diff options
author | Pablo Tello <pablo.tello@arm.com> | 2018-03-14 17:55:27 +0000 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:49:16 +0000 |
commit | 7fad9b1d00f3ee1488ba4038d1371f6ea219f8b7 (patch) | |
tree | ded71e1cfa8e0c085f8bce5dfc26a99786d60e52 /src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp | |
parent | 1562be3e8a449360a90af75f6f1481a30d41be75 (diff) | |
download | ComputeLibrary-7fad9b1d00f3ee1488ba4038d1371f6ea219f8b7.tar.gz |
COMPMID-1021: CPUInfo refactoring.
Removed CPUTarget in favor of the CPUModel type.
CPUInfo now holds a vector of N CPUs.
CPUInfo autoinitialise upon construction with 1 GENERIC CPU.
CPPScheduler fills CPUInfo's vector upon construction (runtime).
IScheduler has a single CPUInfo obj and ThreadInfo always gets a pointer to it (avoid copying the vector)
Change-Id: I30f293258c959c87f6bac5eac8b963beb6a4d365
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/124626
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Diffstat (limited to 'src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp')
-rw-r--r-- | src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp | 18 |
1 files changed, 6 insertions, 12 deletions
diff --git a/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp b/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp index b2dd0227a5..cdbd32373a 100644 --- a/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp +++ b/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp @@ -271,8 +271,7 @@ void NEGEMMConvolutionLayer::configure(const ITensor *input, const ITensor *weig const unsigned int fixed_point_position = input->info()->fixed_point_position(); const ITensor *biases_to_use = (_append_bias) ? biases : nullptr; - bool run_optimised = - (NEScheduler::get().cpu_info().CPU == CPUTarget::ARMV7 && dt == DataType::F32) || (NEScheduler::get().cpu_info().CPU >= CPUTarget::ARMV8 && dt == DataType::F32); + bool run_optimised = dt == DataType::F32; // Reshape weights if needed if(run_optimised) @@ -369,8 +368,10 @@ void NEGEMMConvolutionLayer::configure(const ITensor *input, const ITensor *weig // Configure matrix multiply if(run_optimised) { - run_optimised = setup_assembly_kernel(&_input_im2col_reshaped, weights, nullptr, &_gemm_output, 1.f, 0.f, _workspace, _memory_group, _asm_glue); - ARM_COMPUTE_ERROR_ON_MSG(run_optimised == false, "setup_assembly_kernel failed."); + if(!setup_assembly_kernel(&_input_im2col_reshaped, weights, &_gemm_output, 1.f, 0.f, _workspace, _memory_group, _asm_glue)) + { + ARM_COMPUTE_ERROR("setup_assembly_kernel failed."); + } } else { @@ -450,17 +451,10 @@ Status NEGEMMConvolutionLayer::validate(const ITensorInfo *input, const ITensorI std::unique_ptr<ITensorInfo> reshaped_weights = weights->clone(); bool optimised_kernel = false; -#if defined(__arm__) - if(NEScheduler::get().cpu_info().CPU == CPUTarget::ARMV7 && dt == DataType::F32) - { - optimised_kernel = true; - } -#elif defined(__aarch64__) - if(NEScheduler::get().cpu_info().CPU >= CPUTarget::ARMV8 && dt == DataType::F32) + if(dt == DataType::F32) { optimised_kernel = true; } -#endif /* defined(__arm__) || defined(__aarch64__) */ // Reshape weights if needed if(optimised_kernel) |