Update CPU kernels and add mixed sign GEMM support

 - Add support for mixed sign quantized convolution.
 - Add support for mixed sign dequantized GEMM.
 - Add SME FP16 GEMV kernel.
 - Change SME vector length function to use RDSVL instead of static variable.
 - Add GEMM dilation support internally (not exposed yet).
 - Remove unused "get_default_activation_values" functions.
 - Add SVE fixed format interleaved BF16 DOT kernel.
 - Updates and optimizations to assembly kernels.

Resolves COMPMID-6926

Change-Id: I227f502502611d4cc4111c89e30c53ce94079544
Signed-off-by: Michael Tyler <michael.tyler@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11570
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 5 insertions, 5 deletions

diff --git a/src/core/NEON/kernels/arm_gemm/std_transforms_sve.hpp b/src/core/NEON/kernels/arm_gemm/std_transforms_sve.hpp
index c516bfc456..32d597f4af 100644
--- a/src/core/NEON/kernels/arm_gemm/std_transforms_sve.hpp
+++ b/src/core/NEON/kernels/arm_gemm/std_transforms_sve.hpp
@@ -39,24 +39,24 @@ namespace arm_gemm {
  * The optional 'block' parameter is for kernels using dot-product type
  * instructions like UDOT and SDOT.
  */
-template<typename TOperand, typename TResult, unsigned int height, unsigned int width_vectors, unsigned int block=1, unsigned int mmla=1, bool integrate_sums=false>
+template<typename TInput, typename TWeight, typename TResult, unsigned int height, unsigned int width_vectors, unsigned int block=1, unsigned int mmla=1, bool integrate_sums=false>
 class StdTransformsSVE
 {
 public:
     template<typename TIn>
-    void PrepareA(TOperand *out, const TIn *in, const int stride, const int y0,
+    void PrepareA(TInput *out, const TIn *in, const int stride, const int y0,
                   const int ymax, const int k0, const int kmax, int32_t row_sum_multiplier) {
         Interleave<height, block, VLType::None>(out, in, stride, y0, ymax, k0, kmax, integrate_sums, row_sum_multiplier);
     }
 
     template<typename TIn>
-    void PrepareA_indirect(TOperand *out, const TIn * const * const *ptr, size_t stringlen, size_t rounded_stringlen, const int y0,
+    void PrepareA_indirect(TInput *out, const TIn * const * const *ptr, size_t stringlen, size_t rounded_stringlen, const int y0,
                            const int ymax, const int k0, const int kmax, int32_t row_sum_multiplier) {
         IndirectInterleave<height, block, VLType::None>(out, ptr, stringlen, rounded_stringlen, y0, ymax, k0, kmax, integrate_sums, row_sum_multiplier);
     }
 
     template<typename TIn>
-    void PrepareA_convolution(TOperand *out, const TIn *ptr, size_t stride, const convolver<TIn> &conv, size_t rounded_stringlen,
+    void PrepareA_convolution(TInput *out, const TIn *ptr, size_t stride, const convolver<TIn> &conv, size_t rounded_stringlen,
                               const int y0, const int ymax, const int k0, const int kmax, int32_t row_sum_multiplier) {
         ConvolutionInterleave<height, block, VLType::None>(out, ptr, stride, conv, rounded_stringlen, y0, ymax, k0, kmax, integrate_sums, row_sum_multiplier);
     }
@@ -66,7 +66,7 @@ public:
     }
 
     template<typename TIn>
-    void PrepareB(TOperand *out, const TIn *in, const int stride, const int x0,
+    void PrepareB(TWeight *out, const TIn *in, const int stride, const int x0,
                   const int xmax, const int k0, const int kmax, bool transposed) {
         assert (!transposed);
         Transform<width_vectors, block,  true, VLType::SVE>(out, in, stride, x0, xmax, k0, kmax);