monster/modelbeast
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

benchmarks: Qwen-Layered cross-bench — m3ultra 243.0s (1.50x vs m1), both Ultras serving from baked q8

Browse Source 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
m3ultra 2026-07-19 21:41:28 +10:00
parent 0db2c90481
commit 8849438bce
1 changed files with 11 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
BENCHMARKS.md
View File

@ -362,3 +362,14 @@ q4/q6 quality ladder, m3ultra cross-bench.
Remaining ladder (untested): q6/q4 quality/speed, 1024-res, fewer-steps floor (10?),
m3ultra cross-bench (~1.4× compute but needs 34GB disk — m3 disk currently tight),
MODELBEAST operator wrapper (qwen_layered_local) once defaults settle.
### Cross-bench addendum: m3ultra joins (same baked q8, 20 steps, 640, 4 layers)
| box | wall | note |
|---|---|---|
| m3ultra | **243.0 s** | 1.50× vs m1 — matches fleet-matrix TFLOPS ratio (1.44×) |
| m1ultra | 365.6 s | |
Both Ultras now serve Qwen-Image-Layered from the same 34GB baked-q8
artifact (m3: ~/qwen-layered/, m1: ~/qwen-layered-staging/) + the same
Gitea fork clone. Peak 36.7GB either box. m3 disk after install: ~82GB free.