diff options
author | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2018-06-28 16:29:29 +0100 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:54:54 +0000 |
commit | 215b4ea6c9dee480a22070d5873b0b8cb52531a0 (patch) | |
tree | 398e552c4d01c0b84d03a873098a9183ba8f82e4 /src/runtime/CL/functions/CLGEMM.cpp | |
parent | ad486e21e5870f41774f30825c270762e08ae71e (diff) | |
download | ComputeLibrary-215b4ea6c9dee480a22070d5873b0b8cb52531a0.tar.gz |
COMPMID-1277 - Optimizing CLIm2ColKernel for NHWC.
This patch includes:
- Im2Col optimizations for NHWC using a new data layout
- Refactoring of CLIm2ColKernel adding validation method and auto-init
- Removed im2col_reduced from CLIm2ColKernel and created a new kernel CLFlattenLayerKernel
Change-Id: I1620640b6796baa268324b33ae92cdd8de53e27c
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/141241
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Diffstat (limited to 'src/runtime/CL/functions/CLGEMM.cpp')
-rw-r--r-- | src/runtime/CL/functions/CLGEMM.cpp | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/src/runtime/CL/functions/CLGEMM.cpp b/src/runtime/CL/functions/CLGEMM.cpp index 1d1b17bbf1..a8d7058f2a 100644 --- a/src/runtime/CL/functions/CLGEMM.cpp +++ b/src/runtime/CL/functions/CLGEMM.cpp @@ -171,6 +171,7 @@ void CLGEMM::configure(const ICLTensor *a, const ICLTensor *b, const ICLTensor * Status CLGEMM::validate(const ITensorInfo *a, const ITensorInfo *b, const ITensorInfo *c, const ITensorInfo *output, float alpha, float beta, const GEMMInfo &gemm_info) { ARM_COMPUTE_UNUSED(alpha); + ARM_COMPUTE_UNUSED(output); // Check if we need to reshape the matrix B only on the first run const bool reshape_b_only_on_first_run = gemm_info.reshape_b_only_on_first_run(); @@ -180,7 +181,7 @@ Status CLGEMM::validate(const ITensorInfo *a, const ITensorInfo *b, const ITenso TensorInfo tmp_a_info{}; TensorInfo tmp_b_info{}; - TensorInfo tmp_output_info = *output->clone(); + TensorInfo tmp_output_info{}; // Get the GPU target const GPUTarget gpu_target = CLScheduler::get().target(); |