Persistent segfault for unclear reasons #4
Description
Hi,
Thanks for your hard work implementing the accelerated version of KSSD. I am having a problem generating a distance matrix for around 5000 NCBI genomes as part of a research project, namely a segfault right before the distance matrix is generated.
The run command is:
rabbit_kssd alldist -i /storage/distance_method_testing/mash/rabbit-kssd/sketches/fine.assembly.list --threads 16 -L /usr/bin/shuf_file/L3K10.shuf --metric 0 -o /storage/distance_method_testing/mash/rabbit-kssd/sketches/fine.dist
The final message visible in stdout is:
===================time of computing sketches and save sketches into file is 12.9236
-----not use hash64 in index_tridist()
===================time of read index and offset sketch file is: 0.00546098
success create: /storage/distance_method_testing/mash/rabbit-kssd/sketches/fine.dir
=====total: 5000
=====finish: 0
Segmentation fault (core dumped)
Based on my limited review of dist.cpp, I would guess that the segfault occurs right at line 177 at
int tid = omp_get_thread_num();
since there are no further messages afterwards, but that is very puzzling. I tried with a single thread and got the same result as well.
System specs:
6.11.0-26-generic #26~24.04.1-Ubuntu
x86_64
64 GB ram
AMD Ryzen 7 8845HS w/ Radeon 780M Graphics
RabbitKSSD version: 2be3294
Please let me know if there is any additional information needed to help debug this issue. I have had the same problem analyzing a single genome, so I do not think the genome input is related necessarily.