vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 09:04:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

vulkan multigpu, show uptime

Browse source
This commit is contained in:
Concedo 2024-02-08 16:54:38 +08:00
parent ec2dbd99a3
commit 4cd571db89
6 changed files with 50 additions and 14 deletions
Download patch file Download diff file Expand all files Collapse all files

4
gpttype_adapter.cpp
View file

@ -865,7 +865,7 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
llama_ctx_params.rope_freq_scale = rope_freq_scale;
llama_ctx_params.n_batch = kcpp_params->n_batch;
#if defined(GGML_USE_CUBLAS)
#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_VULKAN)
bool ts_all_zero = true;
for (int i = 0; i < tensor_split_max; ++i) {
if (inputs.tensor_split[i] != 0.0f) {
@ -966,7 +966,7 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
llama_ctx_params.n_threads = kcpp_params->n_threads;
llama_ctx_params.n_threads_batch = kcpp_params->n_threads_batch;
#if defined(GGML_USE_CUBLAS)
#if defined(GGML_USE_CUBLAS) || defined(GGML_USE_VULKAN)
bool ts_all_zero = true;
for (int i = 0; i < tensor_split_max; ++i) {
if (inputs.tensor_split[i] != 0.0f) {