AMDGPU: Don't duplicate implicit operands in 3-address conversion #168426
+14
−14
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
We previously got a duplicate implicit $exec operand. It didn't really hurt anything (other than being a slight drag on compile-time performance). Still, let's keep things clean. commit-id:203b6f66
Member
|
@llvm/pr-subscribers-backend-amdgpu Author: Nicolai Hähnle (nhaehnle) ChangesWe previously got a duplicate implicit $exec operand. It didn't really Full diff: https://github.com/llvm/llvm-project/pull/168426.diff 2 Files Affected:
diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
index 7cb7f47ddb220..a7333e3373f38 100644
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -4187,7 +4187,7 @@ SIInstrInfo::convertToThreeAddressImpl(MachineInstr &MI,
if (NewMFMAOpc != -1) {
MachineInstrBuilder MIB =
BuildMI(MBB, MI, MI.getDebugLoc(), get(NewMFMAOpc));
- for (unsigned I = 0, E = MI.getNumOperands(); I != E; ++I)
+ for (unsigned I = 0, E = MI.getNumExplicitOperands(); I != E; ++I)
MIB.add(MI.getOperand(I));
return MIB;
}
@@ -4196,7 +4196,7 @@ SIInstrInfo::convertToThreeAddressImpl(MachineInstr &MI,
unsigned NewOpc = AMDGPU::mapWMMA2AddrTo3AddrOpcode(MI.getOpcode());
MachineInstrBuilder MIB = BuildMI(MBB, MI, MI.getDebugLoc(), get(NewOpc))
.setMIFlags(MI.getFlags());
- for (unsigned I = 0, E = MI.getNumOperands(); I != E; ++I)
+ for (unsigned I = 0, E = MI.getNumExplicitOperands(); I != E; ++I)
MIB->addOperand(MI.getOperand(I));
return MIB;
}
diff --git a/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir b/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir
index 98d2eca213aae..3aacfcd57b2de 100644
--- a/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir
+++ b/llvm/test/CodeGen/AMDGPU/twoaddr-wmma.mir
@@ -2,7 +2,7 @@
# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 %s --passes=two-address-instruction -verify-each -o - | 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
+# GCN: early-clobber %2:vreg_256 = V_WMMA_F32_16X16X16_F16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_f32_16x16x16_f16_twoaddr_w32
@@ -20,7 +20,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_f32_16x16x16_bf16_twoaddr_w32
-# GCN: early-clobber %2:vreg_256 = V_WMMA_F32_16X16X16_BF16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_256 = V_WMMA_F32_16X16X16_BF16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_f32_16x16x16_bf16_twoaddr_w32
@@ -38,7 +38,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_f16_16x16x16_f16_twoaddr_w32
-# GCN: early-clobber %2:vreg_256 = V_WMMA_F16_16X16X16_F16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_256 = V_WMMA_F16_16X16X16_F16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_f16_16x16x16_f16_twoaddr_w32
@@ -56,7 +56,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_bf16_16x16x16_bf16_twoaddr_w32
-# GCN: early-clobber %2:vreg_256 = V_WMMA_BF16_16X16X16_BF16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_256 = V_WMMA_BF16_16X16X16_BF16_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_bf16_16x16x16_bf16_twoaddr_w32
@@ -74,7 +74,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu8_twoaddr_w32
-# GCN: early-clobber %2:vreg_256 = V_WMMA_I32_16X16X16_IU8_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_256 = V_WMMA_I32_16X16X16_IU8_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_i32_16x16x16_iu8_twoaddr_w32
@@ -92,7 +92,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu4_twoaddr_w32
-# GCN: early-clobber %2:vreg_256 = V_WMMA_I32_16X16X16_IU4_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_256 = V_WMMA_I32_16X16X16_IU4_threeaddr_w32 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_i32_16x16x16_iu4_twoaddr_w32
@@ -110,7 +110,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_f32_16x16x16_f16_twoaddr_w64
-# GCN: early-clobber %2:vreg_128 = V_WMMA_F32_16X16X16_F16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_128 = V_WMMA_F32_16X16X16_F16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_f32_16x16x16_f16_twoaddr_w64
@@ -128,7 +128,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_f32_16x16x16_bf16_twoaddr_w64
-# GCN: early-clobber %2:vreg_128 = V_WMMA_F32_16X16X16_BF16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_128 = V_WMMA_F32_16X16X16_BF16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_f32_16x16x16_bf16_twoaddr_w64
@@ -146,7 +146,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_f16_16x16x16_f16_twoaddr_w64
-# GCN: early-clobber %2:vreg_128 = V_WMMA_F16_16X16X16_F16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_128 = V_WMMA_F16_16X16X16_F16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_f16_16x16x16_f16_twoaddr_w64
@@ -164,7 +164,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_bf16_16x16x16_bf16_twoaddr_w64
-# GCN: early-clobber %2:vreg_128 = V_WMMA_BF16_16X16X16_BF16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_128 = V_WMMA_BF16_16X16X16_BF16_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_bf16_16x16x16_bf16_twoaddr_w64
@@ -182,7 +182,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu8_twoaddr_w64
-# GCN: early-clobber %2:vreg_128 = V_WMMA_I32_16X16X16_IU8_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_128 = V_WMMA_I32_16X16X16_IU8_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_i32_16x16x16_iu8_twoaddr_w64
@@ -200,7 +200,7 @@ body: |
...
# GCN-LABEL: name: test_v_wmma_i32_16x16x16_iu4_twoaddr_w64
-# GCN: early-clobber %2:vreg_128 = V_WMMA_I32_16X16X16_IU4_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec
+# GCN: early-clobber %2:vreg_128 = V_WMMA_I32_16X16X16_IU4_threeaddr_w64 8, killed %1, 8, killed %1, 8, %0, 0, 0, 0, implicit $exec{{$}}
---
name: test_v_wmma_i32_16x16x16_iu4_twoaddr_w64
|
🐧 Linux x64 Test Results
|
Sisyph
approved these changes
Nov 20, 2025
arsenm
approved these changes
Nov 20, 2025
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/10/builds/17744 Here is the relevant piece of the build log for the reference |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
We previously got a duplicate implicit $exec operand. It didn't really
hurt anything (other than being a slight drag on compile-time
performance). Still, let's keep things clean.