Closed partha434 closed 12 months ago
It looks like /cluster/software/HDF5/1.10.7-iimpi-2021a/lib/libhdf5.a
that you are trying to link to was built using the Intel compiler and is using MPI which will need to be linked also.
Is there a reason you are building from source and not using the prebuilt version?
https://cdn.oxfordnanoportal.com/software/analysis/dorado-0.2.4-linux-x64.tar.gz
While using from the pre-built version, following the below command I got some error...If you can check..
/cluster/projects/nn9337k/dorado-0.2.3-linux-x64/bin/dorado basecaller ./models/dna_r9.4.1_e8_sup@v3.3 ./cluster/projects/nn9337k/pod5_files/sm1.pod5 --modified-bases-models ./models/dna_r9.4.1_e8_sup@v3.3_5mCG@v0 > sm1.sam
[2023-04-14 13:58:09.131] [info] > Creating basecall pipeline [2023-04-14 13:58:15.652] [error] toml::parse: file open error -> /dna_r9.4.1_e8_sup@v3.3/config.toml
In the first draft of your last message you had $model/dna_r9.4.1_e8_sup@v3.3
and the error message /dna_r9.4.1_e8_sup@v3.3/config.toml
shows this variable is probably unset so it's looking the fileystem root. You need to make sure the model paths you give to dorado point to the location in which you downloaded the models.
Hello Seymour, I pointed out the error. Thank you so much. Now it is resolved.
But now I am getting other error...
Empty sequence and qstring provided for read id 44ee01ff-ff15-4190-9d78-6ad4e0bd57c9
What GPU hardware are you running on? Can you post the output of nvidia-smi
Also, I see you downloaded 0.2.3 but note 0.2.4 is the latest version.
I tried with the latest 0.2.4 version, but got same error. Here is the output of nvidia-smi...
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 515.43.04 Driver Version: 515.43.04 CUDA Version: 11.7 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 Tesla P100-PCIE... Off | 00000000:39:00.0 Off | 0 |
| N/A 34C P0 32W / 250W | 7681MiB / 16384MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 32630 C ./bin/dorado 7679MiB |
+-----------------------------------------------------------------------------+
@partha434 unfortunately dorado doesn't have support for Pascal cards https://github.com/nanoporetech/dorado#platforms
Is cpu basecalling possible with dorado using more threads?
@partha434 - dorado
has no option to control the number of CPU threads to be used, dorado
always seems to use between 2 and 5 cores on average. This is really no option IMHO.
A single fast5->pod5
file (4K sequences) takes 45s on a fast GPU card and 6h20m when run in cpu-only mode using sup model(s). (Nvidia A100/40G and/or AMD EPYC CPUs).
Thank you so much @sklages I will try the code on V100 or A100 GPU servers in near future.
While running cmake --build cmake-build --config Release -j .... I got followign error. Please help. @iiSeymour