[mlir][spirv] Add LIT test for vector.interleave to spirv.VectorShuffle #93189

angelz913 · 2024-05-23T13:17:37Z

The plan is to replace this with a dedicated pattern fomvector.interleave to spirv.shuffle in the future.

github-actions · 2024-05-23T13:17:56Z

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be
notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write
permissions for the repository. In which case you can instead tag reviewers by
name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review
by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate
is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

llvmbot · 2024-05-23T13:18:26Z

@llvm/pr-subscribers-mlir-spirv

Author: Angel Zhang (angelz913)

Changes

Separate vector interleave lowering and to-shuffle rewrite patterns from dialect conversions.

Full diff: https://github.com/llvm/llvm-project/pull/93189.diff

3 Files Affected:

(modified) mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp (-2)
(modified) mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp (+11)
(modified) mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir (+12)

diff --git a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp
index c2dd37f481466..d05e6962fc8fe 100644
--- a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp
+++ b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp
@@ -830,8 +830,6 @@ void mlir::populateVectorToSPIRVPatterns(SPIRVTypeConverter &typeConverter,
   patterns.add<VectorReductionToFPDotProd>(typeConverter, patterns.getContext(),
                                            PatternBenefit(2));
 
-  // Need this until vector.interleave is handled.
-  vector::populateVectorInterleaveToShufflePatterns(patterns);
 }
 
 void mlir::populateVectorReductionToSPIRVDotProductPatterns(
diff --git a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp
index 1932de1be603b..1de0512783af3 100644
--- a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp
+++ b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp
@@ -15,8 +15,10 @@
 #include "mlir/Conversion/VectorToSPIRV/VectorToSPIRV.h"
 #include "mlir/Dialect/SPIRV/IR/SPIRVDialect.h"
 #include "mlir/Dialect/SPIRV/Transforms/SPIRVConversion.h"
+#include "mlir/Dialect/Vector/Transforms/LoweringPatterns.h"
 #include "mlir/Pass/Pass.h"
 #include "mlir/Transforms/DialectConversion.h"
+#include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 
 namespace mlir {
 #define GEN_PASS_DEF_CONVERTVECTORTOSPIRV
@@ -36,6 +38,15 @@ void ConvertVectorToSPIRVPass::runOnOperation() {
   MLIRContext *context = &getContext();
   Operation *op = getOperation();
 
+  // Rewrite patterns need to be matched separately from the dialect conversion
+  {
+    RewritePatternSet patterns(context);
+    vector::populateVectorInterleaveLoweringPatterns(patterns);
+    vector::populateVectorInterleaveToShufflePatterns(patterns);
+    if (failed(applyPatternsAndFoldGreedily(op, std::move(patterns))))
+      return signalPassFailure();
+  }
+
   auto targetAttr = spirv::lookupTargetEnvOrDefault(op);
   std::unique_ptr<ConversionTarget> target =
       SPIRVConversionTarget::get(targetAttr);
diff --git a/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir b/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir
index cddc4ee385357..6fb40d2beff1a 100644
--- a/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir
+++ b/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir
@@ -483,6 +483,18 @@ func.func @shuffle(%v0 : vector<1xi32>, %v1: vector<1xi32>) -> vector<2xi32> {
 
 // -----
 
+// CHECK-LABEL: func @interleave
+//  CHECK-SAME: (%[[ARG0:.+]]: vector<2xf32>, %[[ARG1:.+]]: vector<2xf32>)
+//       CHECK: %[[SHUFFLE:.*]] = spirv.VectorShuffle [0 : i32, 2 : i32, 1 : i32, 3 : i32] %[[ARG0]], %[[ARG1]] : vector<2xf32>, vector<2xf32> -> vector<4xf32>
+//       CHECK: return %[[SHUFFLE]]
+func.func @interleave(%a: vector<2xf32>, %b: vector<2xf32>) -> vector<4xf32>
+{
+  %0 = vector.interleave %a, %b : vector<2xf32>
+  return %0 : vector<4xf32>
+}
+
+// -----
+
 // CHECK-LABEL: func @reduction_add
 //  CHECK-SAME: (%[[V:.+]]: vector<4xi32>)
 //       CHECK:   %[[S0:.+]] = spirv.CompositeExtract %[[V]][0 : i32] : vector<4xi32>

llvmbot · 2024-05-23T13:18:26Z

@llvm/pr-subscribers-mlir

Author: Angel Zhang (angelz913)

Changes

Separate vector interleave lowering and to-shuffle rewrite patterns from dialect conversions.

Full diff: https://github.com/llvm/llvm-project/pull/93189.diff

3 Files Affected:

(modified) mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp (-2)
(modified) mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp (+11)
(modified) mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir (+12)

diff --git a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp
index c2dd37f481466..d05e6962fc8fe 100644
--- a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp
+++ b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRV.cpp
@@ -830,8 +830,6 @@ void mlir::populateVectorToSPIRVPatterns(SPIRVTypeConverter &typeConverter,
   patterns.add<VectorReductionToFPDotProd>(typeConverter, patterns.getContext(),
                                            PatternBenefit(2));
 
-  // Need this until vector.interleave is handled.
-  vector::populateVectorInterleaveToShufflePatterns(patterns);
 }
 
 void mlir::populateVectorReductionToSPIRVDotProductPatterns(
diff --git a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp
index 1932de1be603b..1de0512783af3 100644
--- a/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp
+++ b/mlir/lib/Conversion/VectorToSPIRV/VectorToSPIRVPass.cpp
@@ -15,8 +15,10 @@
 #include "mlir/Conversion/VectorToSPIRV/VectorToSPIRV.h"
 #include "mlir/Dialect/SPIRV/IR/SPIRVDialect.h"
 #include "mlir/Dialect/SPIRV/Transforms/SPIRVConversion.h"
+#include "mlir/Dialect/Vector/Transforms/LoweringPatterns.h"
 #include "mlir/Pass/Pass.h"
 #include "mlir/Transforms/DialectConversion.h"
+#include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 
 namespace mlir {
 #define GEN_PASS_DEF_CONVERTVECTORTOSPIRV
@@ -36,6 +38,15 @@ void ConvertVectorToSPIRVPass::runOnOperation() {
   MLIRContext *context = &getContext();
   Operation *op = getOperation();
 
+  // Rewrite patterns need to be matched separately from the dialect conversion
+  {
+    RewritePatternSet patterns(context);
+    vector::populateVectorInterleaveLoweringPatterns(patterns);
+    vector::populateVectorInterleaveToShufflePatterns(patterns);
+    if (failed(applyPatternsAndFoldGreedily(op, std::move(patterns))))
+      return signalPassFailure();
+  }
+
   auto targetAttr = spirv::lookupTargetEnvOrDefault(op);
   std::unique_ptr<ConversionTarget> target =
       SPIRVConversionTarget::get(targetAttr);
diff --git a/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir b/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir
index cddc4ee385357..6fb40d2beff1a 100644
--- a/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir
+++ b/mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir
@@ -483,6 +483,18 @@ func.func @shuffle(%v0 : vector<1xi32>, %v1: vector<1xi32>) -> vector<2xi32> {
 
 // -----
 
+// CHECK-LABEL: func @interleave
+//  CHECK-SAME: (%[[ARG0:.+]]: vector<2xf32>, %[[ARG1:.+]]: vector<2xf32>)
+//       CHECK: %[[SHUFFLE:.*]] = spirv.VectorShuffle [0 : i32, 2 : i32, 1 : i32, 3 : i32] %[[ARG0]], %[[ARG1]] : vector<2xf32>, vector<2xf32> -> vector<4xf32>
+//       CHECK: return %[[SHUFFLE]]
+func.func @interleave(%a: vector<2xf32>, %b: vector<2xf32>) -> vector<4xf32>
+{
+  %0 = vector.interleave %a, %b : vector<2xf32>
+  return %0 : vector<4xf32>
+}
+
+// -----
+
 // CHECK-LABEL: func @reduction_add
 //  CHECK-SAME: (%[[V:.+]]: vector<4xi32>)
 //       CHECK:   %[[S0:.+]] = spirv.CompositeExtract %[[V]][0 : i32] : vector<4xi32>

github-actions · 2024-05-23T14:17:52Z

✅ With the latest revision this PR passed the C/C++ code formatter.

kuhar

Just one nit, looks good otherwise!

mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir

Co-authored-by: Jakub Kuderski <kubakuderski@gmail.com>

kuhar

LGTM. Thanks.

github-actions · 2024-05-23T17:32:10Z

@angelz913 Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested
by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as
the builds can include changes from many authors. It is not uncommon for your
change to be included in a build that fails due to someone else's changes, or
infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself.
This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

angelz913 requested review from antiagainst and kuhar as code owners May 23, 2024 13:17

llvmbot added mlir:spirv mlir labels May 23, 2024

angelz913 force-pushed the vector-interleave-fix branch from ac31713 to 5bd7506 Compare May 23, 2024 13:24

kuhar changed the title ~~[MLIR][SPIR-V] Fix bugs in VectorToSPIRV Pass~~ [milir][spirv] Fix vector interleave lowering in VectorToSPIRV May 23, 2024

angelz913 force-pushed the vector-interleave-fix branch from 5bd7506 to 769cfee Compare May 23, 2024 15:24

angelz913 changed the title ~~[milir][spirv] Fix vector interleave lowering in VectorToSPIRV~~ [mlir][spirv] Add LIT test for vector.interleave to spirv.VectorShuffle May 23, 2024

angelz913 force-pushed the vector-interleave-fix branch from 769cfee to 8ac0355 Compare May 23, 2024 15:32

[MLIR][SPIR-V] Add LIT test for vector.interleave to spirv.VectorShuffle

a2178ed

angelz913 force-pushed the vector-interleave-fix branch from 33851ce to a2178ed Compare May 23, 2024 16:18

kuhar reviewed May 23, 2024

View reviewed changes

mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir Outdated Show resolved Hide resolved

angelz913 and others added 2 commits May 23, 2024 17:02

Rename iaddcarry_extended.mlir to addui_extended.mlir

6520e54

Update mlir/test/Conversion/VectorToSPIRV/vector-to-spirv.mlir

9ac1a56

Co-authored-by: Jakub Kuderski <kubakuderski@gmail.com>

kuhar approved these changes May 23, 2024

View reviewed changes

kuhar merged commit d70cab4 into llvm:main May 23, 2024
3 checks passed

angelz913 deleted the vector-interleave-fix branch June 6, 2024 19:34

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[mlir][spirv] Add LIT test for vector.interleave to spirv.VectorShuffle #93189

[mlir][spirv] Add LIT test for vector.interleave to spirv.VectorShuffle #93189

angelz913 commented May 23, 2024 •

edited by kuhar

Loading

github-actions bot commented May 23, 2024

llvmbot commented May 23, 2024

llvmbot commented May 23, 2024

github-actions bot commented May 23, 2024 •

edited

Loading

kuhar left a comment

kuhar left a comment

github-actions bot commented May 23, 2024

[mlir][spirv] Add LIT test for vector.interleave to spirv.VectorShuffle #93189

[mlir][spirv] Add LIT test for vector.interleave to spirv.VectorShuffle #93189

Conversation

angelz913 commented May 23, 2024 • edited by kuhar Loading

github-actions bot commented May 23, 2024

llvmbot commented May 23, 2024

llvmbot commented May 23, 2024

github-actions bot commented May 23, 2024 • edited Loading

kuhar left a comment

Choose a reason for hiding this comment

kuhar left a comment

Choose a reason for hiding this comment

github-actions bot commented May 23, 2024

angelz913 commented May 23, 2024 •

edited by kuhar

Loading

github-actions bot commented May 23, 2024 •

edited

Loading