Skip to content

Misc fixes to make test suite pass on Turnip#2904

Open
HansKristian-Work wants to merge 15 commits intomasterfrom
turnip-fixes
Open

Misc fixes to make test suite pass on Turnip#2904
HansKristian-Work wants to merge 15 commits intomasterfrom
turnip-fixes

Conversation

@HansKristian-Work
Copy link
Copy Markdown
Owner

Fixes some bad assumptions by test and a few real vkd3d-proton bugs were uncovered.

@HansKristian-Work HansKristian-Work force-pushed the turnip-fixes branch 3 times, most recently from c99b0bf to 904ebaf Compare April 1, 2026 12:28
Just too much work.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Makes it possible to query PDP2 separately.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Useless and causes stability issues since we literally OOM the entire
device.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
min16float does not require denorms.
Allows min16float on some Adreno hardware which does not expose denorms
somehow.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Some Adreno hardware cannot pass this without emulation.
Can't be bothered unless content relies on it.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
This is allowed in spec and Turnip hits it as expected.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Vulkan does not guarantee that behavior.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
When resolving with unusual paths, we have to ensure RESOLVE_DEST
state assumptions keep working as intended.

Fixes sync bugs with depth-stencil resolve on Turnip.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Hopefully should not come up ...

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Aim to be spec correct by default, and only if need to hit some
unfortunate edge case do we patch in incremental rendering late.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
It gets really painful to deal with this now and it doesn't really solve
any known issues.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
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.

1 participant