Skip to content

use eR8G8B8A8Unorm image format instead of eR16G16B16A16Sfloat#4079

Closed
evill33t wants to merge 1 commit intoshadps4-emu:mainfrom
evill33t:fsr_hdr_optimization
Closed

use eR8G8B8A8Unorm image format instead of eR16G16B16A16Sfloat#4079
evill33t wants to merge 1 commit intoshadps4-emu:mainfrom
evill33t:fsr_hdr_optimization

Conversation

@evill33t
Copy link
Contributor

if HDR is disabled use eR8G8B8A8Unorm instead of eR16G16B16A16Sfloat
halves bandwidth and memory for fsr pass

@raphaelthegreat
Copy link
Contributor

Is there a large perf benefit to this? Lower bit widths cause banding with FSR

@evill33t
Copy link
Contributor Author

the benefit is not really large, a few ns - I wasn't aware about possible banding issues

@raphaelthegreat
Copy link
Contributor

I wouldn't say its worth it if we are talking about ns

@evill33t
Copy link
Contributor Author

would it be worthwhile to explore FSR 3.1?

@raphaelthegreat
Copy link
Contributor

Later FSR versions require motion vectors which cannot be provided in general fashion

@georgemoralis
Copy link
Collaborator

closed as probably not worthed it based on turtle's comments

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants