jcf94 opened a new pull request #6269: URL: https://github.com/apache/incubator-tvm/pull/6269
For full upstream plan, see [Ansor RFC](https://discuss.tvm.ai/t/rfc-ansor-an-auto-scheduler-for-tvm-autotvm-v2-0/7005/32). In this PR, we bring the basic Sketch search policy which is proposed in our [paper](https://arxiv.org/abs/2006.06762) on GPU ops/subgraphs. The complete search policy still waits for the feature extraction & cost model support, so this will be a basic search policy using random cost model. We'll continue to work on the upstreaming process to bring the complete auto schedule support for various workloads on CPU/GPU with some reproducible performance results, a custom sketch rule support to work like the AutoTVM, as well as the tutorials of how to play with AutoScheduler. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
