Hi all,
I recently ran an inference in IQ-TREE which quickly failed due to an out of memory error. This seemed odd because while it was a large alignment with many partitions, I expect IQ-TREE to be able to handle these data, and I assigned up to 30 hours for inference.
When I look at the scheduler.conf output, it seems to show the issue:
nodes=1
node_exclusive=0
mem=11G
threads_per_process=6
runhours=30
subscribed=yes
By contrast, a successful unpartitioned inference of the same alignment had node_exclusive=1; mem=243G; threads_per_process=128.
Any ideas what went wrong here? The job ID is NGBW-JOB-IQTREE_XSEDE-6DCBDFFF093D4A6C9C69CAC890BF0EEB (submitted under my advisor's account).
Thank you!
-Ziv