tvm
https://github.com/apache/tvm
Python
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported3 Subscribers
Add a CodeTriage badge to tvm
Help out
- Issues
- [TIR][CUDA] Avoid cp.async for noncontiguous vector copies
- [TIR][CUDA] InjectPTXAsyncCopy may generate unaligned cp.async
- [TIR][CUDA] Require aligned addresses for cp.async
- [Bug][Relax] LegalizeOps loses dimensions for global median with keepdims
- [Fix][Relax] Preserve global median keepdims shape
- [Fix][Relax] Dispatch median sorting on GPU
- [Bug][S-TIR][Fuzz] `reverse_compute_at` on a block with no producer in the scope moves it after its consumer: reads before the write
- [Bug][Relax] `y[-3:] = 0` fails in `from_exported_program` when the dimension has fewer than 3 elements
- [Bug][Relax] `torch.cat` of tensors with different dtypes fails in `from_exported_program`: "Concat expects all input tensors to have the same dtype"
- [Bug] InferLayoutStack does not work with LayoutConvert
- Docs
- Python not yet supported