Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[AMDGPU] Run twoaddr tests with -early-live-intervals #66775

Merged
merged 1 commit into from
Sep 20, 2023
Merged

[AMDGPU] Run twoaddr tests with -early-live-intervals #66775

merged 1 commit into from
Sep 20, 2023

Conversation

jayfoad
Copy link
Contributor

@jayfoad jayfoad commented Sep 19, 2023

Sample test case:

%3 = V_FMAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec

With LiveVariables this is converted to three-address form just because
there is no "killed" flag on %2. To make it do the same thing with
LiveIntervals I added a later use of %2:

%3 = V_FMAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec
S_ENDPGM 0, implicit %2

@jayfoad
[AMDGPU] Run twoaddr tests with -early-live-intervals
0a382a8 
Sample test case:

    %3 = V_FMAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec

With LiveVariables this is converted to three-address form just because
there is no "killed" flag on %2. To make it do the same thing with
LiveIntervals I added a later use of %2:

    %3 = V_FMAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec
    S_ENDPGM 0, implicit %2
@jayfoad jayfoad requested review from perlfu, piotrAMD and a team September 19, 2023 14:03
@llvmbot
Copy link
Collaborator

llvmbot commented Sep 19, 2023

@llvm/pr-subscribers-backend-amdgpu

Changes

Sample test case:

%3 = V_FMAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec

With LiveVariables this is converted to three-address form just because
there is no "killed" flag on %2. To make it do the same thing with
LiveIntervals I added a later use of %2:

%3 = V_FMAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec
S_ENDPGM 0, implicit %2

Full diff: https://github.com/llvm/llvm-project/pull/66775.diff

5 Files Affected:

  • (modified) llvm/test/CodeGen/AMDGPU/gfx10-twoaddr-fma.mir (+5-5)
  • (modified) llvm/test/CodeGen/AMDGPU/twoaddr-fma-f64.mir (+5-4)
  • (modified) llvm/test/CodeGen/AMDGPU/twoaddr-fma.mir (+10-5)
  • (modified) llvm/test/CodeGen/AMDGPU/twoaddr-mad.mir (+10-6)
  • (modified) llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir (+13-12)
diff --git a/llvm/test/CodeGen/AMDGPU/gfx10-twoaddr-fma.mir b/llvm/test/CodeGen/AMDGPU/gfx10-twoaddr-fma.mir
index 367c1fb10af53cc..2fe4d9d8e9e9a7c 100644
--- a/llvm/test/CodeGen/AMDGPU/gfx10-twoaddr-fma.mir
+++ b/llvm/test/CodeGen/AMDGPU/gfx10-twoaddr-fma.mir
@@ -1,4 +1,5 @@
 # RUN: llc -march=amdgcn -mcpu=gfx1010 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - | FileCheck --check-prefixes=GFX10 %s
+# RUN: llc -march=amdgcn -mcpu=gfx1010 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - -early-live-intervals | FileCheck --check-prefixes=GFX10 %s
 
 # GFX10-LABEL: name: test_fmamk_reg_imm_f16
 # GFX10: %2:vgpr_32 = IMPLICIT_DEF
@@ -43,7 +44,7 @@ body:             |
 ...
 
 # GFX10-LABEL: name: test_fmaak_f16
-# GFX10: %1:vgpr_32 = IMPLICIT_DEF
+# GFX10: %1:vgpr_32 = V_MOV_B32_e32 1078523331, implicit $exec
 # GFX10-NOT: V_MOV_B32
 # GFX10: V_FMAAK_F16 killed %0.sub0, %0.sub1, 1078523331, implicit $mode, implicit $exec
 ---
@@ -58,10 +59,11 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = V_MOV_B32_e32 1078523331, implicit $exec
     %2 = V_FMAC_F16_e32 killed %0.sub0, %0.sub1, %1, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %1
 ...
 
 # GFX10-LABEL: name: test_fmaak_inline_literal_f16
-# GFX10: %1:vgpr_32 = IMPLICIT_DEF
+# GFX10: %1:vgpr_32 = V_MOV_B32_e32 49664, implicit $exec
 # GFX10-NOT: V_MOV_B32
 # GFX10: %2:vgpr_32 = V_FMAAK_F16 16384, killed %0, 49664, implicit $mode, implicit $exec
 
@@ -78,7 +80,5 @@ body:             |
 
     %1:vgpr_32 = V_MOV_B32_e32 49664, implicit $exec
     %2:vgpr_32 = V_FMAC_F16_e32 16384, killed %0, %1, implicit $mode, implicit $exec
-    S_ENDPGM 0
-
+    S_ENDPGM 0, implicit %1
 ...
-
diff --git a/llvm/test/CodeGen/AMDGPU/twoaddr-fma-f64.mir b/llvm/test/CodeGen/AMDGPU/twoaddr-fma-f64.mir
index db5cf285c109718..717893a9576446f 100644
--- a/llvm/test/CodeGen/AMDGPU/twoaddr-fma-f64.mir
+++ b/llvm/test/CodeGen/AMDGPU/twoaddr-fma-f64.mir
@@ -1,4 +1,5 @@
 # RUN: llc -march=amdgcn -mcpu=gfx90a %s -run-pass twoaddressinstruction -verify-machineinstrs -o - | FileCheck -check-prefix=GCN %s
+# RUN: llc -march=amdgcn -mcpu=gfx90a %s -run-pass twoaddressinstruction -verify-machineinstrs -o - -early-live-intervals | FileCheck -check-prefix=GCN %s
 
 # GCN-LABEL: name: test_fmamk_reg_imm_f64
 # GCN: V_FMA_F64_e64 0, killed %0, 0, %2, 0, killed %1, 0, 0, implicit $mode, implicit $exec
@@ -52,7 +53,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = V_MOV_B64_PSEUDO 4607182418800017408, implicit $exec
     %2 = V_FMAC_F64_e32 killed %0.sub0_sub1, %0.sub2_sub3, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 
 # GCN-LABEL: name: test_fmaak_sgpr_src0_f64
@@ -72,7 +73,7 @@ body:             |
     %1 = V_MOV_B64_PSEUDO 4607182418800017408, implicit $exec
     %2 = IMPLICIT_DEF
     %3 = V_FMAC_F64_e32 killed %0, %1, %2, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %2
 ...
 
 # GCN-LABEL: name: test_fmaak_inlineimm_src0_f64
@@ -90,7 +91,7 @@ body:             |
     %0 = V_MOV_B64_PSEUDO 4607182418800017408, implicit $exec
     %1 = IMPLICIT_DEF
     %2 = V_FMAC_F64_e32 4611686018427387904, %0, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 
 # GCN-LABEL: name: test_fmaak_otherimm_src0_f64
@@ -186,5 +187,5 @@ body:             |
     %1 = COPY %0
     %2 = V_MOV_B64_PSEUDO 123456, implicit $exec
     %3 = V_FMAC_F64_e32 killed %0, killed %1, %2, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %2
 ...
diff --git a/llvm/test/CodeGen/AMDGPU/twoaddr-fma.mir b/llvm/test/CodeGen/AMDGPU/twoaddr-fma.mir
index 7757c758134c750..412457fd0999f4e 100644
--- a/llvm/test/CodeGen/AMDGPU/twoaddr-fma.mir
+++ b/llvm/test/CodeGen/AMDGPU/twoaddr-fma.mir
@@ -1,5 +1,7 @@
 # RUN: llc -march=amdgcn -mcpu=gfx1010 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - | FileCheck --check-prefixes=GCN %s
+# RUN: llc -march=amdgcn -mcpu=gfx1010 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - -early-live-intervals | FileCheck --check-prefixes=GCN %s
 # RUN: llc -march=amdgcn -mcpu=gfx1100 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - | FileCheck --check-prefixes=GCN %s
+# RUN: llc -march=amdgcn -mcpu=gfx1100 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - -early-live-intervals | FileCheck --check-prefixes=GCN %s
 
 # GCN-LABEL: name: test_fmamk_reg_imm_f32
 # GCN: %2:vgpr_32 = IMPLICIT_DEF
@@ -44,7 +46,7 @@ body:             |
 ...
 
 # GCN-LABEL: name: test_fmaak_f32
-# GCN: %1:vgpr_32 = IMPLICIT_DEF
+# GCN: %1:vgpr_32 = V_MOV_B32_e32 1078523331, implicit $exec
 # GCN-NOT: V_MOV_B32
 # GCN: V_FMAAK_F32 killed %0.sub0, %0.sub1, 1078523331, implicit $mode, implicit $exec
 ---
@@ -59,7 +61,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = V_MOV_B32_e32 1078523331, implicit $exec
     %2 = V_FMAC_F32_e32 killed %0.sub0, %0.sub1, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 
 
@@ -82,7 +84,7 @@ body:             |
     %1 = V_MOV_B32_e32 1078523331, implicit $exec
     %2 = IMPLICIT_DEF
     %3 = V_FMAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %2
 ...
 
 # GCN-LABEL: name: test_fmaak_inlineimm_src0_f32
@@ -102,7 +104,7 @@ body:             |
     %0 = V_MOV_B32_e32 1078523331, implicit $exec
     %1 = IMPLICIT_DEF
     %2 = V_FMAC_F32_e32 1073741824, %0, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 
 # GCN-LABEL: name: test_fmaak_otherimm_src0_f32
@@ -120,7 +122,7 @@ body:             |
     %0 = V_MOV_B32_e32 1078523331, implicit $exec
     %1 = IMPLICIT_DEF
     %2 = V_FMAC_F32_e32 1120403456, %0, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 
 # GCN-LABEL: name: test_fmaak_other_constantlike_src0_f32
@@ -204,6 +206,7 @@ body: |
     %1:vgpr_32 = IMPLICIT_DEF
     %2:vgpr_32 = IMPLICIT_DEF
     %3:vgpr_32 = V_FMAC_F32_e64 1, %0, 0, %1, 0, %2, 0, 0, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %2
 ...
 
 # GCN-LABEL: name: test_fma_src1mods
@@ -216,6 +219,7 @@ body: |
     %1:vgpr_32 = IMPLICIT_DEF
     %2:vgpr_32 = IMPLICIT_DEF
     %3:vgpr_32 = V_FMAC_F32_e64 0, %0, 1, %1, 0, %2, 0, 0, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %2
 ...
 
 # GCN-LABEL: name: test_fma_src2mods
@@ -228,4 +232,5 @@ body: |
     %1:vgpr_32 = IMPLICIT_DEF
     %2:vgpr_32 = IMPLICIT_DEF
     %3:vgpr_32 = V_FMAC_F32_e64 0, %0, 0, %1, 1, %2, 0, 0, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %2
 ...
diff --git a/llvm/test/CodeGen/AMDGPU/twoaddr-mad.mir b/llvm/test/CodeGen/AMDGPU/twoaddr-mad.mir
index 1458e8135ef2da3..d470e3adb2a884c 100644
--- a/llvm/test/CodeGen/AMDGPU/twoaddr-mad.mir
+++ b/llvm/test/CodeGen/AMDGPU/twoaddr-mad.mir
@@ -1,4 +1,5 @@
 # RUN: llc -march=amdgcn -mcpu=gfx900 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - | FileCheck -check-prefix=GCN %s
+# RUN: llc -march=amdgcn -mcpu=gfx900 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - -early-live-intervals | FileCheck -check-prefix=GCN %s
 
 # GCN-LABEL: name: test_madmk_reg_imm_f32
 # GCN: V_MADMK_F32 killed %0.sub0, 1078523331, killed %1, implicit $mode, implicit $exec
@@ -52,7 +53,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = V_MOV_B32_e32 1078523331, implicit $exec
     %2 = V_MAC_F32_e32 killed %0.sub0, %0.sub1, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 
 # GCN-LABEL: name: test_madmk_reg_imm_f16
@@ -107,6 +108,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = V_MOV_B32_e32 1078523331, implicit $exec
     %2 = V_MAC_F16_e32 killed %0.sub0, %0.sub1, %1, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %1
 ...
 
 # Make sure constant bus restriction isn't violated if src0 is an SGPR.
@@ -129,7 +131,7 @@ body:             |
     %1 = V_MOV_B32_e32 1078523331, implicit $exec
     %2 = IMPLICIT_DEF
     %3 = V_MAC_F32_e32 killed %0, %1, %2, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %2
 ...
 
 # This can still fold if this is an inline immediate.
@@ -149,7 +151,7 @@ body:             |
     %0 = V_MOV_B32_e32 1078523331, implicit $exec
     %1 = IMPLICIT_DEF
     %2 = V_MAC_F32_e32 1073741824, %0, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 # Non-inline immediate uses constant bus already.
 
@@ -168,7 +170,7 @@ body:             |
     %0 = V_MOV_B32_e32 1078523331, implicit $exec
     %1 = IMPLICIT_DEF
     %2 = V_MAC_F32_e32 1120403456, %0, %1, implicit $mode, implicit $exec
-
+    S_ENDPGM 0, implicit %1
 ...
 # Non-inline immediate uses constant bus already.
 
@@ -204,8 +206,7 @@ body:             |
 
     %26:vgpr_32 = V_MOV_B32_e32 49664, implicit $exec
     %28:vgpr_32 = V_MAC_F16_e32 16384, killed %3, %26, implicit $mode, implicit $exec
-    S_ENDPGM 0
-
+    S_ENDPGM 0, implicit %26
 ...
 
 # GCN-LABEL: name: test_mad_src0mods
@@ -218,6 +219,7 @@ body: |
     %1:vgpr_32 = IMPLICIT_DEF
     %2:vgpr_32 = IMPLICIT_DEF
     %3:vgpr_32 = V_MAC_F32_e64 1, %0, 0, %1, 0, %2, 0, 0, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %2
 ...
 
 # GCN-LABEL: name: test_mad_src1mods
@@ -230,6 +232,7 @@ body: |
     %1:vgpr_32 = IMPLICIT_DEF
     %2:vgpr_32 = IMPLICIT_DEF
     %3:vgpr_32 = V_MAC_F32_e64 0, %0, 1, %1, 0, %2, 0, 0, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %2
 ...
 
 # GCN-LABEL: name: test_mad_src2mods
@@ -242,4 +245,5 @@ body: |
     %1:vgpr_32 = IMPLICIT_DEF
     %2:vgpr_32 = IMPLICIT_DEF
     %3:vgpr_32 = V_MAC_F32_e64 0, %0, 0, %1, 1, %2, 0, 0, implicit $mode, implicit $exec
+    S_ENDPGM 0, implicit %2
 ...
diff --git a/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir b/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir
index a74d442534c128e..fc5b942804ca41f 100644
--- a/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir
+++ b/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir
@@ -1,4 +1,5 @@
 # RUN: llc -march=amdgcn -mcpu=gfx1100 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - | FileCheck -check-prefix=GCN %s
+# RUN: llc -march=amdgcn -mcpu=gfx1100 %s -run-pass twoaddressinstruction -verify-machineinstrs -o - -early-live-intervals | FileCheck -check-prefix=GCN %s
 
 # GCN-LABEL: name: test_v_wmma_f32_16x16x16_f16_twoaddr_w32
 # GCN: early-clobber %2:vreg_256 = V_WMMA_F32_16X16X16_F16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec
@@ -15,7 +16,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_256 = V_WMMA_F32_16X16X16_F16_twoaddr_w32 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_256, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_f32_16x16x16_bf16_twoaddr_w32
@@ -33,7 +34,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_256 = V_WMMA_F32_16X16X16_BF16_twoaddr_w32 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_256, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_f16_16x16x16_f16_twoaddr_w32
@@ -51,7 +52,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_256 = V_WMMA_F16_16X16X16_F16_twoaddr_w32 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_256, 0, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_bf16_16x16x16_bf16_twoaddr_w32
@@ -69,7 +70,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_256 = V_WMMA_BF16_16X16X16_BF16_twoaddr_w32 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_256, 0, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu8_twoaddr_w32
@@ -87,7 +88,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_256 = V_WMMA_I32_16X16X16_IU8_twoaddr_w32 8, killed %1:vreg_128, 8, killed %1:vreg_128, 8, %0:vreg_256, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu4_twoaddr_w32
@@ -105,7 +106,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_256 = V_WMMA_I32_16X16X16_IU4_twoaddr_w32 8, killed %1:vreg_64, 8, killed %1:vreg_64, 8, %0:vreg_256, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_f32_16x16x16_f16_twoaddr_w64
@@ -123,7 +124,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_128 = V_WMMA_F32_16X16X16_F16_twoaddr_w64 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_128, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_f32_16x16x16_bf16_twoaddr_w64
@@ -141,7 +142,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_128 = V_WMMA_F32_16X16X16_BF16_twoaddr_w64 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_128, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_f16_16x16x16_f16_twoaddr_w64
@@ -159,7 +160,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_128 = V_WMMA_F16_16X16X16_F16_twoaddr_w64 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_128, 0, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_bf16_16x16x16_bf16_twoaddr_w64
@@ -177,7 +178,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_128 = V_WMMA_BF16_16X16X16_BF16_twoaddr_w64 8, killed %1:vreg_256, 8, killed %1:vreg_256, 8, %0:vreg_128, 0, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu8_twoaddr_w64
@@ -195,7 +196,7 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_128 = V_WMMA_I32_16X16X16_IU8_twoaddr_w64 8, killed %1:vreg_128, 8, killed %1:vreg_128, 8, %0:vreg_128, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...
 
 # GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu4_twoaddr_w64
@@ -213,5 +214,5 @@ body:             |
     %0 = IMPLICIT_DEF
     %1 = IMPLICIT_DEF
     early-clobber %2:vreg_128 = V_WMMA_I32_16X16X16_IU4_twoaddr_w64 8, killed %1:vreg_64, 8, killed %1:vreg_64, 8, %0:vreg_128, 0, 0, 0, implicit $exec
-
+    S_ENDPGM 0, implicit %0
 ...

@jayfoad
Copy link
Contributor Author

jayfoad commented Sep 19, 2023

The motivation is to remove some spurious test failures when -early-live-intervals is enabled, in the hope that one day we can enable it by default.

perlfu
perlfu approved these changes Sep 20, 2023
View reviewed changes
Copy link
Contributor

@perlfu perlfu left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM - seems like an obvious expansion of test coverage

@jayfoad jayfoad merged commit a68c724 into llvm:main Sep 20, 2023
3 checks passed
This was referenced Sep 20, 2023
Closed
Closed
@kstoimenov kstoimenov mentioned this pull request Sep 22, 2023
Closed
@jayfoad jayfoad deleted the test-twoaddr-early-live-intervals branch October 9, 2023 10:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@perlfu perlfu perlfu approved these changes

@piotrAMD piotrAMD Awaiting requested review from piotrAMD

Assignees
No one assigned
Labels
backend:AMDGPU
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
3 participants
@jayfoad @llvmbot @perlfu