Signal fences on CPU to reduce delays on muxed queues #2082
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
Signed-off-by: Philip Rebohle <[email protected]>
Adding even more parameters to this would get a bit silly. Signed-off-by: Philip Rebohle <[email protected]>
doitsujin
force-pushed
the
cpu-signal
branch
2 times, most recently
from
3ea4dbf to
507210c
Compare
May reduce some signaling delays for multiplexed queues. Signed-off-by: Philip Rebohle <[email protected]>
This way, the wait/signal logic will detect an external submission and may allow pending Signal calls to skip this submission. Signed-off-by: Philip Rebohle <[email protected]>
Signed-off-by: Philip Rebohle <[email protected]>
2 tasks
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Outdated suggestions cannot be applied.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.
Suppose we have two virtual queues
AandBboth sharing the samevkd3d_queue, and the application does something like this:then actually signaling the fence may be delayed until work submitted to
Bcompletes, which can delay both app threads waiting on the fence as well as command submissions to any queues that wait on said fence. To solve this, we instead queue up a CPU-side wait on the last submission performed by the signaling command queue and signal the fence manually.This also takes the early signal idea from #2075 and queues up a CPU signal in case the virtual queue has no pending submissions.
This expands upon the per-queue submission tracker introduced in 5ac0781, which now needs to be updated for all submissions made from d3d12 scope (waits, sparse binding, 11on12 interop stuff, etc) for correctness.