ExternalVendorCode/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-06-22 16:38:58 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

SYCL: Add non contiguous support in RMS_NORM and NORM kernels (llama/13611)

Browse Source 
* SYCL: Add non contiguous input support to norm kernel

* refactor and add RMS_NORM non contiguous input support

ggml-ci

* restore subgroup reduction for multi-subgroup thread blocks in norm kernels

* Swap grid dims of nsamples and nrows

ggml-ci

* Revert "Swap grid dims of nsamples and nrows"

This reverts commit 43be2d657fec7f7fba54e2cd154106bc0fc45adf.

* restore not required changes
ggml-ci

* address review comments: change it to more like SYCL

* Use a common function to calculate offset

* remove wrap around logic for handling broadcasts

* remove static from calculate_offset fn and use ceil_div
This commit is contained in:
Akarshan Biswas
2025-05-26 21:10:36 +05:30
committed by Georgi Gerganov
parent 25e27904ca
commit 195fde8804
3 changed files with 109 additions and 67 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
ggml/src/ggml-sycl/ggml-sycl.cpp
View File

@ -4241,6 +4241,7 @@ static bool ggml_backend_sycl_device_supports_op(ggml_backend_dev_t dev, const g
#endif
case GGML_OP_NORM:
case GGML_OP_RMS_NORM:
return true;
case GGML_OP_L2_NORM:
case GGML_OP_GROUP_NORM:
return ggml_is_contiguous(op->src[0]);