Re:How Is LLaMa.cpp Possible?
by visarga on Wednesday August 16, 2023 @12:19AM
(#37142827)
Some GPU operations give different results depending on the order they are done. This happens because floating point numbers are approximations and lose associativity. Requiring a strict order causes a big slowdown.
[ Parent
| on HN ]
There are no comments on this comment yet.