Basic computers for bioinformatics, revisited

I recently started a postdoc in a lab without any real computer infrastructure so I was in need of a system with a little more juice than my laptop.  While a graduate student at the UW I did everything except for the lightest-weight bioinformatics on either a MacPro workstation (dating from 2008 but running strong) or the Hyak high-performance cluster.  For the next two years I don’t anticipate needing anything like Hyak (though it was always very reassuring to know it was there, if I needed it, which I did far more than I ever anticipated).  My immediate concern was to find a replacement for the MacPro.

In a previous post I suggested that bioinformaticians, as a group, might be a little Mac-happy.  Even if I wanted one however, I couldn’t afford a high end MacPro on my budget.  My two year fellowship came with $5,000 in research funds so I needed to find the most computer that I could for < 5K (lab supplies will have to take care of themselves.  With luck my starter population of pipette tips will breed).  After a bit of shopping I ended up purchasing a custom built workstation from Puget Systems, a small company based outside of Seattle.  For just over $4,300 I got a 12 core machine with 32 Gigs of RAM and 8.5 Tb of storage, including a solid state boot drive.  That won’t be enough RAM or storage if all my pending projects bear fruit, but the system can handle 128 Gb RAM and 32 Tb of storage if I need to expand.  For the operating system I settled on Ubuntu for ease of use.  The box is purring away now but there was enough weirdness in the beginning that I’m not sure I’d go with Ubuntu again.  Redhat might have been a better solution as I need the workstation configured like a server, not a home computer.

Just for kicks I did a quick cost-comparison against a similar MacPro.  Apple’s gone a little crazy with the current generation of MacPros; I believe you can only get a single processor (up to 12 cores) and, consistent with Apple’s design philosophy, the storage has been moved outside the central box.  I’ve been told that the new configuration vastly improves I/O speeds but I’m not tech savvy enough to know how or why.  At any rate the “equivalent” MacPro comes in at $9,198.00 and, while sleek and sexy (minus the weird external storage module), lacks the expandability of the hulking box in my office.

In case it is of interest for other junior researchers on a budget here are the parts that went into my box from Puget Systems.  It took them only 3-4 days to build the system and get it out the door.  If you’ve got the money this basic system could be configured to a pretty decent computer – I believe the motherboard can support 2 x 18 Intel Xeon cores (oh I wish!) and 128 Gb RAM.

Parts for a ~$4,300 workstation for mid-weight bioinformatics.

Parts for a ~$4,300 workstation for mid-weight bioinformatics.

 

7439 Total Views 2 Views Today
This entry was posted in Research. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

WordPress Anti Spam by WP-SpamShield