Where is Casava's Demux Multithreaded?

[lednakashim]

Thanks!

I came to the conclusion that the BCL to Fastq was Parallel in the makefile, although other steps are parallel with pthreads in addition to the make file.

This approach is not necessarily optimal, in my case the CPU is grossly underutilized, and I am considering making some source changes to correct it.

[seb567]

Most of the time, you will launch CASAVA commands using 'make -j <number of processes>.

So yes CASAVA is parallel in the Makefile layer.

I don't know if it uses OpenMP or pthread.

Sébastien

I am trying to get CASAVA setup on my fancy computer but I am having trouble verifying that the code is muddleheaded. Specifically my test files run, but I cannot see the threading for my single lane example. See http://www.biostars.org/post/show/46...hreaded/#46477

On my 32 core machine I see 3% cpu utilization (1/32 threads).

I have been exploring the source with artistically placed "grep" commands and I cannot find where the threading occurs.

1. Is this step parallel?

2. Where is the CPP associated with the process called "demultiplexBcls"

[lednakashim]

I am trying to get CASAVA setup on my fancy computer but I am having trouble verifying that the code is muddleheaded. Specifically my test files run, but I cannot see the threading for my single lane example. See http://www.biostars.org/post/show/46...hreaded/#46477

On my 32 core machine I see 3% cpu utilization (1/32 threads).

I have been exploring the source with artistically placed "grep" commands and I cannot find where the threading occurs.

1. Is this step parallel?

2. Where is the CPP associated with the process called "demultiplexBcls"