Conversation

Notices

1. Embed this notice
   Forest of Enchantment (forestofenchantment@clubcyberia.co)'s status on Thursday, 14-Nov-2024 02:40:33 JST Forest of Enchantment
   Fuck you Nvidia for making every other Linux driver release a buggy underperforming piece of crap
   • snacks likes this.
   • Embed this notice
     iced depresso (icedquinn@blob.cat)'s status on Thursday, 14-Nov-2024 02:44:06 JST iced depresso

     in reply to

     • Maija
     @Forestofenchantment @maija its possible to do, albeit weird sometimes. you either need to run linux (rocm) or use one of the frameworks that speaks vulkan compute
   • Embed this notice
     Forest of Enchantment (forestofenchantment@clubcyberia.co)'s status on Thursday, 14-Nov-2024 02:44:07 JST Forest of Enchantment

     in reply to

     • Maija
     @maija I will once their cards can do AI.
   • Embed this notice
     Maija (maija@netzsphaere.xyz)'s status on Thursday, 14-Nov-2024 02:44:08 JST Maija

     in reply to
     @Forestofenchantment just use ayyynd chud :chungus:
   • Embed this notice
     iced depresso (icedquinn@blob.cat)'s status on Thursday, 14-Nov-2024 02:58:06 JST iced depresso

     in reply to

     • ICScaryThings
     • Maija
     @icst @Forestofenchantment @maija amd sponsored such a thing and nvidia shit an entire costco of diapers about it
   • Embed this notice
     ICScaryThings (icst@clubcyberia.co)'s status on Thursday, 14-Nov-2024 02:58:07 JST ICScaryThings

     in reply to

     • iced depresso
     • Maija
     @icedquinn @Forestofenchantment @maija I wonder if it's possible anyone will ever write a reverse engineered cuda compiler using llvm or something. I would think it should be possible, even if difficult, as long as Nvidia doesn't have some horseshit software patent.