vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-10 04:00:53 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

CUDA: fix crash on uneven context without FA (#16988)

Browse source
This commit is contained in:
Johannes Gäßler 2025-11-06 14:05:47 +01:00 • committed by GitHub
parent 5b180c3d60
commit aa374175c3
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
7 changed files with 44 additions and 38 deletions
Download patch file Download diff file Expand all files Collapse all files

2
src/llama-context.cpp
View file

@ -21,6 +21,8 @@ llama_context::llama_context(
llama_context_params params) :
model(model),
balloc(std::make_unique<llama_batch_allocr>(model.hparams.n_pos_per_embd())) {
// TODO warning when creating llama_context with awkward ctx size that is not a power of 2,
// may need to be backend-dependent
LLAMA_LOG_INFO("%s: constructing llama_context\n", __func__);
t_start_us = model.t_start_us;