tvm
tvm copied to clipboard
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Add a new schedule for the dense operator based on the gemm algorithm.
``` In file included from /wrkdirs/usr/ports/misc/tvm/work/tvm-0.16.0/src/meta_schedule/database/database.cc:20: In file included from /wrkdirs/usr/ports/misc/tvm/work/tvm-0.16.0/src/meta_schedule/database/../utils.h:52: /wrkdirs/usr/ports/misc/tvm/work/tvm-0.16.0/src/meta_schedule/database/../../support/base64.h:232:8: error: virtual function 'Write' has a different return type ('void') than the function it overrides (which has return...
> Hi all, based on our quarterly release schedule, next release is coming on **July**. **The next release is vital for downstream companies, especially depended on TVM, whatever software or...
 Steps to reproduce the behavior: git clone https://github.com/zkkli/I-ViT.git Just follow the TVM_benchmark. One or more operators have not been tuned. Please tune your model for better performance. Use DEBUG...
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,...
### Expected behavior I am following the instructions for AoT compilation of AI models [here](https://github.com/apache/tvm/tree/main/apps/bundle_deploy). Specifically, I built `tvm` from source with `set(USE_LLVM ON)`, `set(USE_MICRO ON)`, and `set(USE_MICRO_STANDALONE_RUNTIME ON)`. I...
Bumps [ws](https://github.com/websockets/ws) from 7.5.9 to 7.5.10. Release notes Sourced from ws's releases. 7.5.10 Bug fixes Backported e55e5106 to the 7.x release line (22c28763). Commits d962d70 [dist] 7.5.10 22c2876 [security] Fix...
Prior to the implementation of `Variant` in https://github.com/apache/tvm/pull/15672, functions that were polymorphic over an argument type would typically accept an `ObjectRef` argument, then downcast to an allowed type. This delays...
Prior to this commit, if a TIR variable was required to compute the output of `BlockBuilder.call_te`, but that TIR variable could not be inferred from the shape of any tensor...