IUBio

Local BLAST server - any experience on dual/quad Intel - Linux platforms

Doo Suk Yang microbe at CHOLLIAN.NET
Fri Mar 12 04:12:17 EST 1999


What about using SGI machine if affordable?
SGI announced parallelized version of BLAST for their machine.
I have no idea how fast it works but just saw its release note.
Check its home page
www.sgi.com/chembio

If you set it up, spread wide how fast it is.

On Thu, 11 Mar 1999, Andrew Martin wrote:

> Harry Mangalam (mangalam at uci.edu) wrote:
> : I'm considering using a dual Pentium II/Celeron, unless I can get
> : feedback that BLAST does not work well on such a system, in which case
> : we would run either a uni-PII/Celeron or if more $ can be scraped up, an
> : Alpha (but having experience with Linux/Alpha, I know some of the
> : problems with that approach).
> 
> Don't bother with Alpha for BLAST - price/performance is not up to it.
> A 533MHz Alpha box costs about 50% more than an Intel PentiumII-400 and 
> gives a 20% speed increase for BLAST. Alpha shines where there is lots of
> floating point. There also seem to be some problems in the maths 
> libraries with over/underflow causing BLAST to core dump or stop if
> the E-value is extremely significant (e.g. when a sequence hits
> itself). Compaq have now released their Digital Unix maths libraries
> for Linux, so this problem may be solved by using those.
> 
> I can't really comment on using an SMP machine, but my impression is
> that only parts of BLAST are threaded, so you will get far from a 2x
> speedup - probably a good option if you want to run 2 BLASTs at once
> though since you don't have the overhead of disks etc. (providing you
> have enough memory to fit the DB in memory twice).
> 
> We've also done some tests on different disks under Linux and don't
> see any significant speedup by going from UDMA to UW-SCSI or U2W-SCSI.
> 
> 
> Andrew
> 
> 
> 

===============================================================================
Science is the game we play with God to find out what his rules are.
-------------------------------------------------------------------------------
Doo Suk Yang                            Biotech Research Institute I
Research Scientist                      LG Chemical Ltd. Research Park
                                        104-1, Moonji-dong, Yusung-ku,
Voice: 82-42-866-2222                   Taejun, 305-390
FAX:   82-42-861-2566                   KOREA
-------------------------------------------------------------------------------
A fool hath no delight in understanding, but that his heart may discover itself.
(Proverbs 18:2)
===============================================================================





More information about the Bio-soft mailing list

Send comments to us at biosci-help [At] net.bio.net