ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-05-04 13:21:37 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add GitHub data: filename sanitization (#640)

Browse Source
This commit is contained in:
Thomas
2025-07-23 13:31:53 +02:00
committed by GitHub
parent 3600d82e98
commit eaa2510a28
626 changed files with 0 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
github-data/pull_requests/610 - q8_k_r8_ experimental AVX512 version.md Normal file
View File

@@ -0,0 +1,17 @@
### 🔀 [#610](https://github.com/ikawrakow/ik_llama.cpp/pull/610) - q8_k_r8: experimental AVX512 version
| **Author** | `ikawrakow` |
| :--- | :--- |
| **State** | ✅ **Open** |
| **Created** | 2025-07-14 |
| **Updated** | 2025-07-18 |
---
#### Description
@ubergarm This is specifically for your 9950X CPU.
On my 7950X this is ~10% slower than what we have on the main branch. The 7950X supports `AVX512`, but 512-bit instructions get executed as two 256-bit instructions. Hence, I'm expecting (hoping?) this `Q8_K_R8` GEMM version to be significantly faster on a CPU with "real" 512-bit instructions such as the 9950X.
Please benchmark it so I can decide if it is worth adding this to the main branch.