How to Configure a GPU Cluster to Scale with PyTorch Lightning (Part 2) | by Adrian Wälchli | PyTorch Lightning Developer Blog
![Multi-Node Multi-GPU Comprehensive Working Example for PyTorch Lightning on AzureML | by Joel Stremmel | Medium Multi-Node Multi-GPU Comprehensive Working Example for PyTorch Lightning on AzureML | by Joel Stremmel | Medium](https://miro.medium.com/max/347/1*q2Vw7zWb_7JSJPRyzILreA.png)
Multi-Node Multi-GPU Comprehensive Working Example for PyTorch Lightning on AzureML | by Joel Stremmel | Medium
![Not using the same GPU as pytorch because pytorch device id doesn't match nvidia-smi id without setting environment variable. What is a good way to select gpu_id for experiments? · Issue #2 · Not using the same GPU as pytorch because pytorch device id doesn't match nvidia-smi id without setting environment variable. What is a good way to select gpu_id for experiments? · Issue #2 ·](https://user-images.githubusercontent.com/12853718/50668165-3ddb9580-0f72-11e9-81a1-f79fca585c2b.png)
Not using the same GPU as pytorch because pytorch device id doesn't match nvidia-smi id without setting environment variable. What is a good way to select gpu_id for experiments? · Issue #2 ·
![GeForce RTX 3080 with CUDA capability sm_86 is not compatible with the current PyTorch installation. · Issue #45028 · pytorch/pytorch · GitHub GeForce RTX 3080 with CUDA capability sm_86 is not compatible with the current PyTorch installation. · Issue #45028 · pytorch/pytorch · GitHub](https://user-images.githubusercontent.com/52276191/93667640-7ca4dc80-fac2-11ea-80de-47cbdcfa9cd5.png)
GeForce RTX 3080 with CUDA capability sm_86 is not compatible with the current PyTorch installation. · Issue #45028 · pytorch/pytorch · GitHub
![Why the two GPUs on my machine have the same ID, so that Pytorch can only choose one? - PyTorch Forums Why the two GPUs on my machine have the same ID, so that Pytorch can only choose one? - PyTorch Forums](https://discuss.pytorch.org/uploads/default/original/2X/c/cdcea4552e44a10d4cdf6c14b99dd60f9f9cd203.png)
Why the two GPUs on my machine have the same ID, so that Pytorch can only choose one? - PyTorch Forums
![Not using the same GPU as pytorch because pytorch device id doesn't match nvidia-smi id without setting environment variable. What is a good way to select gpu_id for experiments? · Issue #2 · Not using the same GPU as pytorch because pytorch device id doesn't match nvidia-smi id without setting environment variable. What is a good way to select gpu_id for experiments? · Issue #2 ·](https://user-images.githubusercontent.com/12853718/50667147-d4a55380-0f6c-11e9-8baf-e3dc3adb5fe9.png)