COMPMID-1380: Pre-work for SVE support.
This patch makes the needed infrastructure changes to allow SVE
kernels to be added later on.
Change-Id: Ide5bccac2f47278e93fff3d648231aee2d5f8c2e
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/139070
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>
diff --git a/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp b/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp
index 6e07064..16fa31e 100644
--- a/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp
+++ b/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp
@@ -30,12 +30,12 @@
// Generic unblocked transposed 6x32-bit sized specialisation
template <>
template <typename T>
-inline void TransformImpl<6, 1, true, 4, 4>::Transform(
+inline void TransformImpl<6, 1, true, 4, 4, false>::Transform(
T* out, const T* const in, const int stride,
const int x0, const int xmax, const int k0, const int kmax
) {
// Redirect to a 12 x uint16_t specialisation
- TransformImpl<12, 1, true, 2, 2>::Transform(
+ TransformImpl<12, 1, true, 2, 2, false>::Transform(
reinterpret_cast<uint16_t *>(out),
reinterpret_cast<const uint16_t * const>(in),
stride*2, x0*2, xmax*2, k0, kmax
@@ -45,7 +45,7 @@
// Generic 12x16-bit sized specialisation
template <>
template <typename T>
-inline void TransformImpl<12, 1, true, 2, 2>::Transform(
+inline void TransformImpl<12, 1, true, 2, 2, false>::Transform(
T* out, const T* const in, const int stride,
const int x0, const int xmax, const int k0, const int kmax
) {
@@ -135,7 +135,7 @@
template <>
template <>
-inline void TransformImpl<12, 1, true, 2, 2>::Transform(
+inline void TransformImpl<12, 1, true, 2, 2, false>::Transform(
uint16_t* out, const uint16_t* const in, const int stride,
const int x0, const int xmax, const int k0, const int kmax
) {