Fix flaky gpu compiler test when building with rocm

Author: alekstheod

xla/backends/gpu/codegen/triton/compilation_pipeline_rocm.cc

@@ -90,7 +90,7 @@ absl::Status CreateTritonPipeline(mlir::OpPassManager* pm,
   pm->addPass(mt::gpu::createTritonGPUCoalesce());
   pm->addPass(mt::gpu::createTritonGPURemoveLayoutConversions());
   pm->addPass(mt::gpu::createTritonGPUOptimizeThreadLocality());
-  pm->addPass(mt::gpu::createTritonGPUAccelerateMatmul());
+  pm->addPass(mlir::createTritonAMDGPUAccelerateMatmulPass());
   pm->addPass(mt::gpu::createTritonGPURemoveLayoutConversions());
   // TODO ROCm Check if we want to compare MI100 and greater
   pm->addPass(mlir::createTritonAMDGPUOptimizeEpiloguePass());