github/redis
1
1
Fork 0
mirror of https://github.com/redis/redis.git synced 2026-04-21 03:01:35 -04:00
Code Issues Packages Projects Releases Wiki Activity

reworded the AVX error message when compiling xxh_x86dispatch.c

Browse Source 
Suggest the XXH_X86DISPATCH_ALLOW_AVX solution upfront.
This commit is contained in:
Yann Collet
2023-06-19 00:31:28 -07:00
parent f11745eb2e
commit 3682673116
1 changed files with 8 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
xxh_x86dispatch.c
View File

@@ -60,29 +60,28 @@ extern "C" {
* @def XXH_X86DISPATCH_ALLOW_AVX * @def XXH_X86DISPATCH_ALLOW_AVX
* @brief Disables the AVX sanity check. * @brief Disables the AVX sanity check.
* *
* Don't compile xxh_x86dispatch.c with options like `-mavx*`, `-march=native`, * xxh_x86dispatch.c is intended to be compiled for the minimum target, and
* or `/arch:AVX*`. It is intended to be compiled for the minimum target, and
* it selectively enables SSE2, AVX2, and AVX512 when it is needed. * it selectively enables SSE2, AVX2, and AVX512 when it is needed.
* *
* Using this option _globally_ allows this feature, and therefore makes it * Compiling with options like `-mavx*`, `-march=native`, or `/arch:AVX*`
* _globally_ will always enable this feature, and therefore makes it
* undefined behavior to execute on any CPU without said feature. * undefined behavior to execute on any CPU without said feature.
* *
* Even if the source code isn't directly using AVX intrinsics in a function, * Even if the source code isn't directly using AVX intrinsics in a function,
* the compiler can still generate AVX code from autovectorization and by * the compiler can still generate AVX code from autovectorization and by
* "upgrading" SSE2 intrinsics to use the VEX prefixes (a.k.a. AVX128). * "upgrading" SSE2 intrinsics to use the VEX prefixes (a.k.a. AVX128).
* *
* Use the same flags that you use to compile the rest of the program; this * Define XXH_X86DISPATCH_ALLOW_AVX to ignore this check,
* file will safely generate SSE2, AVX2, and AVX512 without these flags. * thus accepting that the produced binary will not work correctly
* * on any CPU with less features than the ones stated at compilation time.
* Define XXH_X86DISPATCH_ALLOW_AVX to ignore this check, and feel free to open
* an issue if there is a target in the future where AVX is a default feature.
*/ */
#ifdef XXH_DOXYGEN #ifdef XXH_DOXYGEN
# define XXH_X86DISPATCH_ALLOW_AVX # define XXH_X86DISPATCH_ALLOW_AVX
#endif #endif
#if defined(__AVX__) && !defined(XXH_X86DISPATCH_ALLOW_AVX) #if defined(__AVX__) && !defined(XXH_X86DISPATCH_ALLOW_AVX)
# error "Do not compile xxh_x86dispatch.c with AVX enabled! See the comment above." # error "Error: if xxh_x86dispatch.c is compiled with AVX enabled, the resulting binary will crash on sse2-only cpus !! " \
"If you nonetheless want to do that, please enable the XXH_X86DISPATCH_ALLOW_AVX build variable"
#endif #endif
#ifdef __has_include #ifdef __has_include