COMPMID-3322: Add cl_image support for GEMMReshapedOnlyRHS NT
COMPMID-3323: Add cl_image support for GEMMReshapedOnlyRHS T
- Added support for cl_image in CLGEMMMatrixMultiplyReshapedInlyRHSKernel (both NT and T kernels)
- Extended the tests for the validating rhs_info.export_to_cl_image = true
- Updated doxygen documentation in CLGEMMMatrixMultiplyReshapedOnlyRHSKernel.h
Change-Id: If253794323aac072d84a4d8680b9a2339ab7ad92
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3437
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
diff --git a/tests/validation/fixtures/GEMMFixture.h b/tests/validation/fixtures/GEMMFixture.h
index b2adf2d..e91becf 100644
--- a/tests/validation/fixtures/GEMMFixture.h
+++ b/tests/validation/fixtures/GEMMFixture.h
@@ -986,18 +986,19 @@
public:
template <typename...>
void setup(unsigned int m, unsigned int n, unsigned int k, unsigned int batch_size, unsigned int m0, unsigned int n0, unsigned int k0, unsigned int h0,
- bool interleave_rhs, bool transpose_rhs, DataType data_type, float alpha, float beta, bool broadcast_bias, const ActivationLayerInfo &act_info)
+ bool interleave_rhs, bool transpose_rhs, bool export_to_cl_image, DataType data_type, float alpha, float beta, bool broadcast_bias, const ActivationLayerInfo &act_info)
{
GEMMLHSMatrixInfo lhs_info;
lhs_info.m0 = m0;
lhs_info.k0 = k0;
GEMMRHSMatrixInfo rhs_info;
- rhs_info.n0 = n0;
- rhs_info.k0 = k0;
- rhs_info.h0 = h0;
- rhs_info.interleave = interleave_rhs;
- rhs_info.transpose = transpose_rhs;
+ rhs_info.n0 = n0;
+ rhs_info.k0 = k0;
+ rhs_info.h0 = h0;
+ rhs_info.interleave = interleave_rhs;
+ rhs_info.transpose = transpose_rhs;
+ rhs_info.export_to_cl_image = export_to_cl_image;
// Set the tensor shapes for LHS and RHS matrices
const TensorShape lhs_shape(k, m, batch_size);
@@ -1124,18 +1125,19 @@
public:
template <typename...>
void setup(unsigned int m_w, unsigned int m_h, unsigned int n, unsigned int k, unsigned int batch_size, unsigned int m0, unsigned int n0, unsigned int k0, unsigned int h0,
- bool interleave_rhs, bool transpose_rhs, DataType data_type, float alpha, float beta, const ActivationLayerInfo &act_info)
+ bool interleave_rhs, bool transpose_rhs, bool export_to_cl_image, DataType data_type, float alpha, float beta, const ActivationLayerInfo &act_info)
{
GEMMLHSMatrixInfo lhs_info;
lhs_info.m0 = m0;
lhs_info.k0 = k0;
GEMMRHSMatrixInfo rhs_info;
- rhs_info.n0 = n0;
- rhs_info.k0 = k0;
- rhs_info.h0 = h0;
- rhs_info.interleave = interleave_rhs;
- rhs_info.transpose = transpose_rhs;
+ rhs_info.n0 = n0;
+ rhs_info.k0 = k0;
+ rhs_info.h0 = h0;
+ rhs_info.interleave = interleave_rhs;
+ rhs_info.transpose = transpose_rhs;
+ rhs_info.export_to_cl_image = export_to_cl_image;
// In case of GEMM3D, m is the product between m_w and m_h
const unsigned int m = m_w * m_h;