Strix Halo HomeLab
Attachments
History
Blame
View Source
Documentation
About An Otter Wiki
Toggle dark mode
Login
Home
A - Z
Changelog
Menu
GitHub Mirror
Discord Server
Page Index
Guides
110GB-of-VRAM
AI-Capabilities
C-States
Hardware-Monitoring
Power-Mode-and-Fan-Control
Power-Modes-and-Performance
Replacing-Thermal-Interfaces-On-GMKtec-EVO-X2
VM-iGPU-Passthrough
Hardware
Boards
Sixunited-AXB35
Firmware
PCs
Bosgame-M5
FEVM-FA-EX9
Framework-Desktop
GMKtec-EVO-X2
HP-Z2-Mini-G1a
Peladn-YO1
Home
Guides
AI-Capabilities
0bbe74
Commit
0bbe74
2025-06-18 10:07:25
deseven
: added example config file
Guides/AI-Capabilities.md
..
@@ 13,7 13,7 @@
MoE models work very well, hopefully we'll see some 50-70B ones in the future, they could be the real sweet spot for this hardware.
-
Some real-life examples (KoboldCPP, Vulkan):
+
Some real-life examples (KoboldCPP, Vulkan, [example config file](./gemma-3-27b.kcpps)):
| Model | Quantization | Prompt Processing | Generation Speed |
| ----------------- | ------------ | ----------------- | ---------------- |
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9