Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-10 06:49:12 +02:00
Code Issues Packages Projects Releases Wiki Activity

removed openblas backend, merged into CPU (with llamafile for BLAS). GPU backend is now automatically selected when running from CLI unless noblas is specified.

Browse Source
This commit is contained in:
Concedo
2024-09-15 19:21:52 +08:00
parent 01c7d82185
commit 53bf0fb32d
14 changed files with 115 additions and 159 deletions
Download Patch File Download Diff File Expand all files Collapse all files
gpttype_adapter.cpp
+3 -3
View File
@@ -1624,7 +1624,7 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
gptj_ctx_v3.hparams.rope_freq_scale = neox_ctx_v3.hparams.rope_freq_scale = rope_freq_scale;
gptj_ctx_v3.hparams.rope_freq_base = neox_ctx_v3.hparams.rope_freq_base = rope_freq_base;
//this is used for the mem_per_token eval, openblas needs more RAM
//this is used for the mem_per_token eval, blas needs more RAM
bool v3_use_scratch = ggml_v3_cpu_has_gpublas();
int cu_parseinfo_maindevice = inputs.cublas_info<=0?0:inputs.cublas_info;
@@ -2362,11 +2362,11 @@ int GetThreadsToUse(bool blasmode)
{
if(!ggml_cpu_has_gpublas())
{
return 1;
return std::min(kcpp_data->n_blasthreads, 4);
}
else
{
return kcpp_data->n_blasthreads;
return kcpp_data->n_blasthreads;
}
}
return kcpp_data->n_threads;