llama.cpp
CUDA: implement hmax and hmax2 for CUDA < 11.7
#7019

Merged

Commits

CUDA: CUDART < 11.7 workaround for __hmax, __hmax2

JohannesGaessler committed 1 year ago

Loading