[AMDGPU] Add gfx940 run line to gfx90a mfma test. NFC.

diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll
index 29c3d69..f79406a 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.mfma.gfx90a.ll
@@ -1,4 +1,5 @@
 ; RUN: llc -march=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,GFX90A %s
+; RUN: llc -march=amdgcn -mcpu=gfx940 -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,GFX940 %s
 
 declare <32 x float> @llvm.amdgcn.mfma.f32.32x32x4bf16.1k(<4 x i16>, <4 x i16>, <32 x float>, i32, i32, i32)
 declare <16 x float> @llvm.amdgcn.mfma.f32.16x16x4bf16.1k(<4 x i16>, <4 x i16>, <16 x float>, i32, i32, i32)
@@ -44,9 +45,13 @@ declare i32 @llvm.amdgcn.workitem.id.x()
 ; GFX90A-DAG:  v_accvgpr_write_b32 a{{[0-9]+}}, v{{[0-9]+}}
 ; GFX90A-DAG:  v_accvgpr_write_b32 a{{[0-9]+}}, v{{[0-9]+}}
 ; GFX90A-DAG:  v_accvgpr_write_b32 a{{[0-9]+}}, v{{[0-9]+}}
+; GFX940-DAG:  v_mov_b32_e32 v[[TWO:[0-9]+]], 2
+; GFX940-DAG:  v_mov_b32_e32 v[[ONE:[0-9]+]], 1
+; GFX940-COUNT-32: v_accvgpr_write_b32 a{{[0-9]+}}, s{{[0-9]+}}
 ; GFX90A-DAG:  v_mov_b32_e32 v[[TWO:[0-9]+]], 2
 ; GFX90A-DAG:  v_mov_b32_e32 v[[ONE:[0-9]+]], 1
 ; GFX90A:      v_mfma_f32_32x32x4bf16_1k a[{{[0-9]+:[0-9]+}}], v[[[ONE]]:{{[0-9]+}}], v[[[TWO]]:{{[0-9]+}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
+; GFX940:      v_mfma_f32_32x32x4_2b_bf16 a[{{[0-9]+:[0-9]+}}], v{{\[}}[[ONE]]:{{[0-9+]}}], v{{\[}}[[TWO]]:{{[0-9+]}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
 ; GCN-NOT:     v_accvgpr_read_b32
 ; GCN-COUNT-8: global_store_dwordx4 v{{[0-9]+}}, a[{{[0-9:]+}}]
 define amdgpu_kernel void @test_mfma_f32_32x32x4bf16_1k(<32 x float> addrspace(1)* %arg) #0 {
@@ -64,7 +69,9 @@ bb:
 ; GCN-DAG:         v_mov_b32_e32 v[[TWO:[0-9]+]], 2
 ; GCN-DAG:         v_mov_b32_e32 v[[ONE:[0-9]+]], 1
 ; GFX90A-COUNT-16: v_accvgpr_write_b32 a{{[0-9]+}}, v{{[0-9]+}}
+; GFX940-COUNT-16: v_accvgpr_write_b32 a{{[0-9]+}}, s{{[0-9]+}}
 ; GFX90A:          v_mfma_f32_16x16x4bf16_1k a[{{[0-9]+:[0-9]+}}], v[[[ONE]]:{{[0-9]+}}], v[[[TWO]]:{{[0-9]+}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
+; GFX940:          v_mfma_f32_16x16x4_4b_bf16 a[{{[0-9]+:[0-9]+}}], v{{\[}}[[ONE]]:{{[0-9+]}}], v{{\[}}[[TWO]]:{{[0-9+]}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
 ; GCN-NOT:         v_accvgpr_read_b32
 ; GCN-COUNT-4:     global_store_dwordx4 v{{[0-9]+}}, a[{{[0-9:]+}}]
 define amdgpu_kernel void @test_mfma_f32_16x16x4bf16_1k(<16 x float> addrspace(1)* %arg) #0 {
@@ -82,7 +89,9 @@ bb:
 ; GCN-DAG:        v_mov_b32_e32 v[[TWO:[0-9]+]], 2
 ; GCN-DAG:        v_mov_b32_e32 v[[ONE:[0-9]+]], 1
 ; GFX90A-COUNT-4: v_accvgpr_write_b32 a{{[0-9]+}}, v{{[0-9]+}}
+; GFX940-COUNT-4: v_accvgpr_write_b32 a{{[0-9]+}}, s{{[0-9]+}}
 ; GFX90A:         v_mfma_f32_4x4x4bf16_1k [[RES:a\[[0-9]+:[0-9]+\]]], v[[[ONE]]:{{[0-9]+}}], v[[[TWO]]:{{[0-9]+}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
+; GFX940:         v_mfma_f32_4x4x4_16b_bf16 [[RES:a\[[0-9]+:[0-9]+\]]], v{{\[}}[[ONE]]:{{[0-9+]}}], v{{\[}}[[TWO]]:{{[0-9+]}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
 ; GCN-NOT:        v_accvgpr_read_b32
 ; GCN:            global_store_dwordx4 v{{[0-9]+}}, [[RES]],
 define amdgpu_kernel void @test_mfma_f32_4x4x4bf16_1k(<4 x float> addrspace(1)* %arg) #0 {
@@ -100,7 +109,9 @@ bb:
 ; GCN-DAG:         v_mov_b32_e32 v[[TWO:[0-9]+]], 2
 ; GCN-DAG:         v_mov_b32_e32 v[[ONE:[0-9]+]], 1
 ; GFX90A-COUNT-16: v_accvgpr_write_b32 a{{[0-9]+}}, v{{[0-9]+}}
+; GFX940-COUNT-16: v_accvgpr_write_b32 a{{[0-9]+}}, s{{[0-9]+}}
 ; GFX90A:          v_mfma_f32_32x32x8bf16_1k a[{{[0-9]+:[0-9]+}}], v[[[ONE]]:{{[0-9]+}}], v[[[TWO]]:{{[0-9]+}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
+; GFX940:          v_mfma_f32_32x32x8_bf16 a[{{[0-9]+:[0-9]+}}], v{{\[}}[[ONE]]:{{[0-9+]}}], v{{\[}}[[TWO]]:{{[0-9+]}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
 ; GCN-NOT:         v_accvgpr_read_b32
 ; GCN-COUNT-4:     global_store_dwordx4 v{{[0-9]+}}, a[{{[0-9:]+}}]
 define amdgpu_kernel void @test_mfma_f32_32x32x8bf16_1k(<16 x float> addrspace(1)* %arg) #0 {
@@ -118,7 +129,9 @@ bb:
 ; GCN-DAG:        v_mov_b32_e32 v[[TWO:[0-9]+]], 2
 ; GCN-DAG:        v_mov_b32_e32 v[[ONE:[0-9]+]], 1
 ; GFX90A-COUNT-4: v_accvgpr_write_b32 a{{[0-9]+}}, v{{[0-9]+}}
+; GFX940-COUNT-4: v_accvgpr_write_b32 a{{[0-9]+}}, s{{[0-9]+}}
 ; GFX90A:         v_mfma_f32_16x16x16bf16_1k [[RES:a\[[0-9]+:[0-9]+\]]], v[[[ONE]]:{{[0-9]+}}], v[[[TWO]]:{{[0-9]+}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
+; GFX940:         v_mfma_f32_16x16x16_bf16 [[RES:a\[[0-9]+:[0-9]+\]]], v{{\[}}[[ONE]]:{{[0-9+]}}], v{{\[}}[[TWO]]:{{[0-9+]}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
 ; GCN-NOT:        v_accvgpr_read_b32
 ; GCN:            global_store_dwordx4 v{{[0-9]+}}, [[RES]],
 define amdgpu_kernel void @test_mfma_f32_16x16x16bf16_1k(<4 x float> addrspace(1)* %arg) #0 {
@@ -134,6 +147,8 @@ bb:
 ; GCN-LABEL: {{^}}test_mfma_f64_4x4x4f64:
 ; GFX90A: v_mfma_f64_4x4x4f64 [[M1:a\[[0-9]+:[0-9]+\]]], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], 0{{$}}
 ; GFX90A: v_mfma_f64_4x4x4f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], [[M1]] cbsz:1 abid:2 blgp:3
+; GFX940: v_mfma_f64_4x4x4_4b_f64 [[M1:a\[[0-9]+:[0-9]+\]]], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], 0{{$}}
+; GFX940: v_mfma_f64_4x4x4_4b_f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], [[M1]] cbsz:1 abid:2 neg:[1,1,0]
 ; GCN:    global_store_dwordx2
 define amdgpu_kernel void @test_mfma_f64_4x4x4f64(double addrspace(1)* %arg, double %a, double %b) #0 {
 bb:
@@ -146,6 +161,7 @@ bb:
 ; GCN-LABEL: {{^}}test_mfma_f64_16x16x4f64:
 ; GCN:    s_load_dwordx8
 ; GFX90A: v_mfma_f64_16x16x4f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 blgp:3
+; GFX940: v_mfma_f64_16x16x4_f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], a[{{[0-9]+:[0-9]+}}] cbsz:1 abid:2 neg:[1,1,0]
 ; GCN:    global_store_dwordx4
 ; GCN:    global_store_dwordx4
 define amdgpu_kernel void @test_mfma_f64_16x16x4f64(<4 x double> addrspace(1)* %arg, double %a, double %b) #0 {
@@ -159,6 +175,8 @@ bb:
 ; GCN-LABEL: {{^}}test_mfma_f64_16x16x4f64_splat_imm:
 ; GFX90A: v_mfma_f64_16x16x4f64 [[M1:a\[[0-9]+:[0-9]+\]]], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], 0{{$}}
 ; GFX90A: v_mfma_f64_16x16x4f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], [[M1]] cbsz:1 abid:2 blgp:3
+; GFX940: v_mfma_f64_16x16x4_f64 [[M1:a\[[0-9]+:[0-9]+\]]], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], 0{{$}}
+; GFX940: v_mfma_f64_16x16x4_f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], [[M1]] cbsz:1 abid:2 neg:[1,1,0]
 ; GCN:    global_store_dwordx4
 ; GCN:    global_store_dwordx4
 define amdgpu_kernel void @test_mfma_f64_16x16x4f64_splat_imm(<4 x double> addrspace(1)* %arg, double %a, double %b) #0 {
@@ -171,6 +189,7 @@ bb:
 
 ; GCN-LABEL: {{^}}test_mfma_f64_16x16x4f64_imm:
 ; GFX90A: v_mfma_f64_16x16x4f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], a[{{[0-9]+:[0-9]+}}]{{$}}
+; GFX940: v_mfma_f64_16x16x4_f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], a[{{[0-9]+:[0-9]+}}]{{$}}
 ; GCN:    global_store_dwordx4
 ; GCN:    global_store_dwordx4
 define amdgpu_kernel void @test_mfma_f64_16x16x4f64_imm(<4 x double> addrspace(1)* %arg, double %a, double %b) #0 {
@@ -183,7 +202,9 @@ bb:
 ; GCN-LABEL: {{^}}test_mfma_f64_16x16x4f64_splat_lit:
 ; GCN-DAG:    v_accvgpr_write_b32 a{{[0-9]+}}, 0{{$}}
 ; GFX90A-DAG: v_mov_b32_e32 v{{[0-9]+}}, 0x405ec000
+; GFX940-DAG: s_mov_b32 s{{[0-9]+}}, 0x405ec000
 ; GFX90A:     v_mfma_f64_16x16x4f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], a[{{[0-9]+:[0-9]+}}]{{$}}
+; GFX940:     v_mfma_f64_16x16x4_f64 a[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], a[{{[0-9]+:[0-9]+}}]{{$}}
 ; GCN:        global_store_dwordx4
 ; GCN:        global_store_dwordx4
 define amdgpu_kernel void @test_mfma_f64_16x16x4f64_splat_lit(<4 x double> addrspace(1)* %arg, double %a, double %b) #0 {