From b0c5037d94ba7073ccabb0ebaff54db320f184c4 Mon Sep 17 00:00:00 2001 From: Gian Marco Iodice Date: Fri, 15 Mar 2019 10:13:05 +0000 Subject: COMPMID-2043: Add support for "dummy threads" in CLGEMMReshaped Change-Id: I89403b97503fbb99f6a32f5d62b8c535ab26a7be Signed-off-by: Gian Marco Iodice Reviewed-on: https://review.mlplatform.org/c/877 Reviewed-by: Michele Di Giorgio Tested-by: Arm Jenkins Reviewed-by: Georgios Pinitas Comments-Addressed: Arm Jenkins --- src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp') diff --git a/src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp b/src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp index e9be1a6dfc..a8c1704d91 100644 --- a/src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp +++ b/src/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedKernel.cpp @@ -165,7 +165,7 @@ std::pair validate_and_configure_window(ITensorInfo *input0, ITe } // namespace CLGEMMLowpMatrixMultiplyReshapedKernel::CLGEMMLowpMatrixMultiplyReshapedKernel() - : _input0(nullptr), _input1(nullptr), _output(nullptr), _slide_matrix_b(true), _reinterpret_output_as_3d(false), _k(1) + : _input0(nullptr), _input1(nullptr), _output(nullptr), _slide_matrix_b(true), _reinterpret_output_as_3d(false), _k(1), _use_dummy_work_items(false) { } @@ -181,6 +181,7 @@ void CLGEMMLowpMatrixMultiplyReshapedKernel::configure(const ICLTensor *input0, _output = output; _reinterpret_output_as_3d = (gemm_info.depth_output_gemm3d() != 0); _k = gemm_info.k(); + _use_dummy_work_items = preferred_dummy_work_items_support(CLKernelLibrary::get().get_device()); // Check if we need to slide the matrix B const unsigned int num_dimensions_input0 = _input0->info()->num_dimensions(); @@ -201,6 +202,9 @@ void CLGEMMLowpMatrixMultiplyReshapedKernel::configure(const ICLTensor *input0, build_opts.add_option_if(!_slide_matrix_b, "-DMATRIX_B_DEPTH=" + support::cpp11::to_string(input1->info()->dimension(2))); build_opts.add_option_if(lhs_info.interleave, "-DLHS_INTERLEAVE"); build_opts.add_option_if(rhs_info.interleave, "-DRHS_INTERLEAVE"); + build_opts.add_option_if(_use_dummy_work_items, "-DDUMMY_WORK_ITEMS"); + build_opts.add_option("-DM=" + support::cpp11::to_string(gemm_info.m())); + build_opts.add_option("-DN=" + support::cpp11::to_string(gemm_info.n())); build_opts.add_option("-DM0=" + support::cpp11::to_string(lhs_info.m0)); build_opts.add_option("-DN0=" + support::cpp11::to_string(rhs_info.n0)); build_opts.add_option("-DK0=" + support::cpp11::to_string(lhs_info.k0)); @@ -302,7 +306,7 @@ void CLGEMMLowpMatrixMultiplyReshapedKernel::run(const Window &window, cl::Comma _kernel.setArg(idx++, static_cast(_input0->info()->strides_in_bytes()[2])); _kernel.setArg(idx++, static_cast(_input1->info()->strides_in_bytes()[2])); _kernel.setArg(idx++, static_cast(_output->info()->strides_in_bytes()[2])); - enqueue(queue, *this, slice, lws_hint()); + enqueue(queue, *this, slice, lws_hint(), _use_dummy_work_items); } while(window.slide_window_slice_3D(slice)); } \ No newline at end of file -- cgit v1.2.1