jinsong-mao

Results 7 comments of jinsong-mao

@xuzhao9 I tried to use 4xA100-40G to avoid the OOM issue, looks torchbench.py only use one GPU's memory, I used options like --device-index or --multiprocess, both failed. do you have...

> We have tried this script on AMD GPUs and it works for LoRa and full fine tuning. We have not tried bits-n-bytes. can u share us your flow/script on...

strange to get this feedback: ![image](https://github.com/pytorch/benchmark/assets/146043398/62daebca-5142-47fe-965e-05792ded3fd8) when direct run with command: docker run -it --gpus all ghcr.io/pytorch/torchbench:latest thanks

@xuzhao9 thanks, with this workflow file, I can modify the dockerfile to access volume mounting of my user id and runner, I can run this flow smoothly now. there is...

@xuzhao9 hi xuzhao, if I want to run the test_bench.py in fp16, how should i add configs? thanks

Hi @xuzhao9 , after running the tests and checking the output json file, it's very strange that the GPU clock frequency is only 765, ![image](https://github.com/pytorch/benchmark/assets/146043398/2fc02296-57cc-4744-b6c5-d5d3a73cf231) I am running test on...

how about running test_bency.py on rocm+amd?