On June 19 I started a basecalling & demux job on a computing cluster. I have performed exactly the same task a day before using dorado 0.5.0. I wanted to compare the results just because curiosity but it turned out that, without changing other parameters except the version of dorado, the result using 0.7.1 is just empty files. After that I changed the version parameter back and it worked as before.
I have seen the bug report #860 and I think that it's a related issue.
Hardware (1 cpu and 1 gpu used):
2 x Intel(R) Xeon(R) Gold 6130 CPU @ 2.10GHz (in total 32 cores),
4 x NVIDIA Tesla V100 GPU per node,
16 GB HBM2, 5120 CUDA cores,
192 GB DDR4 2666 MHz ECC, 240 GB SSD,
Infiniband EDR 100 Gb/s.
Source data: pod5 (doesn't matter, large or small)
average read length about 1 kb
Issue Report
On June 19 I started a basecalling & demux job on a computing cluster. I have performed exactly the same task a day before using dorado 0.5.0. I wanted to compare the results just because curiosity but it turned out that, without changing other parameters except the version of dorado, the result using 0.7.1 is just empty files. After that I changed the version parameter back and it worked as before. I have seen the bug report #860 and I think that it's a related issue.
Dorado version: 0.7.1+80da5f5 Commands:
Hardware (1 cpu and 1 gpu used): 2 x Intel(R) Xeon(R) Gold 6130 CPU @ 2.10GHz (in total 32 cores), 4 x NVIDIA Tesla V100 GPU per node, 16 GB HBM2, 5120 CUDA cores, 192 GB DDR4 2666 MHz ECC, 240 GB SSD, Infiniband EDR 100 Gb/s.
Source data: pod5 (doesn't matter, large or small) average read length about 1 kb
piece of error message:
Best regards, Vilnis