aboutsummaryrefslogtreecommitdiff
path: root/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
diff options
context:
space:
mode:
authorGeorgios Pinitas <georgios.pinitas@arm.com>2018-06-04 19:27:13 +0100
committerAnthony Barbier <anthony.barbier@arm.com>2018-11-02 16:52:54 +0000
commit17812ba9f7cf2c8f5121c11760ac45fbbdb7aeaf (patch)
tree28c7bb65a8306e82de91a644fdcc1c0947c6f6d7 /src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
parentf8d8f3aff04faf731f20411ecb91027eab4365c5 (diff)
downloadComputeLibrary-17812ba9f7cf2c8f5121c11760ac45fbbdb7aeaf.tar.gz
COMPMID-817: Tuner: Port kernels to new design.
Change-Id: Iaabb1153c2abe0400ec79d51a21347debe92d642 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/134062 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Diffstat (limited to 'src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp')
-rw-r--r--src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp48
1 files changed, 3 insertions, 45 deletions
diff --git a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
index 7a9760b778..fc52f4e124 100644
--- a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
+++ b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp
@@ -194,51 +194,9 @@ void CLGEMMMatrixMultiplyKernel::configure(const ICLTensor *input0, const ICLTen
_output = output;
_slide_matrix_b = _input1->info()->num_dimensions() >= _input0->info()->num_dimensions();
- const DataType data_type = input0->info()->data_type();
- const int fp_pos = input0->info()->fixed_point_position();
-
- // Get target architecture
- GPUTarget gpu_target = get_target();
-
- // Configure LWS hint
- switch(gpu_target)
- {
- case GPUTarget::MIDGARD:
- case GPUTarget::T600:
- case GPUTarget::T700:
- case GPUTarget::T800:
- if(output->info()->dimension(1) == 196)
- {
- _lws_hint = cl::NDRange(1, 7);
- }
- else
- {
- _lws_hint = cl::NDRange(8, 8);
- }
- break;
- case GPUTarget::G71:
- case GPUTarget::G72:
- case GPUTarget::G51:
- case GPUTarget::G51BIG:
- case GPUTarget::G51LIT:
- case GPUTarget::TNOX:
- if(input1->info()->dimension(1) == 24)
- {
- // LWS optimized for the 11x11 AlexNet convolution on Bifrost.
- _lws_hint = cl::NDRange(2, 2);
- }
- else if(output->info()->dimension(1) == 196)
- {
- _lws_hint = cl::NDRange(1, 7);
- }
- else
- {
- _lws_hint = cl::NDRange(8, 8);
- }
- break;
- default:
- _lws_hint = cl::NullRange;
- }
+ const DataType data_type = input0->info()->data_type();
+ const int fp_pos = input0->info()->fixed_point_position();
+ const GPUTarget gpu_target = get_target();
ElementsProcessed num_elements_processed{};