Skip to content

Commit b3570b9

Browse files
add performance data
1 parent 0e91e0e commit b3570b9

File tree

1 file changed

+8
-1
lines changed

1 file changed

+8
-1
lines changed

docs/backend/SYCL.md

Lines changed: 8 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -43,7 +43,14 @@ The following release is verified with good quality:
4343
## News
4444

4545
- 2025.2
46-
- Optimize MUL_MAT Q4_0 on Intel CPU. Increase the performance of LLM (llama-2-7b.Q4_0.gguf) 21%-87% on Intel GPUs (MTL, ARL-H, Arc, Flex, PVC).
46+
- Optimize MUL_MAT Q4_0 on Intel GPU for all dGPUs and built-in GPUs since MTL. Increase the performance of LLM (llama-2-7b.Q4_0.gguf) 21%-87% on Intel GPUs (MTL, ARL-H, Arc, Flex, PVC).
47+
|GPU|Base tokens/s|Increased tokens/s|Percent|
48+
|-|-|-|-|
49+
|PVC 1550|39|73|+87%|
50+
|Flex 170|39|50|+28%|
51+
|Arc770|42|55|+30%|
52+
|MTL|13|16|+23%|
53+
|ARL-H|14|17|+21%|
4754

4855
- 2024.11
4956
- Use syclcompat to improve the performance on some platforms. This requires to use oneAPI 2025.0 or newer.

0 commit comments

Comments
 (0)