-
-
Notifications
You must be signed in to change notification settings - Fork 1.9k
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
rsx: Fbo fixes #5427
rsx: Fbo fixes #5427
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
best review
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
best review so far
|
X-Men Origins: Wolverine no longer needs SRM to fix flickering models. Sadly it's still needed to fix rainbow-colored effects, but that might be AMD-specific. Skate 3 is very close to playable with this PR, I think the only issues that remain are performance, very rare crashes and teammate naming. |
|
#4429 now fixed |
|
With this PR i have this error |
|
The virtualprotect failures hint that I may be over-estimating the size of some textures, something which was actually much more broken before. Let me know if you encounter any more applications that trigger this, maybe one of them will be easy to debug. |
|
F {RSX [0x037ad40]} RSX: class std::runtime_error thrown: VirtualProtect failed (00000003cd834000, 0x4001000, addr=0x3cf900000, error=0x1e7) God of War Ascension "Prison of the Damned" Demo, right before the menu. F {RSX [0x00508f8]} RSX: class std::runtime_error thrown: VirtualProtect failed (00000003cd85d000, 0x4001000, addr=0x3cf900000, error=0x1e7) Rage Demo, just before ingame |
|
I noticed a regression in Catherine, flickering in main menu and fps cut in half, I haven't had time to test it and compare it to the latest master (maybe another PR caused this) and test if WCB fix this. I can test on Friday again if no one else does. Sorry for this not very helpful report, I'll try to post logs on Friday. |
|
Just tested and no: Tekken Tag Tournament 2 still needs Strict Rendering Mode for the 2 last stages. |
|
Confirm Heavy Rain fixed |
|
Hi, Unfortunately, the game continues to crash and also emulator crash too. event viewer: Nome dell'applicazione che ha generato l'errore: rpcs3.exe, versione: 0.0.0.0, timestamp: 0x5c164d3a Regards |
|
More VirtualProtect Fatals, Yakuza 5, Dead Souls, Ishin and Kiwami now crash after boot.
Yakuza Dead Souls:
Yakuza Ishin:
Yakuza Kiwami:
|
|
Motorstorm speed regression, commit 01f7ad4 |
|
Killzone 3: Battlefield: Bad Company: |
|
In Motorstorm the 2nd and 3rd races go ingame again with this PR but I can also confirm that I'm seeing the same performance regression. |
|
A lot of the regressions reported are due to incorrect texture size approximation, or rather, it was so bad before (seriously under-estimated size) that I do not know if its a real bug or if fixing the problem correctly just uncovered other issues. A solution will be made available in the next few hours though. |
|
The regressions should be fixed now, I forgot to account for number-of-texels decoded per single row of bytes which caused overestimated size of S3TC data. Should be resolved now. |
|
Found another small error, crash may still happen. |
- Implicitly invoke a memory barrier if actively reading from an unsynchronized texture - Simplify memory transfer operations - Should allow more games to work without strict mode
- Also account for variable pitch textures (swizzled scan)
…alidation - If draw call resources consume memory that intersects with NA parts of the texture cache, we get a framebuffer test mismatch. This mismatch is false and happens because the thread has not yet reached the point of relocking the pages
- Remove the required_xxx_pitch constraint as it makes no sense. The pitch controls what can be written per line. - It is possible to have a huge surface width but only render to a small region at the beginning and have a smaller pitch than can fit the surface (NFS carbon)
- gl: Include an execution state wrapper to ensure state changes are consistent. Also removes a lot of required 'cleanup' for helper methods - texture_cache: Make execition context a mandatory field as it is required for all operations. Also removes a lot of situations where duplicate argument is added in for both fixed and vararg fields - Explicit read/write barrier for framebuffer resources depending on usage. Allows for operations like optional memory initialization before reading
- This silly typo broke the flip improvements in the GT fixes PR
- D24S8 targets have 2 aspects that are dealt with separately; Forcefully initialize the remaining data if a partial init is done. Its 'free' anyway - It seems that the stencil mask matters when clearing unlike the depth mask and color mask
- Set up partial transfers - Force clear of target before starting the transfer
- Pitch 0 makes sense if width == 1 and height == 1
- Avoid tagging and rely on read/write barriers and the dirty flag mechanism. Testing is done with a weak 8-byte memory test - Introducing new data when tagging breaks applications with race conditions where tags can overwrite flushed data
|
This PR reintroduced this issue which was actually fixed a few weeks prior: |
|
affect resident evil 5 |
|
This PR fixed some of the issues with black lines going horizontally across the screen when aiming down sight. Edit: When using Vulkan on this PR the auger shows enemies through the wall on the left half of the screen. |












Reimplements some parts of framebuffer management including:
This is the first in a series of pull requests, so gt_fixes will be made available after this one is merged as they touch some of the same common code.