[NFC][PowerPC] Optimize vector compares for not equal to non zero vectors #171635
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
Member
|
@llvm/pr-subscribers-backend-powerpc Author: None (Himadhith) ChangesLockdown instructions for vector compares not equal to non-zero. Current implementation can be made better by removing the negation and using the identity Full diff: https://github.com/llvm/llvm-project/pull/171635.diff 1 Files Affected:
diff --git a/llvm/test/CodeGen/PowerPC/optimize-vector-not-equal.ll b/llvm/test/CodeGen/PowerPC/optimize-vector-not-equal.ll
new file mode 100644
index 0000000000000..c3bb2d5ecc461
--- /dev/null
+++ b/llvm/test/CodeGen/PowerPC/optimize-vector-not-equal.ll
@@ -0,0 +1,71 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+; RUN: llc -verify-machineinstrs -mcpu=pwr9 -mtriple=powerpc64le-unknown-linux-gnu \
+; RUN: -ppc-asm-full-reg-names --ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=POWERPC_64LE
+
+; RUN: llc -verify-machineinstrs -mcpu=pwr9 -mtriple=powerpc64-ibm-aix \
+; RUN: -ppc-asm-full-reg-names --ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=POWERPC_64
+
+; RUN: llc -verify-machineinstrs -mcpu=pwr9 -mtriple=powerpc-ibm-aix \
+; RUN: -ppc-asm-full-reg-names --ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=POWERPC_32
+
+; The current implementation is comparing vector of non-zeros in register v2 with v3. v3 is then negated and converts:
+; 0XFFFF -> 0
+; 0 -> 1
+; An optimized version is to follow this NFC patch
+
+define i32 @cols_needed(<4 x i16> %wide.load) {
+; POWERPC_64LE-LABEL: cols_needed:
+; POWERPC_64LE: # %bb.0: # %entry
+; POWERPC_64LE-NEXT: xxlxor v3, v3, v3
+; POWERPC_64LE-NEXT: li r3, 0
+; POWERPC_64LE-NEXT: vcmpequh v2, v2, v3
+; POWERPC_64LE-NEXT: xxleqv v3, v3, v3
+; POWERPC_64LE-NEXT: vmrglh v2, v2, v2
+; POWERPC_64LE-NEXT: vsubuwm v2, v2, v3
+; POWERPC_64LE-NEXT: xxswapd v3, v2
+; POWERPC_64LE-NEXT: vadduwm v2, v2, v3
+; POWERPC_64LE-NEXT: xxspltw v3, v2, 2
+; POWERPC_64LE-NEXT: vadduwm v2, v2, v3
+; POWERPC_64LE-NEXT: vextuwrx r3, r3, v2
+; POWERPC_64LE-NEXT: blr
+;
+; POWERPC_64-LABEL: cols_needed:
+; POWERPC_64: # %bb.0: # %entry
+; POWERPC_64-NEXT: xxlxor v3, v3, v3
+; POWERPC_64-NEXT: li r3, 0
+; POWERPC_64-NEXT: vcmpequh v2, v2, v3
+; POWERPC_64-NEXT: xxleqv v3, v3, v3
+; POWERPC_64-NEXT: vmrghh v2, v2, v2
+; POWERPC_64-NEXT: vsubuwm v2, v2, v3
+; POWERPC_64-NEXT: xxswapd v3, v2
+; POWERPC_64-NEXT: vadduwm v2, v2, v3
+; POWERPC_64-NEXT: xxspltw v3, v2, 1
+; POWERPC_64-NEXT: vadduwm v2, v2, v3
+; POWERPC_64-NEXT: vextuwlx r3, r3, v2
+; POWERPC_64-NEXT: blr
+;
+; POWERPC_32-LABEL: cols_needed:
+; POWERPC_32: # %bb.0: # %entry
+; POWERPC_32-NEXT: xxlxor v3, v3, v3
+; POWERPC_32-NEXT: vcmpequh v2, v2, v3
+; POWERPC_32-NEXT: xxleqv v3, v3, v3
+; POWERPC_32-NEXT: vmrghh v2, v2, v2
+; POWERPC_32-NEXT: vsubuwm v2, v2, v3
+; POWERPC_32-NEXT: xxswapd v3, v2
+; POWERPC_32-NEXT: vadduwm v2, v2, v3
+; POWERPC_32-NEXT: xxspltw v3, v2, 1
+; POWERPC_32-NEXT: vadduwm v2, v2, v3
+; POWERPC_32-NEXT: stxv v2, -16(r1)
+; POWERPC_32-NEXT: lwz r3, -16(r1)
+; POWERPC_32-NEXT: blr
+entry:
+ %0 = icmp ne <4 x i16> %wide.load, zeroinitializer
+ %1 = zext <4 x i1> %0 to <4 x i32>
+ %2 = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %1)
+ ret i32 %2
+}
+
+; Function Attrs: nocallback nofree nosync nounwind speculatable willreturn memory(none)
+declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>) #0
+
+attributes #0 = { nocallback nofree nosync nounwind speculatable willreturn memory(none) }
|
🐧 Linux x64 Test Results
✅ The build succeeded and all tests passed. |
🪟 Windows x64 Test Results
✅ The build succeeded and all tests passed. |
|
This is a comment for testing the issue write workflow |
|
This is another comment for testing the issue write workflow that was placed in a separate file |
Himadhith
force-pushed
the
NFC/VCMP_NE_NON_ZERO_OPT
branch
from
a37d89d to
d056c90
Compare
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/39/builds/9073 Here is the relevant piece of the build log for the reference |
anonymouspc
pushed a commit
to anonymouspc/llvm
that referenced
this pull request
Dec 15, 2025
…tors (llvm#171635) Lockdown instructions for vector compares `not equal to non-zero (Ex: vec[i]!=7)`. Current implementation can be made better by removing the negation and using the identity ``` 0XFFFF + 1 = 0 and 0 + 1 = 0 ``` Co-authored-by: himadhith <himadhith.v@ibm.com>
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.
Lockdown instructions for vector compares
not equal to non-zero (Ex: vec[i]!=7). Current implementation can be made better by removing the negation and using the identity0XFFFF + 1 = 0 and 0 + 1 = 0