vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 01:24:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

hopefully fixed the ooms for good

Browse source
This commit is contained in:
Concedo 2023-04-23 13:49:50 +08:00
parent 3f21bd81f3
commit 4e1ea2ac61
4 changed files with 8 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

6
otherarch/gptj_v2.cpp
View file

@ -382,11 +382,11 @@ bool gptj_eval(
const int d_key = n_embd/n_head;
//todo: there is a bug that causes the buffer to oom and I cannot figure it out, hack to increase size for now
static size_t buf_size = 512u*1024*1024;
static size_t buf_size = 256u*1024*1024;
static void * buf = malloc(buf_size);
if (mem_per_token > 0 && mem_per_token*N*1.4 > buf_size) {
const size_t buf_size_new = 1.6*(mem_per_token*N); // add 10% to account for ggml object overhead
if (mem_per_token > 0 && mem_per_token*N*1.6 > buf_size) {
const size_t buf_size_new = 1.8*(mem_per_token*N); // add 10% to account for ggml object overhead
//printf("\n%s: reallocating buffer from %zu to %zu bytes\n", __func__, buf_size, buf_size_new);
// reallocate