gpu_benchmarking.md

1. Nanopore basecalling with Dorado

https://github.com/nanoporetech/dorado

Setup

• dorado: https://cdn.oxfordnanoportal.com/software/analysis/dorado-0.4.3-linux-x64.tar.gz
• Dataset:

  https://dropoff.nbi.ac.uk/pickup?claimID=HejmBN7XJVbMRv74&claimPasscode=mxpYSGxyhKCmJG5z&emailAddr=1017
  Claim ID:	HejmBN7XJVbMRv74
  Claim Passcode:	mxpYSGxyhKCmJG5z
  

Running

We'd like to test simplex basecalling with three models https://github.com/nanoporetech/dorado#simplex-basecalling

dna_r10.4.1_e8.2_400bps_fast@v4.2.0
dna_r10.4.1_e8.2_400bps_hac@v4.2.0
dna_r10.4.1_e8.2_400bps_sup@v4.2.0

#!/sur/bin/env bash

DORADO=/home/boston/dorado-0.4.3-linux-x64/bin/dorado
MODELS=("dna_r10.4.1_e8.2_400bps_fast@v4.2.0" "dna_r10.4.1_e8.2_400bps_hac@v4.2.0" "dna_r10.4.1_e8.2_400bps_sup@v4.2.0")
DATA=/home/boston/benchmarking/
for model in "${MODELS[@]}";
do
echo "Running $model"
$DORADO basecaller -v models/$model $DATA 2>> $model.log > $model.bam
done

Metrics

• Runtime and samples/s which are reported when dorado completes successfully.
• GPU memory usage

2. ai-benchmarking

https://pypi.org/project/ai-benchmark/