[RISCV] Skip DAG combine for bitcast fabs/fneg #115325
Merged
+6
−12
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
Disable the DAG combine for bitcast fabs/fneg in case of the zdinx extension. The combine folds the fabs/fneg nodes in some cases. This might result in suboptimal code if compiled with the zdinx extension. In case of the zdinx extension, there is no need to load the double value from an x register to an f register, so the combine can be skipped.
Member
|
@llvm/pr-subscribers-backend-risc-v Author: Gergely Futo (futog) ChangesDisable the DAG combine for bitcast fabs/fneg in case of the zdinx extension. The combine folds the fabs/fneg nodes in some cases. This might result in suboptimal code if compiled with the zdinx extension. In case of the zdinx extension, there is no need to load the double value from an x register to an f register, so the combine can be skipped. Full diff: https://github.com/llvm/llvm-project/pull/115325.diff 4 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 5600524b69a620..5e824fde78e859 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -17075,7 +17075,7 @@ SDValue RISCVTargetLowering::PerformDAGCombine(SDNode *N,
// fold (bitconvert (fneg x)) -> (xor (bitconvert x), signbit)
// fold (bitconvert (fabs x)) -> (and (bitconvert x), (not signbit))
if (!(Op0.getOpcode() == ISD::FNEG || Op0.getOpcode() == ISD::FABS) ||
- !Op0.getNode()->hasOneUse())
+ !Op0.getNode()->hasOneUse() || Subtarget.hasStdExtZdinx())
break;
SDValue NewSplitF64 =
DAG.getNode(RISCVISD::SplitF64, DL, DAG.getVTList(MVT::i32, MVT::i32),
diff --git a/llvm/test/CodeGen/RISCV/double-arith.ll b/llvm/test/CodeGen/RISCV/double-arith.ll
index fa74dcb4810060..5f06398daa8b9a 100644
--- a/llvm/test/CodeGen/RISCV/double-arith.ll
+++ b/llvm/test/CodeGen/RISCV/double-arith.ll
@@ -844,8 +844,7 @@ define double @fnmadd_d_3(double %a, double %b, double %c) nounwind {
; RV32IZFINXZDINX-LABEL: fnmadd_d_3:
; RV32IZFINXZDINX: # %bb.0:
; RV32IZFINXZDINX-NEXT: fmadd.d a0, a0, a2, a4
-; RV32IZFINXZDINX-NEXT: lui a2, 524288
-; RV32IZFINXZDINX-NEXT: xor a1, a1, a2
+; RV32IZFINXZDINX-NEXT: fneg.d a0, a0
; RV32IZFINXZDINX-NEXT: ret
;
; RV64IZFINXZDINX-LABEL: fnmadd_d_3:
@@ -890,9 +889,7 @@ define double @fnmadd_nsz(double %a, double %b, double %c) nounwind {
;
; RV32IZFINXZDINX-LABEL: fnmadd_nsz:
; RV32IZFINXZDINX: # %bb.0:
-; RV32IZFINXZDINX-NEXT: fmadd.d a0, a0, a2, a4
-; RV32IZFINXZDINX-NEXT: lui a2, 524288
-; RV32IZFINXZDINX-NEXT: xor a1, a1, a2
+; RV32IZFINXZDINX-NEXT: fnmadd.d a0, a0, a2, a4
; RV32IZFINXZDINX-NEXT: ret
;
; RV64IZFINXZDINX-LABEL: fnmadd_nsz:
diff --git a/llvm/test/CodeGen/RISCV/double-bitmanip-dagcombines.ll b/llvm/test/CodeGen/RISCV/double-bitmanip-dagcombines.ll
index f7d57178b03d41..01aa25c15c8d2b 100644
--- a/llvm/test/CodeGen/RISCV/double-bitmanip-dagcombines.ll
+++ b/llvm/test/CodeGen/RISCV/double-bitmanip-dagcombines.ll
@@ -36,8 +36,7 @@ define double @fneg(double %a) nounwind {
;
; RV32IZFINXZDINX-LABEL: fneg:
; RV32IZFINXZDINX: # %bb.0:
-; RV32IZFINXZDINX-NEXT: lui a2, 524288
-; RV32IZFINXZDINX-NEXT: xor a1, a1, a2
+; RV32IZFINXZDINX-NEXT: fneg.d a0, a0
; RV32IZFINXZDINX-NEXT: ret
;
; RV64I-LABEL: fneg:
@@ -79,8 +78,7 @@ define double @fabs(double %a) nounwind {
;
; RV32IZFINXZDINX-LABEL: fabs:
; RV32IZFINXZDINX: # %bb.0:
-; RV32IZFINXZDINX-NEXT: slli a1, a1, 1
-; RV32IZFINXZDINX-NEXT: srli a1, a1, 1
+; RV32IZFINXZDINX-NEXT: fabs.d a0, a0
; RV32IZFINXZDINX-NEXT: ret
;
; RV64I-LABEL: fabs:
diff --git a/llvm/test/CodeGen/RISCV/double-intrinsics.ll b/llvm/test/CodeGen/RISCV/double-intrinsics.ll
index bf21ee6696a282..a65fd09613424c 100644
--- a/llvm/test/CodeGen/RISCV/double-intrinsics.ll
+++ b/llvm/test/CodeGen/RISCV/double-intrinsics.ll
@@ -678,8 +678,7 @@ define double @fabs_f64(double %a) nounwind {
;
; RV32IZFINXZDINX-LABEL: fabs_f64:
; RV32IZFINXZDINX: # %bb.0:
-; RV32IZFINXZDINX-NEXT: slli a1, a1, 1
-; RV32IZFINXZDINX-NEXT: srli a1, a1, 1
+; RV32IZFINXZDINX-NEXT: fabs.d a0, a0
; RV32IZFINXZDINX-NEXT: ret
;
; RV64IZFINXZDINX-LABEL: fabs_f64:
|
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/195/builds/784 Here is the relevant piece of the build log for the reference |
Groverkss
pushed a commit
to iree-org/llvm-project
that referenced
this pull request
Nov 15, 2024
Disable the DAG combine for bitcast fabs/fneg in case of the zdinx extension. The combine folds the fabs/fneg nodes in some cases. This might result in suboptimal code if compiled with the zdinx extension. In case of the zdinx extension, there is no need to load the double value from an x register to an f register, so the combine can be skipped.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Disable the DAG combine for bitcast fabs/fneg in case of the zdinx extension.
The combine folds the fabs/fneg nodes in some cases. This might result in suboptimal code if compiled with the zdinx extension. In case of the zdinx extension, there is no need to load the double value from an x register to an f register, so the combine can be skipped.