tvm
tvm copied to clipboard
Open deep learning compiler stack for cpu, gpu and specialized accelerators
I am encountering linker errors when trying to build my project using CMake with PyTorch. The build process fails with several undefined reference errors related to PyTorch symbols. Below are...
This PR adds fusing of the astype operation to matmul for cublas. This change is needed to improve the performance for fp8.
I was tensorizing BERT using meta schedule and get a runtime error. ### Expected behavior It should - Extract task - Tune task - Give me a tuned model to...
Fixes a naming mismatch in MSCGraph where tensor_name could formatted as 'string:index:index',and the corresponding node.name is 'string:index'. Splitting tensor_name from the right aligns it correctly. For example, the TFLite default...
This commit extends the functionality of the SME dense and matmul schedules to support operations with fp16 inputs and an fp32 output, where `transpose_a=False` and `transpose_b=True`. For convenience, it also...
### This issue is to track progress for [SME enablement](https://github.com/apache/tvm-rfcs/pull/107) upstream - [x] P0. Fixed Virtual Platform (FVP) and toolchain installation - https://github.com/apache/tvm/pull/16755 - [x] P1. Fixed Virtual Platform (FVP)...
The refactor of the unittest folder in #16110 did not include the `tests/python/disco` folder in the list of folders to run in CI.
This commit adds a unit test to verify that a single `IRModule` can contain functions that will be used on multiple distinct targets. Previously, this test case caused errors when...
This commit introduces the Particle Swarm Optimization (PSO) tuner `PSOTuner` to AutoTVM, enhancing tuning efficiency by better exploring the configuration space compared to GATuner. Key additions include: ### Major Changes:...
This commit updates TVM implementations of `dmlc::Stream`. With https://github.com/dmlc/dmlc-core/pull/686, this API now requires the `Write` method to return the number of bytes written. This change allows partial writes to be...