Comments on: Bioinformatics, multi-core CPUs and grid computing: User perspective (2/4) http://tiago.org/ps/2007/07/17/bioinformatics-multi-core-cpus-and-grid-computing-user-perspective-24/ Computational Biology, Epidemiology, Infectious diseases, Open Science Mon, 02 Nov 2009 14:07:57 +0000 http://wordpress.org/?v=2.8.5 hourly 1 By: Cognitive Consonance » Blog Archive » Groovy/Scala/Ruby/Python on JVM http://tiago.org/ps/2007/07/17/bioinformatics-multi-core-cpus-and-grid-computing-user-perspective-24/comment-page-1/#comment-33444 Cognitive Consonance » Blog Archive » Groovy/Scala/Ruby/Python on JVM Thu, 27 Nov 2008 17:58:27 +0000 http://tiago.org/ps/2007/07/17/bioinformatics-multi-core-cpus-and-grid-computing-user-perspective-24/#comment-33444 [...] Bioinformatics, multi-core CPUs and grid computing: User perspective (2/4) [...] [...] Bioinformatics, multi-core CPUs and grid computing: User perspective (2/4) [...]

]]> By: Perfect Storm » Blog Archive » Groovy/Scala/Ruby/Python on JVM http://tiago.org/ps/2007/07/17/bioinformatics-multi-core-cpus-and-grid-computing-user-perspective-24/comment-page-1/#comment-4061 Perfect Storm » Blog Archive » Groovy/Scala/Ruby/Python on JVM Sun, 24 Feb 2008 20:22:30 +0000 http://tiago.org/ps/2007/07/17/bioinformatics-multi-core-cpus-and-grid-computing-user-perspective-24/#comment-4061 [...] Bioinformatics, multi-core CPUs and grid computing: User perspective (2/4) [...] [...] Bioinformatics, multi-core CPUs and grid computing: User perspective (2/4) [...]

]]> By: Bo Peng http://tiago.org/ps/2007/07/17/bioinformatics-multi-core-cpus-and-grid-computing-user-perspective-24/comment-page-1/#comment-2140 Bo Peng Thu, 13 Dec 2007 17:21:40 +0000 http://tiago.org/ps/2007/07/17/bioinformatics-multi-core-cpus-and-grid-computing-user-perspective-24/#comment-2140 As the author of simuPOP, I have basically given up the MPI version of simuPOP and I am actively researching openMP. The reasons are: 1. It is difficult to MPIify a scripting language (but doable). 2. The mating process is hard to parallelize. It is doable but the performance gain is questionable. 3. The major reason for a MPI version is for simulating huge populations, but this is not a bug problem with 64bit computers, and 4G ram as standard (highend) configuration for 32bit machines. 4. OpenMP can make better use of multi-core, mutlti-thread CPUs than MPI. We can discuss this further in the simuPOP mailing list. As the author of simuPOP, I have basically given up the MPI version of simuPOP and I am actively researching openMP. The reasons are:

1. It is difficult to MPIify a scripting language (but doable).
2. The mating process is hard to parallelize. It is doable but the performance gain is questionable.
3. The major reason for a MPI version is for simulating huge populations, but this is not a bug problem with 64bit computers, and 4G ram as standard (highend) configuration for 32bit machines.
4. OpenMP can make better use of multi-core, mutlti-thread CPUs than MPI.

We can discuss this further in the simuPOP mailing list.

]]>