diff options
author | David Mansell <David.Mansell@arm.com> | 2023-08-23 16:00:13 +0100 |
---|---|---|
committer | Viet-Hoa Do <viet-hoa.do@arm.com> | 2023-08-29 10:14:10 +0000 |
commit | b7aefd71d07d56b001e795410700cae71a518eca (patch) | |
tree | d913041260c04ad56872a4fb45d39c28a18a9528 /src/core/NEON/kernels/arm_gemm/interleave_indirect_impl.hpp | |
parent | cea7060684ae6c33fc8e16affc1c7998d17815ae (diff) | |
download | ComputeLibrary-b7aefd71d07d56b001e795410700cae71a518eca.tar.gz |
GEMM: AArch32: Split assembler block in a32_merge_float_8x6.hpp
Inline assembler blocks attempting to bind 8 integer
registers don't compile in certain configurations (notably GCC 13.2 debug
builds with -O0 -g). Fix this by splitting the offending block into two
separate parts (straightforward as there is no flow control in the block).
Fixes: COMPMID-6532
Signed-off-by: David Mansell <David.Mansell@arm.com>
Change-Id: I80e9a10e6a91574176d50e63c45fab055aefa659
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10197
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Emanuele Rocca <ema@linux.it>
Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/interleave_indirect_impl.hpp')
0 files changed, 0 insertions, 0 deletions