[RISCV] Use LiveIntervals to determine if AVL dominates when coalescing #118285
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
In order to coalesce a vsetvli with a register AVL into a previous vsetvli, we need to make sure that the AVL register is reachable at the previous vsetvli. Back in pre-RA vsetvli insertion we just checked to see if the two virtual registers were the same virtual register, and then this was hacked around in the move. We can instead use live intervals to check that the reaching definition is the same at both instructions. On its own this doesn't have much of an impact, but helps a lot in llvm#118283 and enables coalescing in about 60 of the test cases from that PR.
Your org has enabled the Graphite merge queue for merging into mainAdd the label “FP Bundles” to the PR and Graphite will automatically add it to the merge queue when it’s ready to merge. You must have a Graphite account and log in to Graphite in order to use the merge queue. Sign up using this link. |
Member
|
@llvm/pr-subscribers-backend-risc-v Author: Luke Lau (lukel97) ChangesIn order to coalesce a vsetvli with a register AVL into a previous vsetvli, we need to make sure that the AVL register is reachable at the previous vsetvli. Back in pre-RA vsetvli insertion we just checked to see if the two virtual registers were the same virtual register, and then this was hacked around in the move. We can instead use live intervals to check that the reaching definition is the same at both instructions. On its own this doesn't have much of an impact, but helps a lot in #118283 and enables coalescing in about 60 of the test cases from that PR. Full diff: https://github.com/llvm/llvm-project/pull/118285.diff 2 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
index 052b4a61298223..93f9bf5ceba138 100644
--- a/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
+++ b/llvm/lib/Target/RISCV/RISCVInsertVSETVLI.cpp
@@ -1619,14 +1619,15 @@ bool RISCVInsertVSETVLI::canMutatePriorConfig(
}
auto &AVL = MI.getOperand(1);
- auto &PrevAVL = PrevMI.getOperand(1);
- // If the AVL is a register, we need to make sure MI's AVL dominates PrevMI.
- // For now just check that PrevMI uses the same virtual register.
- if (AVL.isReg() && AVL.getReg() != RISCV::X0 &&
- (!MRI->hasOneDef(AVL.getReg()) || !PrevAVL.isReg() ||
- PrevAVL.getReg() != AVL.getReg()))
- return false;
+ // If the AVL is a register, we need to make sure its definition is the same
+ // at PrevMI as it was at MI.
+ if (AVL.isReg() && AVL.getReg() != RISCV::X0) {
+ VNInfo *VNI = getVNInfoFromReg(AVL.getReg(), MI, LIS);
+ VNInfo *PrevVNI = getVNInfoFromReg(AVL.getReg(), PrevMI, LIS);
+ if (!VNI || !PrevVNI || VNI->id != PrevVNI->id)
+ return false;
+ }
}
assert(PrevMI.getOperand(2).isImm() && MI.getOperand(2).isImm());
diff --git a/llvm/test/CodeGen/RISCV/rvv/compressstore.ll b/llvm/test/CodeGen/RISCV/rvv/compressstore.ll
index bfb2d0a3accc44..a407cd048ffe3f 100644
--- a/llvm/test/CodeGen/RISCV/rvv/compressstore.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/compressstore.ll
@@ -652,15 +652,14 @@ define void @test_compresstore_v64i32(ptr %p, <64 x i1> %mask, <64 x i32> %data)
; RV64-NEXT: vse32.v v24, (a0)
; RV64-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64-NEXT: vslidedown.vi v8, v0, 4
-; RV64-NEXT: vsetvli zero, zero, e32, m2, ta, ma
-; RV64-NEXT: vmv.x.s a2, v0
; RV64-NEXT: vsetvli zero, a1, e32, m8, ta, ma
+; RV64-NEXT: vmv.x.s a1, v0
; RV64-NEXT: vcompress.vm v24, v16, v8
-; RV64-NEXT: vcpop.m a1, v8
-; RV64-NEXT: cpopw a2, a2
-; RV64-NEXT: slli a2, a2, 2
-; RV64-NEXT: add a0, a0, a2
-; RV64-NEXT: vsetvli zero, a1, e32, m8, ta, ma
+; RV64-NEXT: vcpop.m a2, v8
+; RV64-NEXT: cpopw a1, a1
+; RV64-NEXT: slli a1, a1, 2
+; RV64-NEXT: add a0, a0, a1
+; RV64-NEXT: vsetvli zero, a2, e32, m8, ta, ma
; RV64-NEXT: vse32.v v24, (a0)
; RV64-NEXT: ret
;
|
topperc
reviewed
Dec 2, 2024
…lsewhere in RISCVInsertVSETVLI and in LiveIntervals.cpp
lukel97
added a commit
to lukel97/llvm-project
that referenced
this pull request
Dec 2, 2024
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/73/builds/9451 Here is the relevant piece of the build log for the reference |
lukel97
added a commit
to lukel97/llvm-project
that referenced
this pull request
Dec 3, 2024
lukel97
added a commit
to lukel97/llvm-project
that referenced
this pull request
Dec 4, 2024
lukel97
added a commit
to lukel97/llvm-project
that referenced
this pull request
Dec 5, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 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.
In order to coalesce a vsetvli with a register AVL into a previous vsetvli, we need to make sure that the AVL register is reachable at the previous vsetvli.
Back in pre-RA vsetvli insertion we just checked to see if the two virtual registers were the same virtual register, and then this was hacked around in the move to post-RA. We can instead use live intervals to check that the reaching definition is the same at both instructions.
On its own this doesn't have much of an impact, but helps a lot in #118283 and enables coalescing in about 60 of the test cases from that PR.