Re:Kimi K3 is now #1 in the Front end Code Arena with 1679 p...
by ALLTaken on Friday July 17, 2026 @02:20AM
(#48943896)
I wish we could stream active data to RAM, directly from the NVMe drive for the 2TB K3 model. Can't wait for prism-ml/ to release a ternary 2bit model, which would make this a bit more tenable.
[ Parent
| on HN ]
Re:Kimi K3 is now #1 in the Front end Code Arena with 1679 p...
by adrian_b on Friday July 17, 2026 @08:07AM
(#48946362)
You can, but the reported speeds for such big models were in the range from a token per a dozen seconds to slightly better than 1 or 2 tokens per second.
Re:Kimi K3 is now #1 in the Front end Code Arena with 1679 p...
by halJordan on Friday July 17, 2026 @01:44PM
(#48950103)
A 2 bit model is still 700gbs you need to stream through.
I know we say reading is free (vs writing). But on a 2tb drive, you're doing a full drive read for every three tokens. Thats 333,000 drive reads just to fill up the context. Well, this is at least an moe model so not that terrible. But i think the point remains