COMPMID-2178: Update GEMM assembly code.
Perform offset reduction and requantization within the assembly wrapper.
Change-Id: I5d5b3e1f6f9ef4c71805362c57f88ff199c027a3
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/1541
Comments-Addressed: Pablo Marquez <pablo.tello@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
diff --git a/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_24way_16bit.hpp b/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_24way_16bit.hpp
index 80420dd..6d62733 100644
--- a/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_24way_16bit.hpp
+++ b/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_24way_16bit.hpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2018 ARM Limited.
+ * Copyright (c) 2017-2019 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -37,7 +37,7 @@
// Redirect to a 24 x uint16_t specialisation
TransformImpl<24, 1, true, 2, 2, false>::Transform(
reinterpret_cast<uint16_t *>(out),
- reinterpret_cast<const uint16_t *>(in),
+ reinterpret_cast<const uint16_t * const>(in),
stride*2, x0*2, xmax*2, k0, kmax
);
}
@@ -52,7 +52,7 @@
// Redirect to a uint16_t specialisation
Transform(
reinterpret_cast<uint16_t *>(out),
- reinterpret_cast<const uint16_t *>(in),
+ reinterpret_cast<const uint16_t * const>(in),
stride, x0, xmax, k0, kmax
);
}