ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (#13014)

Browse source 
* CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID

* fix logic for RoPE support, CUDA graphs
This commit is contained in:
Johannes Gäßler 2025-04-22 21:27:40 +02:00 • committed by GitHub
parent dc39a5e7a8
commit 658987cfc9
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
9 changed files with 548 additions and 426 deletions
Download patch file Download diff file Expand all files Collapse all files

2
tests/test-backend-ops.cpp
View file

@ -2071,7 +2071,7 @@ struct test_mul_mat_id : public test_case {
const ggml_type type_b;
const int n_mats;
const int n_used;
const bool b; // brodcast b matrix
const bool b; // broadcast b matrix
const int64_t m;
const int64_t n;
const int64_t k;