COMPMID-1687: Optimize CLGEMMMatrixMultiplyKernel
Change-Id: I040478ff7aa04f0523ed6e302129b829442cb194
Reviewed-on: https://review.mlplatform.org/534
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
diff --git a/src/runtime/CL/functions/CLGEMM.cpp b/src/runtime/CL/functions/CLGEMM.cpp
index a3612f3..cd40fc6 100644
--- a/src/runtime/CL/functions/CLGEMM.cpp
+++ b/src/runtime/CL/functions/CLGEMM.cpp
@@ -33,11 +33,13 @@
#include "arm_compute/core/Validate.h"
#include "arm_compute/core/utils/misc/ShapeCalculator.h"
#include "arm_compute/runtime/CL/CLScheduler.h"
+#include "arm_compute/runtime/CL/gemm_reshaped/CLGEMMReshapedConfiguration.h"
#include "arm_compute/runtime/ITensorAllocator.h"
namespace arm_compute
{
using namespace arm_compute::misc::shape_calculator;
+using namespace arm_compute::cl_gemm;
namespace
{
@@ -77,43 +79,6 @@
return flag;
}
-
-inline void select_gemm_configuration(unsigned int m, unsigned int n, GEMMLHSMatrixInfo &lhs_info, GEMMRHSMatrixInfo &rhs_info)
-{
- // Heuristic selection for GEMM
- if(n <= 4)
- {
- // Configure GEMMLHSMatrixInfo
- lhs_info.m0 = 4;
- lhs_info.k0 = 8;
- lhs_info.v0 = lhs_info.m0 * 16 < m ? 2 : 16;
- lhs_info.interleave = true;
- lhs_info.transpose = false;
-
- // Configure GEMMRHSMatrixInfo
- rhs_info.n0 = 2;
- rhs_info.k0 = lhs_info.k0;
- rhs_info.h0 = rhs_info.n0 * 16 < n ? 2 : 16;
- rhs_info.interleave = false;
- rhs_info.transpose = true;
- }
- else
- {
- // Configure GEMMLHSMatrixInfo
- lhs_info.m0 = (m * n) / 24 > 2048 ? 6 : 5;
- lhs_info.k0 = 4;
- lhs_info.v0 = 32;
- lhs_info.interleave = false;
- lhs_info.transpose = false;
-
- // Configure GEMMRHSMatrixInfo
- rhs_info.n0 = 4;
- rhs_info.k0 = lhs_info.k0;
- rhs_info.h0 = 32;
- rhs_info.interleave = true;
- rhs_info.transpose = true;
- }
-}
} // namespace
CLGEMM::CLGEMM(std::shared_ptr<IMemoryManager> memory_manager)
@@ -130,7 +95,7 @@
_run_addition(false),
_reshape_b_only_on_first_run(false),
_is_prepared(false),
- _is_G76_path(false)
+ _is_new_gemm_reshaped(false)
{
}
@@ -164,6 +129,7 @@
const unsigned int m = reinterpret_input_as_3d ? (a->info()->dimension(1) * a->info()->dimension(2)) : a->info()->dimension(1);
const unsigned int n = b->info()->dimension(0);
const unsigned int k = a->info()->dimension(0);
+ const unsigned int batch_size = reinterpret_input_as_3d ? a->info()->dimension(3) : a->info()->dimension(2);
const int depth_output_gemm3d = gemm_info.depth_output_gemm3d();
int mult_transpose1xW_width = 1;
int mult_interleave4x4_height = 1;
@@ -191,7 +157,8 @@
_is_interleaved_transposed = is_interleaved_transposed(m, n, k, a->info()->data_type(), _reshape_b_only_on_first_run, gpu_target);
// Check if we can run the new reshaped GEMM
- _is_G76_path = (gpu_target == GPUTarget::G76) && _is_interleaved_transposed && (data_type == DataType::F32);
+ const auto workload = static_cast<float>((m * n) / 20.0f);
+ _is_new_gemm_reshaped = (workload > 1600.0f) && (get_arch_from_target(gpu_target) == GPUTarget::BIFROST) && _is_interleaved_transposed && (data_type == DataType::F32);
// if _is_interleaved_transposed is set, force reinterpret_input_as_3d to be false as the output of CLGEMMInterleaveKernel will be 2D
if(_is_interleaved_transposed)
@@ -209,12 +176,12 @@
}
// _tmp_a and _tmp_b will be auto configured in _interleave_kernel and in _transpose_kernel
- if(_is_G76_path)
+ if(_is_new_gemm_reshaped)
{
GEMMLHSMatrixInfo lhs_info;
- // Pick up the GEMM configuration based on M,N and K
- select_gemm_configuration(m, n, lhs_info, rhs_info);
+ // Pick up the GEMM configuration
+ std::tie(lhs_info, rhs_info) = CLGEMMReshapedConfigurationFactory::create()->configure(m, n, k, batch_size, data_type);
_reshape_lhs_kernel.configure(a, &_tmp_a, lhs_info, gemm_info.reinterpret_input_as_3d());
_reshape_rhs_kernel.configure(b, &_tmp_b, rhs_info);
@@ -232,7 +199,7 @@
}
}
- if(!_is_G76_path)
+ if(!_is_new_gemm_reshaped)
{
// Configure and tune matrix multiply kernel
_mm_kernel.configure(matrix_a, matrix_b, output, alpha, _is_interleaved_transposed, GEMMReshapeInfo(m, n, k,
@@ -285,6 +252,7 @@
const unsigned int m = reinterpret_input_as_3d ? (a->dimension(1) * a->dimension(2)) : a->dimension(1);
const unsigned int n = b->dimension(0);
const unsigned int k = a->dimension(0);
+ const unsigned int batch_size = reinterpret_input_as_3d ? a->dimension(3) : a->dimension(2);
int mult_transpose1xW_width = 1;
int mult_interleave4x4_height = 1;
const int depth_output_gemm3d = gemm_info.depth_output_gemm3d();
@@ -313,7 +281,8 @@
const bool run_interleave_transpose = is_interleaved_transposed(m, n, k, a->data_type(), reshape_b_only_on_first_run, gpu_target);
// Check if we can run the new reshaped GEMM
- const bool is_G76_path = (gpu_target == GPUTarget::G76) && run_interleave_transpose && (data_type == DataType::F32);
+ const auto workload = static_cast<float>((m * n) / 20.0f);
+ const bool is_new_gemm_reshaped = (workload > 1600.f) && (get_arch_from_target(gpu_target) == GPUTarget::BIFROST) && run_interleave_transpose && (data_type == DataType::F32);
// if _is_interleaved_transposed is set, force reinterpret_input_as_3d to be false as the output of CLGEMMInterleaveKernel will be 2D
if(run_interleave_transpose)
@@ -328,12 +297,12 @@
matrix_a_info = &tmp_a_info;
matrix_b_info = &tmp_b_info;
- if(is_G76_path)
+ if(is_new_gemm_reshaped)
{
GEMMLHSMatrixInfo lhs_info;
- // Pick up the GEMM configuration based on M,N and K
- select_gemm_configuration(m, n, lhs_info, rhs_info);
+ // Pick up the GEMM configuration
+ std::tie(lhs_info, rhs_info) = CLGEMMReshapedConfigurationFactory::create()->configure(m, n, k, batch_size, data_type);
auto_init_if_empty(tmp_a_info, a->clone()->set_tensor_shape(compute_lhs_reshaped_shape(*a, lhs_info, gemm_info.reinterpret_input_as_3d())));
ARM_COMPUTE_RETURN_ON_ERROR(CLGEMMReshapeLHSMatrixKernel::validate(a, &tmp_a_info, lhs_info, gemm_info.reinterpret_input_as_3d()));
@@ -356,7 +325,7 @@
}
}
- if(!is_G76_path)
+ if(!is_new_gemm_reshaped)
{
// Validate matrix multiply
ARM_COMPUTE_RETURN_ON_ERROR(CLGEMMMatrixMultiplyKernel::validate(matrix_a_info, matrix_b_info, output, alpha, run_interleave_transpose, reshape_info, gpu_target, gemm_info.fp_mixed_precision()));
@@ -390,7 +359,7 @@
}
// Run matrix multiply kernel
- if(_is_G76_path)
+ if(_is_new_gemm_reshaped)
{
CLScheduler::get().enqueue(_mm_reshaped_kernel, !_run_addition);
}