tvm
tvm copied to clipboard
apache
Open deep learning compiler stack for cpu, gpu and specialized accelerators
### Expected behavior I try to test auto_tvm on real hexagon device by using `pytest tests/python/contrib/test_hexagon/test_autotvm.py` (After all steps are followed in [this](https://github.com/apache/tvm/blob/main/tests/python/contrib/test_hexagon/README.md) but i got the error log :...
It seems there is an issue while generating c source of dense operation using cutlass. Even though the dense operation contains bias parameters, the generated c code doesn't reflect that...
Flash attention recently added support for loading from paged KV cache in https://github.com/Dao-AILab/flash-attention/commit/54e80a3829c6d2337570d01e78ebd9529c02d342. The support was added to the [Flash-Decoding kernel](https://pytorch.org/blog/flash-decoding/), which we haven't used so far. This PR lets...
In the debug_executor's `debug_get_output` function, the out parameter can be set to None, but in the C++ implementation of the function, there is no check to see if the out...
- [x] [M0] Build MSCGraph core parts. Enable translation between Relay, Relax and MSCGraph without lossing information. - [x] [M0.1] Passes for set name and layout for expressions (src/contrib/msc/transform) -...
Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion,...
> [Last release v0.14.0](https://github.com/apache/tvm/releases/tag/v0.14.0) was proposed at the end of September and the release day is 23 Oct, more detail refer [v0.14.0 release schedule](https://github.com/apache/tvm/issues/15812). It has been almost **three months**...
Hi all， I try to use the autotuning module to search for the optimal library, but an error was reported. Attached is the model file [picodet_new.onnx.zip](https://github.com/apache/tvm/files/10174174/picodet_new.onnx.zip) pool is not optimized...
1. Enable VM executor in tvmc command line script (--use-vm) 2. Enable cross-compiler option for VM executor compile, it fixed opencl compilation for adreno platform.
This change introduces: - A `TVMExtension` interface to extend TVMC with custom code. Currently its only use is to add `UMABackend`s. - A command line option `--experimental-tvm-extension ` to point...
