diff options
Diffstat (limited to 'src/core/CL/kernels')
-rw-r--r-- | src/core/CL/kernels/CLGEMMMatrixAdditionKernel.cpp | 9 | ||||
-rw-r--r-- | src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp | 15 |
2 files changed, 14 insertions, 10 deletions
diff --git a/src/core/CL/kernels/CLGEMMMatrixAdditionKernel.cpp b/src/core/CL/kernels/CLGEMMMatrixAdditionKernel.cpp index 343838f2f9..5883dd698b 100644 --- a/src/core/CL/kernels/CLGEMMMatrixAdditionKernel.cpp +++ b/src/core/CL/kernels/CLGEMMMatrixAdditionKernel.cpp @@ -28,6 +28,7 @@ #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/CL/OpenCL.h" #include "arm_compute/core/Error.h" +#include "arm_compute/core/FixedPoint.h" #include "arm_compute/core/Helpers.h" #include "arm_compute/core/Types.h" #include "arm_compute/core/Validate.h" @@ -40,10 +41,9 @@ CLGEMMMatrixAdditionKernel::CLGEMMMatrixAdditionKernel() { } -void CLGEMMMatrixAdditionKernel::configure(const ICLTensor *input, ICLTensor *output, const float beta) +void CLGEMMMatrixAdditionKernel::configure(const ICLTensor *input, ICLTensor *output, float beta) { - ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::F16, DataType::F32); - ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::F16, DataType::F32); + ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QS8, DataType::F16, DataType::F32); ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input, output); ARM_COMPUTE_ERROR_ON(input->info()->dimension(0) != output->info()->dimension(0)); ARM_COMPUTE_ERROR_ON(input->info()->dimension(1) != output->info()->dimension(1)); @@ -53,7 +53,8 @@ void CLGEMMMatrixAdditionKernel::configure(const ICLTensor *input, ICLTensor *ou const unsigned int num_elems_processed_per_iteration = max_cl_vector_width / data_size_from_type(input->info()->data_type()); std::ostringstream ma_arguments; - ma_arguments << "-DBETA=" << beta; + ma_arguments << "-DBETA=" << (input->info()->data_type() == DataType::QS8 ? scvt_qs8_f32(beta, input->info()->fixed_point_position()) : beta) << " "; + ma_arguments << "-DFIXED_POINT_POSITION=" << input->info()->fixed_point_position(); std::set<std::string> build_opts; build_opts.emplace(ma_arguments.str()); diff --git a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp index d7388e8579..7c5b3d7866 100644 --- a/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp +++ b/src/core/CL/kernels/CLGEMMMatrixMultiplyKernel.cpp @@ -25,12 +25,12 @@ #include "arm_compute/core/AccessWindowStatic.h" #include "arm_compute/core/AccessWindowTranspose.h" - #include "arm_compute/core/CL/CLHelpers.h" #include "arm_compute/core/CL/CLKernelLibrary.h" #include "arm_compute/core/CL/ICLTensor.h" #include "arm_compute/core/CL/OpenCL.h" #include "arm_compute/core/Error.h" +#include "arm_compute/core/FixedPoint.h" #include "arm_compute/core/Helpers.h" #include "arm_compute/core/Types.h" #include "arm_compute/core/Utils.h" @@ -50,10 +50,10 @@ CLGEMMMatrixMultiplyKernel::CLGEMMMatrixMultiplyKernel() void CLGEMMMatrixMultiplyKernel::configure(const ICLTensor *input0, const ICLTensor *input1, ICLTensor *output, float alpha) { - ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input0, 1, DataType::F16, DataType::F32); - ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input1, 1, DataType::F16, DataType::F32); - ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::F16, DataType::F32); + ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input0, 1, DataType::QS8, DataType::F16, DataType::F32); ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(input0, input1, output); + ARM_COMPUTE_ERROR_ON_MISMATCHING_FIXED_POINT(input0, input1, output); + if(output->info()->dimension(1) == 1) { ARM_COMPUTE_ERROR_ON(input0->info()->dimension(0) != input1->info()->dimension(1)); @@ -74,7 +74,8 @@ void CLGEMMMatrixMultiplyKernel::configure(const ICLTensor *input0, const ICLTen std::ostringstream mm_arguments; mm_arguments << "-DWIDTH_MATRIX_B=" << input1->info()->dimension(0) << " "; - mm_arguments << "-DALPHA=" << alpha << " "; + mm_arguments << "-DALPHA=" << (input0->info()->data_type() == DataType::QS8 ? scvt_qs8_f32(alpha, input0->info()->fixed_point_position()) : alpha) << " "; + mm_arguments << "-DFIXED_POINT_POSITION=" << input0->info()->fixed_point_position() << " "; std::set<std::string> build_opts; // Check if the output tensor is a vector. If so,the kernel runs the vector-matrix multiplication @@ -98,7 +99,9 @@ void CLGEMMMatrixMultiplyKernel::configure(const ICLTensor *input0, const ICLTen update_window_and_padding(win, input0_access, input1_access, output_access); - output_access.set_valid_region(win, ValidRegion(Coordinates(0, 0), output->info()->tensor_shape())); + Coordinates coord; + coord.set_num_dimensions(output->info()->num_dimensions()); + output_access.set_valid_region(win, ValidRegion(coord, output->info()->tensor_shape())); ICLKernel::configure(win); } |