Hi Purvesh,

clearly some bottleneck exists in the system, however it's difficult to say what's going on based on your description alone. Can you post
1) the mpiBLAST command-line used
2) the stderr output produced when running that command with the --debug argument added
3) the amount of memory consumed by mpiblast processes on the active node
4) the amount of physical memory in the active node

If (2) is > 5kb, please send it directly to me instead of posting it to the list.

At startup mpiBLAST computes the effective search space for all query sequences. In the process, a low-complexity filter is applied to the sequences which can take non-trivial compute resources. It may be helpful to disable the filter using the -F F command line argument and see whether the job runs faster.

-Aaron


Purvesh Khatri wrote:
Hi,

I am running mpiBLAST on a 7 node cluster, where each node has 4 processors (2 dual-core AMD processors). I am running mpiBLAST with -n 30 using MPICH2. I use 30 since the mpiBLAST documentation says that add 2 more processes for scheduling and output.

Before submitting my job to the cluster, I benchmarked the cluster and found that I can blast 125 sequences in less than 4 minutes. However, when I submitted a job with 40,000 mRNA sequences (with the same parameters and against the same database as benchmarks) with -n 30 I find that all of the time mpiBLAST processes are sleeping, except on the master node. On the master node all 4 processes are either running or in uninterruptible sleep mode. In the first 12 hours, after submitting the request, only 14 sequences were alinged. Also, in the last 12 hours not a single sequence has been aligned. Based on my benchmarks I was expecting at least a few thousands of sequences to have been aligned already.

Can anyone help me please?

Thank you for your help,

Purvesh Khatri

--
Purvesh Khatri
Dept. of Computer Science
Wayne State University
Detroit, MI 48202
Phone: 313 577 5070
Fax: 313 577 6868




-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language that extends applications into web and mobile media. Attend the live webcast and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
Mpiblast-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mpiblast-users



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
Mpiblast-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mpiblast-users

Reply via email to