Large >14Gb target genome vs >200Mb query contigs fail

[Ural-Yunusbaev]

Hey there,

I am trying to align >14Gb allopolyploid wheat genome assembly contigs (N50>30Mb, longest>270Mb) to a >14Gb reference (longest chromosome>850Mb) with the following parameters:
minimap2 -x asm5 -t 16 -I 16G reference.fasta contigs.fasta > asm.paf
in a node with 400Gb memory
and after 4 hours getting an error
RuntimeError: Failed : minimap2 -x asm5 -t 8 -I 16G reference.fasta contigs.fasta > asm.paf

slurm's seff output:
Nodes: 1
Cores per node: 16
CPU Utilized: 09:11:30
CPU Efficiency: 14.96% of 2-13:26:08 core-walltime
Job Wall-clock time: 03:50:23
Memory Utilized: 278.09 GB
Memory Efficiency: 69.52% of 400.00 GB

When I broke apart all the contigs longer than 75Mb and it reran, the same minimap2 command worked perfectly.
What is the reason? How can I go with contigs longer than 75Mb?
Thank you.