AuroraMiddleware/qt5base-lts
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

AVX512 test: an intrinsic that GCC forgot to implement prior to GCC8

Browse Source 
_mm512_mask_cvtepi32_storeu_epi8 is VPMOVDB (convert from 32-bit to 8-bit
with truncation) where the destination is a memory address, with an
OpMask register used to indicate which of the lanes in the vector to
store. Similarly, _mm512_mask_cvtepi16_storeu_epi8 is VPMOVWB (convert
from 16-bit o 8-bit), which is useful for UTF-16 to Latin1 conversion.

Change-Id: I8f261579aad648fdb4f0fffd15542ea306841ce6
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
This commit is contained in:
Thiago Macieira 2018-09-13 22:57:30 -07:00
parent 6bb0cbc686
commit d241616e41
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
config.tests/avx512/avx512.cpp
View File

@ -65,6 +65,10 @@ int main(int, char**argv)
d = _mm512_loadu_pd((double *)argv + 64);
f = _mm512_loadu_ps((float *)argv + 128);
// some intrinsic that GCC forgot until GCC 8
i = _mm512_maskz_set1_epi32(m, '?');
_mm512_mask_cvtepi32_storeu_epi8(argv, m, i);
#ifdef WANT_AVX512ER
/* AVX512 Exponential and Reciprocal */
f = _mm512_exp2a23_round_ps(f, 8);
@ -84,6 +88,7 @@ int main(int, char**argv)
#ifdef WANT_AVX512BW
/* AVX512 Byte and Word support */
i = _mm512_mask_loadu_epi8(i, m, argv - 8);
_mm512_mask_cvtepi16_storeu_epi8(argv + 8, m, i);
#endif
#ifdef WANT_AVX512VL
/* AVX512 Vector Length */