diff options
| author | Teresa Charlin <teresa.charlinreyes@arm.com> | 2023-01-13 18:44:00 +0000 |
|---|---|---|
| committer | Teresa Charlin <teresa.charlinreyes@arm.com> | 2023-01-13 18:45:01 +0000 |
| commit | c074cdb1b035b0bcd0be9aa25feab97d2613140f (patch) | |
| tree | 88c0cb42d1f21f428ea09510414c6ff092f8f264 | |
| parent | 9fea7d9c0164cd30dee8f49c2835acf9364fa3c3 (diff) | |
| download | armnn-c074cdb1b035b0bcd0be9aa25feab97d2613140f.tar.gz | |
IVGCVSW-6493 Bug Fix on RHS permute GpuAcc Batch MatMul workload Fp32
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I60e9284b90467f58e0acd74d3f1493546b6f1b9b
| -rw-r--r-- | src/backends/cl/workloads/ClBatchMatMulWorkload.cpp | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/src/backends/cl/workloads/ClBatchMatMulWorkload.cpp b/src/backends/cl/workloads/ClBatchMatMulWorkload.cpp index 4acdef5e5c..ece87c2672 100644 --- a/src/backends/cl/workloads/ClBatchMatMulWorkload.cpp +++ b/src/backends/cl/workloads/ClBatchMatMulWorkload.cpp @@ -1,5 +1,5 @@ // -// Copyright © 2022 Arm Ltd and Contributors. All rights reserved. +// Copyright © 2022-2023 Arm Ltd and Contributors. All rights reserved. // SPDX-License-Identifier: MIT // @@ -158,13 +158,13 @@ ClBatchMatMulWorkload::ClBatchMatMulWorkload(const BatchMatMulQueueDescriptor& d if (descriptor.m_Parameters.m_TransposeY == true) { armnn::PermutationVector permutationYVector - = GeneratePermutationVectorOnLastTwoDimensions(info.m_InputTensorInfos[0].GetNumDimensions()); - const TensorInfo permutedYInfo = armnnUtils::Permuted(info.m_InputTensorInfos[0], permutationYVector); + = GeneratePermutationVectorOnLastTwoDimensions(info.m_InputTensorInfos[1].GetNumDimensions()); + const TensorInfo permutedYInfo = armnnUtils::Permuted(info.m_InputTensorInfos[1], permutationYVector); const auto aclPermutationYVector = armcomputetensorutils::BuildArmComputePermutationVector(permutationYVector); armcomputetensorutils::BuildArmComputeTensor(m_PermutedTensorY, permutedYInfo); armcomputetensorutils::InitialiseArmComputeTensorEmpty(m_PermutedTensorY); - std::unique_ptr<arm_compute::CLPermute> permuteLayerY(new arm_compute::CLPermute()); + auto permuteLayerY = std::make_unique<arm_compute::CLPermute>(); permuteLayerY->configure(clCompileContext, &inputY, &m_PermutedTensorY, |