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

minor fixes

Browse source 
* update outdated comment
* fix coding style
This commit is contained in:
Reinforce-II 2024-01-22 08:12:57 +00:00
parent 2c0ed7a638
commit 130b0812d0
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml.c
View file

@ -16796,7 +16796,7 @@ struct ggml_cplan ggml_graph_plan(const struct ggml_cgraph * cgraph, int n_threa
#if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS) #if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS)
if (ggml_compute_forward_mul_mat_use_blas(node)) { if (ggml_compute_forward_mul_mat_use_blas(node)) {
if (node->src[0]->type != GGML_TYPE_F32) { if (node->src[0]->type != GGML_TYPE_F32) {
// here we need memory just for single 2D matrix from src0 // here we need memory for fully dequantized matrix from src0
cur = ggml_type_size(GGML_TYPE_F32) cur = ggml_type_size(GGML_TYPE_F32)
* node->src[0]->ne[0]*node->src[0]->ne[1] * node->src[0]->ne[0]*node->src[0]->ne[1]
* node->src[1]->ne[2]*node->src[1]->ne[3]; * node->src[1]->ne[2]*node->src[1]->ne[3];