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

Update ggml-sycl.cpp

Browse source 
Co-authored-by: Neo Zhang Jianyu <jianyu.zhang@intel.com>
This commit is contained in:
Meng, Hengyu 2024-05-28 21:00:30 +08:00 • committed by GitHub
parent fc08e1a729
commit d63f6b66d4
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-sycl.cpp
View file

@ -15260,7 +15260,7 @@ static void ggml_sycl_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1
// mmvq and mmq need the __dp4a instruction which is available for gen12+ // mmvq and mmq need the __dp4a instruction which is available for gen12+
// Workaround in https://github.com/ggerganov/llama.cpp/commit/95f84d5ce8b449a9b16009434aca800df504a02e // Workaround in https://github.com/ggerganov/llama.cpp/commit/95f84d5ce8b449a9b16009434aca800df504a02e
use_mul_mat_q = use_mul_mat_q && (src0->type != GGML_TYPE_IQ2_XXS); use_mul_mat_q = use_mul_mat_q && (src0->type != GGML_TYPE_IQ2_XXS);
#ifdef SYCL_USE_XMX #ifdef SYCL_USE_XMX
use_mul_mat_q = use_mul_mat_q && (src1->ne[1] <= MMQ_MAX_BATCH_SIZE); use_mul_mat_q = use_mul_mat_q && (src1->ne[1] <= MMQ_MAX_BATCH_SIZE);
#endif // SYCL_USE_XMX #endif // SYCL_USE_XMX