If you have a 5090 run an nvfp4 quant to get Blackwell performance gains for minimal quality loss (some performing as well as q4_1 in divergence)
2026-08-25 22:16:42
4
libera_et_impera (marius) :
This is LARPing
2026-08-26 12:26:10
0
AJ :
cries in 1660 ti
2026-08-25 22:49:59
0
danisalmi :
can i run It with 16gbvram? i got a 7800xt
2026-08-22 12:55:27
3
Malik :
What are the best setting for rtx 5080 16gb vram
2026-08-22 17:54:41
2
Marc :
Since when have they released an MTP head? Also, doesn’t it just use jinja by default?
2026-08-22 16:45:30
0
Gotxi :
rocking mine on my rx 9070 xt, 34-60 tok/s 80k context
2026-08-23 17:12:24
3
Realni :
I tried Dspark2 on a 4080 16gb it got ~80tok/s on a short context at 4Q really wish I had a 50s series card to try the new inference engine from Barkley that uses NVFP4
2026-08-22 12:56:09
0
cpstest_ :
can i run it on 2060?
2026-08-22 12:56:51
0
zenasked :
any optimal settings for the 3080?
2026-08-22 18:43:34
0
Kyrox :
can it run with 4gb vram?
2026-08-24 15:57:48
0
Dr.Papa :
Ornith 1.5 397b is better
2026-08-22 19:11:12
0
ً :
Content is 🔥
2026-08-22 12:38:36
1
valesia :
can I run something on 5070 & 48gb DDR4 2400mt/s?
2026-08-22 14:37:49
1
/home/wvga :
3080 10gb vram?
2026-08-22 18:35:39
1
King Ragz :
My 4080 isn’t good enough…
2026-08-24 12:20:59
4
subliminalistic :
2 Tesla P40s
2026-08-22 21:52:43
1
YuriTarded :
I got 5070 12GB and 5060 ti 16GB
2026-08-24 23:00:09
0
лигмаム :
my 3060 is sobbing
2026-08-24 21:42:48
3
Elvis :
eat
2026-08-22 12:47:23
0
el_framerte :
How does this work in LM Studio?
2026-08-22 17:20:59
3
♡Hannah♡ :
I have a 5090 but i cant run it in q8
2026-08-22 13:41:27
0
krystian ☭ :
święta stasia z piłka wiedzą
2026-08-23 19:47:51
1
magnificent king felix the 24h :
will this work on my intel hd graphics 4600?
2026-08-22 22:23:58
3
𝓕𝓻𝓮𝓪𝓴𝔂 ant :
I honestly don’t see how the frontier models are gouging to compete 😭
2026-08-23 21:34:24
1
To see more videos from user @larpu_acc, please go to the Tikwm
homepage.