Toot

Written by Marc Edwards on 2025-01-30 at 01:01

If you’ve read that Apple Silicon Macs are great at AI inference because unified memory lets all the system memory be used, there’s some nuance to the discussion. It’s mostly true. We did some testing here to see how much memory the system would allow for GPU use:

M1 with 16GB: 11GB (71.58%)

M4 Pro with 24GB: 17GB (71.58%)

M4 Max with 48GB: 38GB (80.53%)

=> More informations about this toot | View the thread | More toots from marcedwards@mastodon.social

Mentions

Tags

Proxy Information
Original URL
gemini://mastogem.picasoft.net/toot/113914605653352790
Status Code
Success (20)
Meta
text/gemini
Capsule Response Time
221.936068 milliseconds
Gemini-to-HTML Time
0.432758 milliseconds

This content has been proxied by September (3851b).