Page MenuHomePhabricator
Differential D57737 Diff 186244 llvm/trunk/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll

Changeset View

Standalone View

View Options

llvm/trunk/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll

; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX7LESS %s ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX7LESS %s
; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX8MORE %s ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX8MORE %s
; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX8MORE %s ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX8MORE %s
declare i32 @llvm.amdgcn.workitem.id.x() declare i32 @llvm.amdgcn.workitem.id.x()
declare i32 @llvm.amdgcn.raw.buffer.atomic.add(i32, <4 x i32>, i32, i32, i32) declare i32 @llvm.amdgcn.raw.buffer.atomic.add(i32, <4 x i32>, i32, i32, i32)
declare i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32, <4 x i32>, i32, i32, i32) declare i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32, <4 x i32>, i32, i32, i32)
; Show that what the atomic optimization pass will do for raw buffers. ; Show that what the atomic optimization pass will do for raw buffers.
; GCN-LABEL: add_i32_constant: ; GCN-LABEL: add_i32_constant:
; GCN: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec ; GCN: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0
; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]] ; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]]
; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]] ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]]
; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} ; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
; GCN: buffer_atomic_add v[[value]] ; GCN: buffer_atomic_add v[[value]]
define amdgpu_kernel void @add_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) { define amdgpu_kernel void @add_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
entry: entry:
%old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0) %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0)
store i32 %old, i32 addrspace(1)* %out store i32 %old, i32 addrspace(1)* %out
ret void ret void
} }
; GCN-LABEL: add_i32_uniform: ; GCN-LABEL: add_i32_uniform:
; GCN: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec ; GCN: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0
; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]] ; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]]
; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]] ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]]
; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} ; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]] ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]] ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
; GCN: buffer_atomic_add v[[value]] ; GCN: buffer_atomic_add v[[value]]
define amdgpu_kernel void @add_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %additive) { define amdgpu_kernel void @add_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %additive) {
Show All 28 Lines
entry: entry:
%lane = call i32 @llvm.amdgcn.workitem.id.x() %lane = call i32 @llvm.amdgcn.workitem.id.x()
%old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 1, <4 x i32> %inout, i32 %lane, i32 0, i32 0) %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 1, <4 x i32> %inout, i32 %lane, i32 0, i32 0)
store i32 %old, i32 addrspace(1)* %out store i32 %old, i32 addrspace(1)* %out
ret void ret void
} }
; GCN-LABEL: sub_i32_constant: ; GCN-LABEL: sub_i32_constant:
; GCN: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec ; GCN: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0
; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]] ; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]]
; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]] ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]]
; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} ; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
; GCN: buffer_atomic_sub v[[value]] ; GCN: buffer_atomic_sub v[[value]]
define amdgpu_kernel void @sub_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) { define amdgpu_kernel void @sub_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
entry: entry:
%old = call i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0) %old = call i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0)
store i32 %old, i32 addrspace(1)* %out store i32 %old, i32 addrspace(1)* %out
ret void ret void
} }
; GCN-LABEL: sub_i32_uniform: ; GCN-LABEL: sub_i32_uniform:
; GCN: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec ; GCN: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_lo:[0-9]+]], s[[exec_lo]], 0
; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]] ; GCN: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt_hi:[0-9]+]], s[[exec_hi]], v[[mbcnt_lo]]
; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]] ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc, 0, v[[mbcnt_hi]]
; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}} ; GCN: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]] ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]] ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
; GCN: buffer_atomic_sub v[[value]] ; GCN: buffer_atomic_sub v[[value]]
define amdgpu_kernel void @sub_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %subitive) { define amdgpu_kernel void @sub_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %subitive) {
Show All 34 Lines