Balint Cristian
Balint Cristian
@fzi-peccia , Tests were done by tuning a [resnet18](https://huggingface.co/frgfm/resnet18/blob/main/model.onnx) model Here is the TVM program and results after 5000 trials: [rvv-resnet18-mstune-rpc-2025Aug22.tar.gz](https://github.com/user-attachments/files/21939266/rvv-resnet18-mstune-rpc-2025Aug22.tar.gz) --- ### Tests In a rpc setup, I used...
@fzi-peccia , >``` > $ ./msch-database-tir-parse.py > Parsed #5000 records > No tensorized schedules found. > ``` > This needs investigation. Based on #18224 investigation, it seems the RVV intrinsic...
Further, investigated the corectness of the proposed tensorization kernels. The proposed ```multimvul``` does multiple dotproducts that would yield highest benefits inside RVV. All tests here needs https://github.com/apache/tvm/pull/18232 --- * Proposed...
@fzi-peccia , I dont know how to help to forward this, fell free to reuse [this working draft](https://github.com/apache/tvm/pull/18243). Thank you 🙏
@Aristide021 , Very nice work, congratulations! As a simple user who might have such interests, walking through the ```contrib``` section, could there be a simple ```README.md``` companion in this ```python/tvm/relax/backend/contrib/example_npu```...