You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm curious what kind of performance you guys are getting using the --lowvram option on GPUs with 2GB VRAM, and what optimization flags everyone is using.
Currently I'm using a 2GB 920MX, which is probably one of the slowest GPUs out there that can run SD. It's run on Linux with an Optimus setup so my display runs off the iGPU and I can devote the full 2GB to SD.
With euler_a I get around 11s/it with flags --lowvram --opt-split-attention --always-batch-cond-uncond. --lowvram is obviously a must to get it running on 2GB, and I get around 16s/it with only that flag. --opt-channelslast didn't do anything for me so I left it off, and I have not tried xformers due to the complex setup. Thankfully my GPU supports fp16 so I don't need to take the hit of running in fp32.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
I'm curious what kind of performance you guys are getting using the
--lowvram
option on GPUs with 2GB VRAM, and what optimization flags everyone is using.Currently I'm using a 2GB 920MX, which is probably one of the slowest GPUs out there that can run SD. It's run on Linux with an Optimus setup so my display runs off the iGPU and I can devote the full 2GB to SD.
With euler_a I get around 11s/it with flags
--lowvram --opt-split-attention --always-batch-cond-uncond
.--lowvram
is obviously a must to get it running on 2GB, and I get around 16s/it with only that flag.--opt-channelslast
didn't do anything for me so I left it off, and I have not tried xformers due to the complex setup. Thankfully my GPU supports fp16 so I don't need to take the hit of running in fp32.Beta Was this translation helpful? Give feedback.
All reactions