Geeta Chauhan chauhang
chauhang
/ gist:044fc13074477faed8e7599ba1186a82
Created
January 28, 2024 00:16
torch compile inductor benchmark
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Following the steps for https://pytorch.org/docs/stable/torch.compiler_inductor_profiling.html#torchinductor-gpu-profiling | |
| =========================================== On A10g g3.xlarge AWS instance ========================================= | |
| $> TORCHINDUCTOR_UNIQUE_KERNEL_NAMES=1 TORCHINDUCTOR_BENCHMARK_KERNEL=1 python -u benchmarks/dynamo/timm_models.py --backend inductor --amp --performance --dashboard --only mixnet_l --disable-cudagraphs --training | |
| /opt/conda/envs/nllm3/lib/python3.10/site-packages/transformers/utils/generic.py:441: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead. | |
| _torch_pytree._register_pytree_node( | |
| loading model: 0it [00:03, ?it/s] | |
| cuda train mixnet_l |
chauhang
/ Open AI Gym + Jupyter
Last active
December 4, 2019 22:18
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| # Open AI Gym Render does not work with default NVidia OpenGL install. Use the steps below to install NVidia Drivers / Cuda without OpenGL | |
| # Install Pre-req | |
| sudo apt install gcc make | |
| # Download installers | |
| mkdir ~/Downloads/nvidia | |
| cd ~/Downloads/nvidia | |
| wget https://developer.nvidia.com/compute/cuda/10.1/Prod/local_installers/cuda_10.1.243_418.87.00_linux.run | |
| wget http://us.download.nvidia.com/XFree86/Linux-x86_64/440.36/NVIDIA-Linux-x86_64-440.36.run |