Cross-device DMA-BUF sharing for VAAPI with KMS grab #2053
+64
−19
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Just tested this out and this indeed fixes #2044 for me. |
|
I tested it and it appears to work for me as well with the exception of the connection seemingly terminating at random. Perhaps once this is rebased on the the latest release, then I can test it again to make sure that it wasn't some other part of the code base causing the problem. |
|
This is interesting. I checked out the PR, but unfortunately I could note compile it: Somehow the |
This was referenced May 5, 2024
|
tl:dr: I manually cloned the json library to give it another go. I'd just like to leave some feedback, in case this is unexpected behavior: With an AMD dGPU (RX6650XT) as card1/renderD129 and an Intel iGPU (card0/D128) cross-encoding does not work (display/render on AMD, encode on Intel). This is on a regular desktop PC, not a highly integrated dual-GPU-notebook. Here's the log in case it's helpful: |
3 tasks
|
I tried running this on my dual-GPU laptop and got a crash: GDB can't seem to produce a proper stacktrace, even with debug info enabled. This is what I've got: If I force Sunshine to use my integrated gpu with the option adapter_name = /dev/dri/renderD128, it runs, but the integrated gpu's video encoder is too weak to work smoothly and I would prefer to be able to use the nvidia GPU's encoder (through nvidia-vaapi-driver). |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
This introduces support for cross-device VAAPI encoding without having to copy each frame from the capture adapter through RAM to the encoding adapter. If the encoding device can successfully import the DMA-BUF directly from the capture device, we will pass FBs between the adapters.
In the event that the capture device doesn't support VAAPI encoding (which is the case for VKMS), we will try to find another device that can successfully import the FBs from the capture device for encoding. This allows VKMS streaming to work as long as the render device (if any) support VAAPI. Software encoding with VKMS with imported DMA-BUFs from a primary render GPU is still unsupported.
Screenshot
Issues Fixed or Closed
Fixes #2044 for GPU encoding only
Type of Change
.github/...)Checklist
Branch Updates
LizardByte requires that branches be up-to-date before merging. This means that after any PR is merged, this branch
must be updated before it can be merged. You must also
Allow edits from maintainers.