optimize VP6 decoding for ARM
* Review VP6 open-source decoders
* Define test cases and ensure requirements are taking account in free test content spec
* document performance hot spot identification mechanisms
* create test-harness and keep track of performance numbers over time
* hand optimize VP6 decoders working together with upstream community
Blueprint information
- Status:
- Not started
- Approver:
- Tom Gall
- Priority:
- Undefined
- Drafter:
- Rony Nandy
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- Discussion
- Series goal:
- None
- Implementation:
-
Deferred
- Milestone target:
- None
- Started by
- Completed by
Whiteboard
[asac Nov 21]: renamed wiki to CamelCase
[asac Jan 3]: moving to 11.11 series
Work Items
Work items:
Get a a9 board to work on: TODO
GDB and Oprofiler on Cortex a9(required for profiling or look for a workaround): TODO
Comparative analysis of Google's vp6 and ffmpeg vp6 for a9: TODO
Check for feasibility of using OpenMP or Open CL for multithreading vp8: TODO
Have the basic vp6 selected running with test setup on board: TODO
Check for Performance Parameter and finalize them: TODO
Document the basic design(Code analysis and read standard): TODO
Do Multithreading in host x86 system using pure C: TODO
PORT into board and do performance measurement: TODO
Interact with Kernel and tools team for support in scheduling: TODO
Assmebly optimisation: TODO
Profile in target using oprofiler: TODO
Check for alignment issues and do data alignment if required: TODO
Check for dual issue in NEON and code using intrinscs if required: TODO
NEON assembly coding: TODO
Configuration Management and packaging: TODO
Dependency tree
* Blueprints in grey have been implemented.
