Yuanjing Shi

[PyTorch] Fix pad_common for float pad_value

2

This PR intends to fix the impl of `pad_common` for float pad_value found in models like `pytorch_unet` and `timm_efficientdet`, as well as add a couple new tests for the `torch.nn.functional.pad`.

[Meta Schedule][XGBoost] Update the custom callback function of xgboost in meta schedule

8

This PR intends to update the custom callback function of xgboost in meta schedule. This change is tested against xgboost==(1.2.0, 1.5.2 & 1.6.0) to ensure backwards compatibility on `tests/python/unittest/test_meta_schedule_cost_model.py`. This...

Repro: ``` # try to run linting locally on ubuntu python tests/scripts/ci.py lint ``` Stack trace: ``` Linting the Python code with flake8... + tests/lint/flake8.sh ./_xxx/xxx.py:93:22: E999 SyntaxError: invalid syntax...

As mentioned in the title, the per store feature extractor of meta schedule seems not be able to handle TIR with one simply block of storing constant tensor. ### Actual...

`/usr/bin/ld: cannot find -lz` while installing triton from source

3

Trying to install triton from source according to the guide [here](https://triton-lang.org/master/index.html). My setup is ubuntu20.04 with cuda 11.3 and llvm-11. Complete stacktrace: ``` /home/ubuntu/triton/lib/driver/llvm.cc: In function ‘std::string triton::driver::ptx_to_cubin(const string&, const...

Performance and Prospects: MLIR-Generated CUDA vs CUTLASS/TensorRT?

5

Hi team, I formulated the question while reviewing the latest advancements in OpenAI's Triton, particularly the transition to using MLIR for replacing LLVM PTX and adding support for Nvidia Hopper....

buildSerializedNetwork failure of TensorRT 10.1 on GPU A10G/3070 - `Error Code 2: Internal Error (Assertion mConfig.caskKlibMapPtr failed. )`

2

## Description I am using the pytorch tensorrt lib to compile a simple pytorch model to tensorrt: ``` def func(x): return torch.ops.aten.clamp(x, 0, 1) ``` This works before `tensorrt==10.1.0` but...