Hi,zhangmiao
I think these todos can be separated to 3 parts
1. framework enhancement, e.g. Automatic Bottleneck Detection
- algorithm enhancement, e.g. Unsupervised and semi-supervised
Classification Algorithm
- more workload type support, e.g More Platforms Support
Am I understand it properly?
More Platforms Support Refers to more hardware support.
Is it possible to share your priority plan and milestone of the todo list? e.g the target of 2020 2nd quarter, 3rd quarter, 4th quarter?
Sorry, such a detailed target plan is currently not available. Our goal is to make performance tuning more efficient. This year we plan to make the tuning function smarter.
Our current priority is making some currently known performance optimization methods integrated into the A-Tune project Some ideas in TODO will be reinvested when appropriate in the future, of course, interested developers are welcome to implement some of them in advance.
Is there any other open-source projects, papers, or ideas is used as references , or is there any conferences that you are following, when developing a-tune project ?
We have analyzed some papers about performance tuning and AI. Maybe we can put some public paper links in the readme later :)
Thanks, Zhipeng Xie