Add support for non-constant weights and biases in CpuFullyConnected
Changing the approach for specifying that weights and biases tensors are
non-constant by making it a member of TensorInfo rather than an option
of the functions.
Resolves: COMPMID-4222
Change-Id: I96e6f3868f51785c9700a3ef6a1fe7b05747862c
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6162
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
diff --git a/src/core/NEON/kernels/arm_gemm/gemm_hybrid_quantized.hpp b/src/core/NEON/kernels/arm_gemm/gemm_hybrid_quantized.hpp
index c72dca2..efb5bd1 100644
--- a/src/core/NEON/kernels/arm_gemm/gemm_hybrid_quantized.hpp
+++ b/src/core/NEON/kernels/arm_gemm/gemm_hybrid_quantized.hpp
@@ -269,12 +269,16 @@
return get_col_sum_size() + (roundup(_Nsize, strategy::out_width()) * roundup(_Ksize, strategy::k_unroll()) * _nmulti * sizeof(Toi));
}
- void pretranspose_B_array(void *in_buffer, const To *B, const int ldb, const int B_multi_stride) override {
+ void requantize_bias(void *in_buffer, const To *B, const int ldb, const int B_multi_stride) override {
col_bias = reinterpret_cast<int32_t *>(in_buffer);
for (unsigned int i=0; i<_nmulti; i++) {
compute_col_sums(_qp, _Nsize, _Ksize, B + (i * B_multi_stride), ldb, col_bias + (i * _Nsize), _Ksize, i, 0);
}
+ }
+
+ void pretranspose_B_array(void *in_buffer, const To *B, const int ldb, const int B_multi_stride) override {
+ requantize_bias(in_buffer, B, ldb, B_multi_stride);
uintptr_t buffer_int = reinterpret_cast<uintptr_t>(in_buffer);
Toi *buffer = reinterpret_cast<Toi *>(buffer_int + get_col_sum_size());