[AMDGPU] Remove one case of vmcnt loop header flushing for GFX12 #105550
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
This was referenced Aug 21, 2024
This stack of pull requests is managed by Graphite. Learn more about stacking. |
|
@llvm/pr-subscribers-backend-amdgpu Author: Jay Foad (jayfoad) ChangesWhen a loop contains a VMEM load whose result is only used outside the Full diff: https://github.com/llvm/llvm-project/pull/105550.diff 2 Files Affected:
diff --git a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
index 4262e7b5d9c25..eafe20be17d5b 100644
--- a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
@@ -2390,7 +2390,7 @@ bool SIInsertWaitcnts::shouldFlushVmCnt(MachineLoop *ML,
}
if (!ST->hasVscnt() && HasVMemStore && !HasVMemLoad && UsesVgprLoadedOutside)
return true;
- return HasVMemLoad && UsesVgprLoadedOutside;
+ return HasVMemLoad && UsesVgprLoadedOutside && ST->hasVmemWriteVgprInOrder();
}
bool SIInsertWaitcnts::runOnMachineFunction(MachineFunction &MF) {
diff --git a/llvm/test/CodeGen/AMDGPU/waitcnt-vmcnt-loop.mir b/llvm/test/CodeGen/AMDGPU/waitcnt-vmcnt-loop.mir
index bdef55ab956a0..0ddd2aa285b26 100644
--- a/llvm/test/CodeGen/AMDGPU/waitcnt-vmcnt-loop.mir
+++ b/llvm/test/CodeGen/AMDGPU/waitcnt-vmcnt-loop.mir
@@ -295,7 +295,7 @@ body: |
# GFX12-LABEL: waitcnt_vm_loop2
# GFX12-LABEL: bb.0:
# GFX12: BUFFER_LOAD_FORMAT_X_IDXEN
-# GFX12: S_WAIT_LOADCNT 0
+# GFX12-NOT: S_WAIT_LOADCNT 0
# GFX12-LABEL: bb.1:
# GFX12: S_WAIT_LOADCNT 0
# GFX12-LABEL: bb.2:
@@ -342,7 +342,7 @@ body: |
# GFX12-LABEL: waitcnt_vm_loop2_store
# GFX12-LABEL: bb.0:
# GFX12: BUFFER_LOAD_FORMAT_X_IDXEN
-# GFX12: S_WAIT_LOADCNT 0
+# GFX12-NOT: S_WAIT_LOADCNT 0
# GFX12-LABEL: bb.1:
# GFX12: S_WAIT_LOADCNT 0
# GFX12-LABEL: bb.2:
@@ -499,9 +499,9 @@ body: |
# GFX12-LABEL: waitcnt_vm_loop2_reginterval
# GFX12-LABEL: bb.0:
# GFX12: GLOBAL_LOAD_DWORDX4
-# GFX12: S_WAIT_LOADCNT 0
-# GFX12-LABEL: bb.1:
# GFX12-NOT: S_WAIT_LOADCNT 0
+# GFX12-LABEL: bb.1:
+# GFX12: S_WAIT_LOADCNT 0
# GFX12-LABEL: bb.2:
name: waitcnt_vm_loop2_reginterval
body: |
@@ -600,7 +600,7 @@ body: |
# GFX12-LABEL: bb.0:
# GFX12: BUFFER_LOAD_FORMAT_X_IDXEN
# GFX12: BUFFER_LOAD_FORMAT_X_IDXEN
-# GFX12: S_WAIT_LOADCNT 0
+# GFX12-NOT: S_WAIT_LOADCNT 0
# GFX12-LABEL: bb.1:
# GFX12: S_WAIT_LOADCNT 0
# GFX12-LABEL: bb.2:
|
9a2103d to
c3cbf18
Compare
Base automatically changed from
users/foad/vmem-write-vgpr-in-order_split
to
main
August 22, 2024 10:46
e53f758 to
283d345
Compare
arsenm
approved these changes
Aug 22, 2024
When a loop contains a VMEM load whose result is only used outside the loop, do not bother to flush vmcnt in the loop head on GFX12. A wait for vmcnt will be required inside the loop anyway, because VMEM instructions can write their VGPR results out of order.
283d345 to
ba06857
Compare
llvmbot
pushed a commit
to llvmbot/llvm-project
that referenced
this pull request
Aug 23, 2024
…m#105550) When a loop contains a VMEM load whose result is only used outside the loop, do not bother to flush vmcnt in the loop head on GFX12. A wait for vmcnt will be required inside the loop anyway, because VMEM instructions can write their VGPR results out of order. (cherry picked from commit fa2dccb)
cjdb
pushed a commit
to cjdb/llvm-project
that referenced
this pull request
Aug 23, 2024
…m#105550) When a loop contains a VMEM load whose result is only used outside the loop, do not bother to flush vmcnt in the loop head on GFX12. A wait for vmcnt will be required inside the loop anyway, because VMEM instructions can write their VGPR results out of order.
tru
pushed a commit
to llvmbot/llvm-project
that referenced
this pull request
Sep 1, 2024
…m#105550) When a loop contains a VMEM load whose result is only used outside the loop, do not bother to flush vmcnt in the loop head on GFX12. A wait for vmcnt will be required inside the loop anyway, because VMEM instructions can write their VGPR results out of order. (cherry picked from commit fa2dccb)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
When a loop contains a VMEM load whose result is only used outside the
loop, do not bother to flush vmcnt in the loop head on GFX12. A wait for
vmcnt will be required inside the loop anyway, because VMEM instructions
can write their VGPR results out of order.