Conversation

Notices

1. Embed this notice
   PC-9801 Enjoyer (pawlicker@bae.st)'s status on Monday, 15-Apr-2024 12:49:08 JST PC-9801 Enjoyer
   intel made a gpu for poverty kids
   image.png
   • Embed this notice
     iced depresso (icedquinn@blob.cat)'s status on Monday, 15-Apr-2024 12:49:07 JST iced depresso

     in reply to
     @Pawlicker good for AV1 encoding i guess
   • Embed this notice
     :blobancap: :blobcattrans: :blobancap: :blobcattrans: :blobancap: :blobcattrans: (allison@hidamari.apartments)'s status on Monday, 15-Apr-2024 12:52:05 JST :blobancap: :blobcattrans: :blobancap: :blobcattrans: :blobancap: :blobcattrans:

     in reply to

     • iced depresso
     @icedquinn @Pawlicker Good for AV1 encoding, good for emulators, good for "just works" under Linux for things that use oneAPI compared to whatever baroque mess AMD's compute stack is
   • Embed this notice
     bronze (bronze@pl.kitsunemimi.club)'s status on Monday, 15-Apr-2024 12:53:32 JST bronze

     in reply to

     • Vivi Nella Verita
     @verita84eva @Pawlicker you know I hate windows but DirectML is one of the few things that's actually made me consider a pass-through VM :holo_think:
   • Embed this notice
     dotnet@loli.church's status on Monday, 15-Apr-2024 12:53:32 JST dotnet

     in reply to

     • bronze
     • Vivi Nella Verita

     @bronze@pl.kitsunemimi.club @verita84eva@poster.place @Pawlicker@bae.st last time I tried it, DirectML was pretty bad, its only value was that it was better than using the CPU.

     :blobancap: :blobcattrans: :blobancap: :blobcattrans: :blobancap: :blobcattrans: likes this.
   • Embed this notice
     Vivi Nella Verita (verita84eva@poster.place)'s status on Monday, 15-Apr-2024 12:53:38 JST Vivi Nella Verita

     in reply to

     • bronze
     @bronze @Pawlicker
     
     Yeah....For stable diffusion, I had to use this fork for Windows
     
     https://github.com/lshqqytiger/stable-diffusion-webui-directml/issues
   • Embed this notice
     Vivi Nella Verita (verita84eva@poster.place)'s status on Monday, 15-Apr-2024 12:53:40 JST Vivi Nella Verita

     in reply to

     • bronze
     @Pawlicker @bronze
     
     It does not looks like all the AI stuff supports ARC yet. Hell, even having Radeons are limited
   • Embed this notice
     bronze (bronze@pl.kitsunemimi.club)'s status on Monday, 15-Apr-2024 12:53:40 JST bronze

     in reply to

     • Vivi Nella Verita
     @verita84eva @Pawlicker Radeon for ML is such a piece of shit its not even funny. AMD does not give a fuck about ROCm for the gaming GPUs, so you maybe get support or maybe not depending on the model.
     Meanwhile I've seen someone with a piece of shit GTX760 churn out some 400x400 pics with the CUDA support he STILL has.
   • Embed this notice
     PC-9801 Enjoyer (pawlicker@bae.st)'s status on Monday, 15-Apr-2024 12:53:41 JST PC-9801 Enjoyer

     in reply to

     • bronze
     • Vivi Nella Verita
     @verita84eva @bronze they have a cheap 16gb card too
   • Embed this notice
     Vivi Nella Verita (verita84eva@poster.place)'s status on Monday, 15-Apr-2024 12:53:42 JST Vivi Nella Verita

     in reply to

     • bronze
     @bronze @Pawlicker
     
     My 12 GB card is barely enough
   • Embed this notice
     bronze (bronze@pl.kitsunemimi.club)'s status on Monday, 15-Apr-2024 12:53:43 JST bronze

     in reply to
     @Pawlicker you too will be able to do stable diffusion for a mere $100 :blobfoxlaughsweat:
   • Embed this notice
     djsumdog (djsumdog@djsumdog.com)'s status on Tuesday, 16-Apr-2024 00:15:05 JST djsumdog

     in reply to

     I have one of these. I was using it while trying to debug issues with amdgpu drivers causing kernel panics. It's not a bad car for basic stuff honestly.