| SPECIAL REPORT
Life science companies are solving big problems with home-grown clusters, racks of blades, and gangs of PCs working overtime.
By Salvatore Salamone
March 17, 2004
| One common thread that links all bio-IT organizations, from industry titans to academic departments, is the need for huge amounts of computational power. But what is the best high-performance computing (HPC) solution to meet these extraordinary data-crunching demands?
The short answer: There is no single HPC solution that is "best" for everyone. Indeed, most life science companies rely on a mix of HPC technologies to meet their diverse computing requirements.
In this special report, Bio·IT World presents a collection of some of the best life science HPC deployments in the world. This is not a top 10 list by any means, nor does it pretend to be an all-inclusive list, or a collection of the most powerful systems — although one, the Virginia Tech supercluster, is rated the third most powerful in the world on the Top500 Supercomputer Sites list (www.top500.org).
Rather, what we present here is a sampling of cutting-edge solutions that apply HPC to critical life science problems. The intent is to show the breadth of applications that are being supported by today's incredibly wide choice in clusters, grids, and specialized hardware.
Some HPC trends are worth noting. First, there is the move from proprietary and RISC-based 64-bit processors to commodity 64-bit processors. Many vendors are now using 64-bit Intel Itanium, AMD Opteron, and Apple/IBM PowerPC chips as the cornerstone of their systems.
Some vendors, such as SGI and OctigaBay (being acquired by Cray), have married commodity processors with large-scale shared memory technology to produce systems that rival proprietary RISC machines.
A second major trend is the increased reliance on clusters. In the most recent Top500 list (released last November), clusters accounted for seven of the top 10 supercomputers, and 208 of the top 500. Virtually all the traditional systems vendors, including IBM, HP, and Dell, have cluster offerings. But smaller companies such as RLX Technologies, Linux NetworX, and Microway have also delivered some impressively powerful systems to life science organizations.
Finally, distributed computing is gaining momentum. Many life science implementations to date have been large-scale academic or institutional efforts (for example, the National Science Foundation's TeraGrid). But some companies, such as Novartis, are tapping PC-based grid capabilities within their walls