You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
crc32c already contains runtime dispatch, but only if you compiles with sse4.2.
So if I want to compile x86_64 binary without sse42 it will be always slow, even if I run on modern cpu.
It's possible to have runtime dispatch between different architecture, you can see example in simdjson https://github.com/simdjson/simdjson/blob/master/HACKING.md#runtime-dispatching
So what do you think?
From another side probably most of users compiles with sse4.2 now
But it will be possible to try to improve things with avx, but without switching to avx for whole library
Not sure is it good or not, because switching between avx instructions and sse has significant runtime cost. Probably depends on input size
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
crc32c already contains runtime dispatch, but only if you compiles with sse4.2.
So if I want to compile x86_64 binary without sse42 it will be always slow, even if I run on modern cpu.
It's possible to have runtime dispatch between different architecture, you can see example in simdjson https://github.com/simdjson/simdjson/blob/master/HACKING.md#runtime-dispatching
So what do you think?
From another side probably most of users compiles with sse4.2 now
But it will be possible to try to improve things with avx, but without switching to avx for whole library
Not sure is it good or not, because switching between avx instructions and sse has significant runtime cost. Probably depends on input size
Beta Was this translation helpful? Give feedback.
All reactions