Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

GS/Vulkan: Always issue first barrier on RDNA3 #11223

Conversation

stenzek
Copy link
Member

@stenzek stenzek commented May 10, 2024

Description of Changes

So I bought a 7900GRE to debug this issue and...

It turns out not doing this causes GPU resets on RDNA3, specifically Windows drivers. Despite the layout changing enforcing the execution dependency between previous draws and the first input attachment read, it still wants the region/fragment-local barrier...

Rationale behind Changes

(Hopefully) Fixes the RDNA3 crashes.
Fixes #10720.

Suggested Testing Steps

Smoke test Vulkan renderer.
See if RDNA3 no longer crashes.

It turns out *not* doing this causes GPU resets on RDNA3, specifically
Windows drivers. Despite the layout changing enforcing the execution
dependency between previous draws and the first input attachment read,
it still wants the region/fragment-local barrier...
Completely redundant. We also don't need to use the drawlist.
@stenzek stenzek force-pushed the the-barrier-man-giveth-and-the-barrier-man-taketh-away branch from 9d3a536 to bbdda4a Compare May 10, 2024 17:08
@stenzek stenzek merged commit 20dbcfd into PCSX2:master May 12, 2024
12 checks passed
@stenzek stenzek deleted the the-barrier-man-giveth-and-the-barrier-man-taketh-away branch May 12, 2024 01:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[BUG]: Driver timeouts with RDNA3 GPUs on Windows + Vulkan
3 participants