Commit e6d00c4b authored by Papatheodore, Thomas's avatar Papatheodore, Thomas
Browse files

edited README to correct statement about CUDA_VISIBLE_DEVICES

parent 4a7e29ce
...@@ -56,7 +56,7 @@ If you pass `verbose` as a command line argument to the executable, you can see ...@@ -56,7 +56,7 @@ If you pass `verbose` as a command line argument to the executable, you can see
* For more information about the GPU compute modes and MPS, please see [https://docs.nvidia.com/deploy/pdf/CUDA_Multi_Process_Service_Overview.pdf](https://docs.nvidia.com/deploy/pdf/CUDA_Multi_Process_Service_Overview.pdf). * For more information about the GPU compute modes and MPS, please see [https://docs.nvidia.com/deploy/pdf/CUDA_Multi_Process_Service_Overview.pdf](https://docs.nvidia.com/deploy/pdf/CUDA_Multi_Process_Service_Overview.pdf).
* The `RT_GPU_id` in the output refers to the runtime device id, whereas the `GPU_id` refers to the device id that would be reported by `CUDA_VISIBLE_DEVICES`. * The `RT_GPU_id` in the output refers to the runtime device id, whereas the `GPU_id` refers to the actual device ID on the node.
* This code uses hard-coded values of the Bus IDs to map GPUs, so it will only work on Summit, Ascent, and SummitDev. You would need to modify the Bus IDs for other systems. * This code uses hard-coded values of the Bus IDs to map GPUs, so it will only work on Summit, Ascent, and SummitDev. You would need to modify the Bus IDs for other systems.
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment