danielkrause/colibri-strix
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
Vincenzo
2026-07-05 23:16:58 +02:00
committed by GitHub
parent 31d91b2c5b
commit 1e33e9b17d
1 changed files with 3 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+3
View File
@@ -45,6 +45,9 @@ The engine is a single C file (`c/glm.c`, ~1,300 lines) plus small headers. No B
| disk ceiling (VHDX random) | ~1 GB/s → ~0.05–0.1 tok/s cold | | disk ceiling (VHDX random) | ~1 GB/s → ~0.05–0.1 tok/s cold |
| MTP speculation | 2.0 tok/forward measured | | MTP speculation | 2.0 tok/forward measured |
### SSD Wear Warning
Cold starts are heavy on random reads. While reads are mostly safe, the OS page cache can cause writes. Heavy use may accelerate SSD wear, especially on cheaper drives. Use with caution and monitor your drive health.
This is not fast. It is a 744B frontier-class model **answering correctly on a machine that costs less than one H100 fan**. Warm cache, pinned hot experts and MTP push the useful-response latency down considerably; the physics of the disk does the rest. This is not fast. It is a 744B frontier-class model **answering correctly on a machine that costs less than one H100 fan**. Warm cache, pinned hot experts and MTP push the useful-response latency down considerably; the physics of the disk does the rest.
## Quick start ## Quick start