Enable premultiplication for depthwise convolution

with fp16 and quantized types

Resolves: COMPMID-6337
Change-Id: I81542e51c9c0329f202ac8452f173b138e51a0f6
Signed-off-by: Michael Tyler <michael.tyler@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/9883
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 1 insertions, 1 deletions

diff --git a/src/core/NEON/kernels/arm_conv/depthwise/interleaves/sve_s8q_3x3_dot.cpp b/src/core/NEON/kernels/arm_conv/depthwise/interleaves/sve_s8q_3x3_dot.cpp
index 3a4999296a..5d7b54f235 100644
--- a/src/core/NEON/kernels/arm_conv/depthwise/interleaves/sve_s8q_3x3_dot.cpp
+++ b/src/core/NEON/kernels/arm_conv/depthwise/interleaves/sve_s8q_3x3_dot.cpp
@@ -42,7 +42,7 @@ size_t interleave_sve_s8q_3x3_dot::get_packed_size(const DepthwiseArgs &args)
 {
   // We store 7 vectors for every <vector_of_ints> of channels.
   const unsigned int n = arm_gemm::roundup(
-    arm_gemm::iceildiv((long unsigned int) args.input_channels,
+    arm_gemm::iceildiv((long unsigned int) args.input_channels * args.channel_multiplier,
                        get_vector_length<int32_t>(arm_gemm::VLType::SVE)), 4lu
   );
   return n * 7 * get_vector_length<int8_t>(arm_gemm::VLType::SVE);