diff options
author | Michalis Spyrou <michalis.spyrou@arm.com> | 2018-02-22 18:07:43 +0000 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:48:35 +0000 |
commit | a9676118fd2a0e5bc916969af83ecee049bae76b (patch) | |
tree | f67ff64d962a2f802700f6b26a9ab160c04c721d /src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp | |
parent | 2bc74410251dcbaf17a7c5447317aa6d0171972a (diff) | |
download | ComputeLibrary-a9676118fd2a0e5bc916969af83ecee049bae76b.tar.gz |
COMPMID-886 Don't use LWS hints by default for GPU post Mali-G72
Change-Id: I64cb2d7f9513d69aebd9307a803b1b2c9c0e04c3
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/121929
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp')
-rw-r--r-- | src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp b/src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp index c01a6660a7..56ac0c7250 100644 --- a/src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp +++ b/src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp @@ -134,7 +134,7 @@ std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input, ITen unsigned int num_elems_written_per_iteration_x = 0; unsigned int num_elems_written_per_iteration_y = 0; - if((target == GPUTarget::BIFROST) && (kernel_size <= 5) && (conv_stride_x == 1) && (conv_stride_y == 1) && (data_type == DataType::F32)) + if(gpu_target_is_in(target, GPUTarget::G71, GPUTarget::G72) && (kernel_size <= 5) && (conv_stride_x == 1) && (conv_stride_y == 1) && (data_type == DataType::F32)) { // Configure kernel window @@ -301,7 +301,7 @@ void CLDirectConvolutionLayerKernel::configure(const ICLTensor *input, const ICL _output = output; _biases = biases; - const GPUTarget gpu_target = get_arch_from_target(get_target()); + const GPUTarget gpu_target = get_target(); std::stringstream kernel_name; kernel_name << "direct_convolution" << kernel_size << "x" << kernel_size; @@ -309,7 +309,7 @@ void CLDirectConvolutionLayerKernel::configure(const ICLTensor *input, const ICL CLBuildOptions build_options; build_options.add_option_if(_biases != nullptr, std::string("-DHAS_BIAS")); - if((gpu_target == GPUTarget::BIFROST) && (kernel_size <= 5) && (_conv_stride_x == 1) && (_conv_stride_y == 1) && (data_type == DataType::F32)) + if(gpu_target_is_in(gpu_target, GPUTarget::G71, GPUTarget::G72) && (kernel_size <= 5) && (_conv_stride_x == 1) && (_conv_stride_y == 1) && (data_type == DataType::F32)) { build_options.add_option(std::string("-DWEIGHTS_DEPTH=" + support::cpp11::to_string(_weights->info()->dimension(2)))); |