Basecalling on RTX 3060, dorado's ram usage never goes above 8GB~ on 12GB GPU. What are the chances the memory usage is specifically tuned for 8GB range to account for majority of consumer level GPUs out there?

I wonder if I can manually tinker to improve performance here.

#bioinformatics #basecalling #dorado

Can you imagine calling #m6A modifications from raw #RNA #nanopore reads just with #basecalling?

Let me introduce you to #m6ABasecaller

https://doi.org/10.1101/2023.11.13.566801