storage

storage copied to clipboard

adding hack to add git hashing to report

2

In the Storage WG meeting I made a suggestion for having the benchmark check the code. This is an example of how that could work. It will add a 'git'...

jaredeh

datagen example gives an error 'Invalid option --hosts'

1

The datagen example: ./benchmark.sh datagen --hosts 10.117.61.121,10.117.61.165 --workload unet3d --accelerator-type h100 --num-parallel 8 --param dataset.num_files_train=1200 --param dataset.data_folder=unet3d_data Returns an invalid --hosts error. Without the --hosts option datagen works correctly (looks...

wpower-work

Checkpointing is a critical piece of AI/ML and there are frequently systems deployed solely to hold checkpoints. I have a basic PoC for running a checkpoint here: https://github.com/wvaske/checkpoint_bench/blob/main/do_checkpoints.py What features...

wvaske

I get the following stack trace while executing datagen, but after that the datagen continues normally. It does not finish though. I left it running overnight by morning it has...

harisphnx

Single node with resnet-50 and 32 accelerators but only single log

1

I'm trying to run a single node benchmark with resnet-50 and 32 accelerators on v1.0 tag. ``` ubuntu@ip-xxx-xxx-xxx-xxx:/mnt/training_volume/benchmark/storage$ ./benchmark.sh run --hosts xxx.xxx.xxx.xxx --workload resnet50 --accelerator-type h100 --num-accelerators 32 --results-dir run2...

noctarius

Unexpected keyword argument 'image_size' when running benchmark

5

Hey folks! I saw that someone asked the same question yesterday on the mailinglist, but nobody has answered so I thought I bring it here since I'm running into the...

noctarius

Benchmark.py script for v2.0

9

Benchmark script is getting migrated from bash to python for better integration with results checking scripts. - Update to latest version of DLIO - Started updating rules document - Separate...

wvaske

We should consider sorting the definitions in section 3 to be ordered alphabetically.

dslik

Phase 1 of the power benchmarking additions involves requiring submitters to include additional documentation about their system related to power supply unit ratings and power supply unit topology. Definitions and...

dslik

The current section "1.1 Timeline" in the submission guidelines document has content from last year's benchmark. This needs to be updated to include the dates for the upcoming 2.0 benchmark.

dslik