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

CUDA: use mul_mat_q kernels by default (#2683)

Browse source
This commit is contained in:
Johannes Gäßler 2023-08-22 22:47:05 +02:00 • committed by GitHub
parent 3b6cfe7c92
commit c63bb1d16a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 16 additions and 17 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-cuda.cu
View file

@ -287,7 +287,7 @@ static int g_device_count = -1;
static int g_main_device = 0;
static int g_compute_capabilities[GGML_CUDA_MAX_DEVICES];
static float g_tensor_split[GGML_CUDA_MAX_DEVICES] = {0};
static bool g_mul_mat_q = false;
static bool g_mul_mat_q = true;
static void * g_scratch_buffer = nullptr;
static size_t g_scratch_size = 1024*1024*1024; // 1 GB by default