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

Merged

CUDA: implement hmax and hmax2 for CUDA < 11.7 #7019

JohannesGaessler merged 1 commit into ggml-org:master from JohannesGaessler:cuda-hmax-fix

JohannesGaessler force pushed from 0ea9cd2e to bc8ac98d 1 year ago

JohannesGaessler force pushed from bc8ac98d to 24ea3c6d 1 year ago

CUDA: CUDART < 11.7 workaround for __hmax, __hmax2

859734ee

JohannesGaessler force pushed from 24ea3c6d to 859734ee 1 year ago

slaren commented on 2024-05-01

slaren approved these changes on 2024-05-01

JohannesGaessler merged 1613ef8d into master 1 year ago

Reviewers

slaren

Assignees

No one assigned

Labels

None yet

Milestone

No milestone