vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-09 19:46:11 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

llama: use FA + max. GPU layers by default (#15434)

Browse source 
* llama: use max. GPU layers by default, auto -fa

* ggml-backend: abort instead of segfault
This commit is contained in:
Johannes Gäßler 2025-08-30 16:32:10 +02:00 • committed by GitHub
parent 38ad381f9f
commit e81b8e4b7f
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
19 changed files with 235 additions and 72 deletions
Download patch file Download diff file Expand all files Collapse all files

12
src/llama.cpp
View file

@ -25,6 +25,18 @@
// interface implementation
//
const char * llama_flash_attn_type_name(enum llama_flash_attn_type flash_attn_type) {
switch (flash_attn_type) {
case LLAMA_FLASH_ATTN_TYPE_AUTO:
return "auto";
case LLAMA_FLASH_ATTN_TYPE_DISABLED:
return "disabled";
case LLAMA_FLASH_ATTN_TYPE_ENABLED:
return "enabled";
}
GGML_ABORT("fatal error");
}
struct llama_sampler_chain_params llama_sampler_chain_default_params() {
struct llama_sampler_chain_params result = {
/*.no_perf =*/ true,