root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-01-02 14:54:35 +00:00
Code Issues Actions 9 Packages Projects Releases Wiki Activity

Sync from device back to host at begin of new prompt.

Browse Source
This commit is contained in:
Adam Treat 2023-09-13 20:47:40 -04:00 committed by Cebtenzzre
parent bd5f6399bb
commit 4ed25b2f88
1 changed files with 4 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
llama.cpp
View File

@ -3820,6 +3820,10 @@ static bool llama_eval_internal(
ggml_vk_graph_compute(lctx.ctx_kompute, gf);
ggml_vk_d2h_tensor(lctx.ctx_kompute, res);
} else {
if (lctx.ctx_kompute) {
ggml_vk_d2h_tensor(lctx.ctx_kompute, kv_self.k);
ggml_vk_d2h_tensor(lctx.ctx_kompute, kv_self.v);
}
ggml_graph_compute_helper(lctx.work_buffer, gf, n_threads);
if (lctx.ctx_kompute) {
ggml_vk_h2d_tensor(lctx.ctx_kompute, kv_self.k);