Merge lp:~vanvugt/mir/always-double-buffers into lp:mir
Status: | Merged |
---|---|
Approved by: | Daniel van Vugt |
Approved revision: | no longer in the source branch. |
Merged at revision: | 2334 |
Proposed branch: | lp:~vanvugt/mir/always-double-buffers |
Merge into: | lp:mir |
Diff against target: |
12 lines (+1/-2) 1 file modified
src/server/compositor/buffer_stream_factory.cpp (+1/-2) |
To merge this branch: | bzr merge lp:~vanvugt/mir/always-double-buffers |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Daniel van Vugt | Abstain | ||
Kevin DuBois (community) | Approve | ||
Robert Carr (community) | Approve | ||
Alan Griffiths | Approve | ||
PS Jenkins bot (community) | continuous-integration | Approve | |
Review via email: mp+250418@code.launchpad.net |
Commit message
Always use double buffers instead of triple. (LP: #1240909)
This reduces queue lag by ~50% and resource usage by ~33%. It just
wasn't safe to do till we fixed all the BufferQueue-related bugs as of release 0.11.0.
Description of the change
Tested on desktop with bypass, frame dropping, multi-monitors. The improvement is easy to notice using a mouse.
Tested on phone (vivid mako). The improvement is quite a bit harder to notice. Not just because we've only removed a third of the rendering lag (was 6 now 4 frames), but it's probably much less than a third when you account for the fact that touch screens themselves usually have a significant latency.
All works fine now. Does it work for you?
This also significantly improves resize latency of course :)