The San Diego Supercomputer Center (SDSC) reports that the world's first supercomputer based on SSDs came online last week using 1,024 high-performance Intel 710 series drives. It's also using software designed to aggregate resources from multiple physical server nodes into "super nodes." This grants users immediate access to data so that they don't have to wait for the system to access specific drives.
Named Gordon (as in Flash Gordon), the supercomputer uses 300 TB of flash-based storage to run massive databases up to ten times faster than traditional memory. It will officially become a research tool in early January and have 16,384 compute cores, a theoretical peak performance of 280+ Teraflops per second, and aggregate flash memory capable of reading and writing at just over 200 GB per second.
"Think of Gordon as the world’s largest thumb drive, but with the capability to ingest about 220 movies per second from Netflix, or consume the entire catalog of about 100,000 Netflix movies – while still having room for another 200,000 titles," the SDSC muses. "That’s a lot of popcorn."
Last week the SDSC said Gordon uses 64 TB of RAM, 4 PB of disk storage, performs at 36 million IOPS (input/output operations per second), and comes with a hefty pricetag of $20 million USD paid by the National Science Foundation (NSF). Gordon is also now ranked at #48 on the official Top 500 list of the fastest supercomputers in the world, and deemed the most powerful supercomputer ever commissioned by the NSF for doing I/O and breaking the previous (2010) record of only 4.2 million IOPS.
“The era of data-intensive supercomputing begins with Gordon,” said SDSC Director Michael Norman. Every year we double the amount of information being generated, and we now are being overwhelmed by the data we are able to produce with our own computers. So it stands to reason that we needed a new kind of computer."
To read more about Gordon, head here.