kate/BadFish
1
0
Fork 0
mirror of https://github.com/sockspls/badfish synced 2025-04-30 08:43:09 +00:00
Code Issues Releases Wiki Activity

Fix AVX512 build with older compilers

Browse source 
avoids an intrinsic that is missing in gcc < 10.

For this target, might trigger another gcc bug on windows that
requires up-to-date gcc 8, 9, or 10, or usage of clang.

Fixes https://github.com/official-stockfish/Stockfish/issues/2975

closes https://github.com/official-stockfish/Stockfish/pull/2976

No functional change
This commit is contained in:
mstembera 2020-08-10 12:52:46 -07:00 • committed by Joost VandeVondele
parent 399cddf444
commit f46c73040c
2 changed files with 2 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
src/Makefile
View file

@ -416,7 +416,7 @@ endif
ifeq ($(avx512),yes)
CXXFLAGS += -DUSE_AVX512
ifeq ($(comp),$(filter $(comp),gcc clang mingw))
CXXFLAGS += -mavx512bw
CXXFLAGS += -mavx512f -mavx512bw
endif
endif

3
src/nnue/layers/affine_transform.h
View file

@ -126,8 +126,7 @@ namespace Eval::NNUE::Layers {
const auto iv256 = reinterpret_cast<const __m256i*>(&input_vector[kNumChunks]);
const auto row256 = reinterpret_cast<const __m256i*>(&row[kNumChunks]);
__m256i product256 = _mm256_maddubs_epi16(_mm256_loadA_si256(&iv256[0]), _mm256_load_si256(&row256[0]));
product256 = _mm256_madd_epi16(product256, _mm256_set1_epi16(1));
sum = _mm512_add_epi32(sum, _mm512_zextsi256_si512(product256));
sum = _mm512_add_epi32(sum, _mm512_cvtepi16_epi32(product256));
}
output[i] = _mm512_reduce_add_epi32(sum) + biases_[i];