diff --git a/mlir/include/mlir/Dialect/AMDGPU/Transforms/Passes.h b/mlir/include/mlir/Dialect/AMDGPU/Transforms/Passes.h
index 94dd9e3a2933..a52ee2ee89ca 100644
--- a/mlir/include/mlir/Dialect/AMDGPU/Transforms/Passes.h
+++ b/mlir/include/mlir/Dialect/AMDGPU/Transforms/Passes.h
@@ -14,6 +14,7 @@
 #define MLIR_DIALECT_AMDGPU_TRANSFORMS_PASSES_H_
 
 #include "mlir/Dialect/AMDGPU/Utils/Chipset.h"
+#include "mlir/IR/PatternMatch.h"
 #include "mlir/Pass/Pass.h"
 
 namespace mlir {
@@ -28,11 +29,14 @@ namespace amdgpu {
 
 void populateAmdgpuEmulateAtomicsPatterns(ConversionTarget &target,
                                           RewritePatternSet &patterns,
-                                          Chipset chipset);
+                                          Chipset chipset,
+                                          PatternBenefit benefit = 1);
 
-void populateAmdgpuResolveStridedMetadataPatterns(RewritePatternSet &patterns);
+void populateAmdgpuResolveStridedMetadataPatterns(RewritePatternSet &patterns,
+                                                  PatternBenefit benefit = 1);
 
-void populateAmdgpuTransferReadToLoadPatterns(RewritePatternSet &patterns);
+void populateAmdgpuTransferReadToLoadPatterns(RewritePatternSet &patterns,
+                                              PatternBenefit benefit = 1);
 
 } // namespace amdgpu
 } // namespace mlir
diff --git a/mlir/lib/Dialect/AMDGPU/Transforms/EmulateAtomics.cpp b/mlir/lib/Dialect/AMDGPU/Transforms/EmulateAtomics.cpp
index 7dd4be66d2bd..fd2ba0683786 100644
--- a/mlir/lib/Dialect/AMDGPU/Transforms/EmulateAtomics.cpp
+++ b/mlir/lib/Dialect/AMDGPU/Transforms/EmulateAtomics.cpp
@@ -164,7 +164,8 @@ LogicalResult RawBufferAtomicByCasPattern<AtomicOp, ArithOp>::matchAndRewrite(
 }
 
 void mlir::amdgpu::populateAmdgpuEmulateAtomicsPatterns(
-    ConversionTarget &target, RewritePatternSet &patterns, Chipset chipset) {
+    ConversionTarget &target, RewritePatternSet &patterns, Chipset chipset,
+    PatternBenefit benefit) {
   // gfx10 has no atomic adds.
   if (chipset.majorVersion == 10 || chipset < Chipset(9, 0, 8)) {
     target.addIllegalOp<RawBufferAtomicFaddOp>();
@@ -204,7 +205,7 @@ void mlir::amdgpu::populateAmdgpuEmulateAtomicsPatterns(
       RawBufferAtomicByCasPattern<RawBufferAtomicFmaxOp, arith::MaximumFOp>,
       RawBufferAtomicByCasPattern<RawBufferAtomicSmaxOp, arith::MaxSIOp>,
       RawBufferAtomicByCasPattern<RawBufferAtomicUminOp, arith::MinUIOp>>(
-      patterns.getContext());
+      patterns.getContext(), benefit);
 }
 
 void AmdgpuEmulateAtomicsPass::runOnOperation() {
diff --git a/mlir/lib/Dialect/AMDGPU/Transforms/ResolveStridedMetadata.cpp b/mlir/lib/Dialect/AMDGPU/Transforms/ResolveStridedMetadata.cpp
index 4b3d94b4ce2a..195f59d62555 100644
--- a/mlir/lib/Dialect/AMDGPU/Transforms/ResolveStridedMetadata.cpp
+++ b/mlir/lib/Dialect/AMDGPU/Transforms/ResolveStridedMetadata.cpp
@@ -66,9 +66,9 @@ struct ExtractStridedMetadataOnFatRawBufferCastFolder final
 } // namespace
 
 void mlir::amdgpu::populateAmdgpuResolveStridedMetadataPatterns(
-    RewritePatternSet &patterns) {
+    RewritePatternSet &patterns, PatternBenefit benefit) {
   patterns.add<ExtractStridedMetadataOnFatRawBufferCastFolder>(
-      patterns.getContext());
+      patterns.getContext(), benefit);
 }
 
 void AmdgpuResolveStridedMetadataPass::runOnOperation() {
diff --git a/mlir/lib/Dialect/AMDGPU/Transforms/TransferReadToLoad.cpp b/mlir/lib/Dialect/AMDGPU/Transforms/TransferReadToLoad.cpp
index 96925dbf9f28..f5b12a9524cc 100644
--- a/mlir/lib/Dialect/AMDGPU/Transforms/TransferReadToLoad.cpp
+++ b/mlir/lib/Dialect/AMDGPU/Transforms/TransferReadToLoad.cpp
@@ -222,8 +222,8 @@ struct TransferReadLowering final : OpRewritePattern<vector::TransferReadOp> {
 } // namespace
 
 void mlir::amdgpu::populateAmdgpuTransferReadToLoadPatterns(
-    RewritePatternSet &patterns) {
-  patterns.add<TransferReadLowering>(patterns.getContext());
+    RewritePatternSet &patterns, PatternBenefit benefit) {
+  patterns.add<TransferReadLowering>(patterns.getContext(), benefit);
 }
 
 struct AmdgpuTransferReadToLoadPass final