Conversation
Notices
-
Embed this notice
snacks (snacks@netzsphaere.xyz)'s status on Thursday, 27-Feb-2025 16:33:57 JST snacks
Should i get the framework thing? Gpu is prob about as fast as my desktop and 110gb unified sounds insanely neat -
Embed this notice
lainy (lain@lain.com)'s status on Thursday, 27-Feb-2025 16:40:57 JST lainy
@snacks if you want to do AI related stuff it seems like a great alternative to macs snacks likes this.snacks repeated this. -
Embed this notice
snacks (snacks@netzsphaere.xyz)'s status on Thursday, 27-Feb-2025 16:43:25 JST snacks
@lain also interesting for some graphics stuff like gaussian splatting tbh -
Embed this notice
lainy (lain@lain.com)'s status on Thursday, 27-Feb-2025 16:43:53 JST lainy
@snacks although the mac studio has ram that's over 3 times faster. will be interesting to see what nvidia actually puts into their AI desktop. -
Embed this notice
snacks (snacks@netzsphaere.xyz)'s status on Thursday, 27-Feb-2025 16:46:35 JST snacks
@lain the new apu boom seems to just be powered by lpddr5x? And apple pioneering it. Fucking weird considering that was supposed to be amds thing -
Embed this notice
lainy (lain@lain.com)'s status on Thursday, 27-Feb-2025 16:47:55 JST lainy
@snacks i have no idea why AMD isn't dropping everything to just produce software that makes them compatible with CUDA. Everyone wants to run AI workloads now, but nobody makes high ram cards. snacks and Polychrome :blabcat: like this. -
Embed this notice
snacks (snacks@netzsphaere.xyz)'s status on Thursday, 27-Feb-2025 17:00:32 JST snacks
@lain actually why? Even nvidias digits thing seems to just use lpddr5 -
Embed this notice
Polychrome :blabcat: (polychrome@poly.cybre.city)'s status on Thursday, 27-Feb-2025 17:23:41 JST Polychrome :blabcat:
@lain @snacks I chose AMD because Linux and now it bites me every time I want to run something and it turns out to be a CUDA exclusive.
There's ROCm but in reality that thing is a pain to get working with Conda, which is the usual combination for what I keep encountering. -
Embed this notice
lainy (lain@lain.com)'s status on Thursday, 27-Feb-2025 17:35:33 JST lainy
@Polychrome @snacks yeah, usually it's possible to use rocm but it's just such a huge amount of work, because people are either running on nvidia or mac, and nobody really pays attention to amd snacks likes this. -
Embed this notice
lainy (lain@lain.com)'s status on Thursday, 27-Feb-2025 18:24:38 JST lainy
@k1tteh @snacks it's still going, i just don't get why AMD isn't prioritizing this. -
Embed this notice
plat (k1tteh@uwu.social)'s status on Thursday, 27-Feb-2025 18:24:39 JST plat
-
Embed this notice
snacks (snacks@netzsphaere.xyz)'s status on Thursday, 27-Feb-2025 22:16:05 JST snacks
@nerthos @lain is this a joke about indians? -
Embed this notice
Qadse Sanur Qut ibn Quaadibz :catsader: (nerthos@shitposter.world)'s status on Thursday, 27-Feb-2025 22:16:06 JST Qadse Sanur Qut ibn Quaadibz :catsader:
@snacks @lain >apu boom snacks likes this. -
Embed this notice
Qadse Sanur Qut ibn Quaadibz :catsader: (nerthos@shitposter.world)'s status on Thursday, 27-Feb-2025 22:17:19 JST Qadse Sanur Qut ibn Quaadibz :catsader:
@snacks @lain No snacks likes this.
-
Embed this notice