getelementptr i32, i32 addrspace(4)* %in, i32 %x
%load = load i32, i32 addrspace(4)* %ptr, align 4
%and = and i32 %load, 524288
%cmp = icmp eq i32 %and, 0
%sel = select i1 %cmp, i32 0, i32 -1
store i32 %sel, i32 addrspace(1)* %out
ret void
}
; GCN-LABEL: const_load_no_shrink_dword_to_aligned_byte:
; GCN: s_load_dword s{{[0-9]+}}
; GCN: s_load_dword [[LD:s[0-9]+]],
; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10003
define amdgpu_kernel void @const_load_no_shrink_dword_to_aligned_byte(i32 addrspace(1)* %out, i32 addrspace(4)* %in, i32 %x) {
%ptr = getelementptr i32, i32 addrspace(4)* %in, i32 %x
%load = load i32, i32 addrspace(4)* %ptr, align 4
%and = and i32 %load, 8
%cmp = icmp eq i32 %and, 0
%sel = select i1 %cmp, i32 0, i32 -1
store i32 %sel, i32 addrspace(1)* %out
ret void
}
; GCN-LABEL: global_load_no_shrink_dword_to_unaligned_byte:
; GCN: s_load_dword s{{[0-9]+}}
; GCN: s_load_dword [[LD:s[0-9]+]],
; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10013
define amdgpu_kernel void @global_load_no_shrink_dword_to_unaligned_byte(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in, i32 %x) {
%ptr = getelementptr i32, i32 addrspace(1)* %in, i32 %x
%load = load i32, i32 addrspace(1)* %ptr, align 4
%and = and i32 %load, 524288
%cmp = icmp eq i32 %and, 0
%sel = select i1 %cmp, i32 0, i32 -1
store i32 %sel, i32 addrspace(1)* %out
ret void
}
; GCN-LABEL: global_load_no_shrink_dword_to_aligned_byte:
; GCN: s_load_dword s{{[0-9]+}}
; GCN: s_load_dword [[LD:s[0-9]+]],
; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10003
define amdgpu_kernel void @global_load_no_shrink_dword_to_aligned_byte(i32 addrspace(1)* %out, i32 addrspace(1)* %in, i32 %x) {
%ptr = getelementptr i32, i32 addrspace(1)* %in, i32 %x
%load = load i32, i32 addrspace(1)* %ptr, align 4
%and = and i32 %load, 8
%cmp = icmp eq i32 %and, 0
%sel = select i1 %cmp, i32 0, i32 -1
store i32 %sel, i32 addrspace(1)* %out
ret void
}
; GCN-LABEL: const_load_shrink_dword_to_unaligned_byte:
; GCN: global_load_ushort
define amdgpu_kernel void @const_load_shrink_dword_to_unaligned_byte(i32 addrspace(1)* %out, i32 addrspace(4)* %in, i32 %x) {
%ptr = getelementptr i32, i32 addrspace(4)* %in, i32 %x
%load = load i32, i32 addrspace(4)* %ptr, align 2
%and = and i32 %load, 524288
%cmp = icmp eq i32 %and, 0
%sel = select i1 %cmp, i32 0, i32 -1
store i32 %sel, i32 addrspace(1)* %out
ret void
}