vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 09:34:37 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fixed mmap prefetch for GPU offloading

Browse source
This commit is contained in:
JohannesGaessler 2023-08-06 10:18:05 +02:00
parent 86c3219895
commit d9024df759
2 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

2
llama-util.h
View file

@ -219,7 +219,7 @@ struct llama_mmap {
// prefetch/readahead impairs performance on NUMA systems
if (numa) { prefetch = 0; }
#ifdef __linux__
if (prefetch) { flags |= MAP_POPULATE; }
if (prefetch >= file->size) { flags |= MAP_POPULATE; }
#endif
addr = mmap(NULL, file->size, PROT_READ, flags, fd, 0);
if (addr == MAP_FAILED) {