SGI has unveiled Altix UV, which is targeted at high-end supercomputing, very large-scale databases and data analytic environments.
According to SGI, based on open standards, the system’s x86 architecture leverages quad, six or eight-core Intel Xeon processors, codenamed Nehalem-EX. It allows for the use of completely unmodified Novell SUSE or Red Hat Linux operating systems. It is suitable for open source, custom and commercial applications, ranging from technical computing applications like Ansys Fluent to enterprise applications like Oracle.
The company makes deployment and management easy by integrating SGI ISLE and supporting third party and open source tools for performance and management optimisation. Using ISLE, customers are able to manage compute environments of varied capabilities from a single interface and schedule applications and operating systems’ images among the elements.
The company claims that the new offering will generate and manage huge amounts of data. This data can be archived and retrieved on the SGI InfiniteStorage family of RAID, EBOD, SAN and NAS hardware products through the total control suite, which includes DMF, CXFS and LiveSAN.
Content from our partners
Altix UV is equally capable of running shared memory and distributed memory applications, the Altix UV offers particularly high performance when it comes to I/O-bound and memory-bound applications such as in-memory and very large databases, SGI said.
Mark Barrenechea, president and CEO of SGI, said: “Altix UV is expanding the definition of supercomputing to include HPC applications, very large-scale databases and data analytic environments. Altix UV will allow our customers to think differently and solve problems they never thought solvable.”
The new offering is available in two models, Altix UV 1000, which ships as a fully integrated cabinet-level offering with up to 256 sockets and 16TB of shared memory in four racks for up to 18.6 Tflops of performance; and Altix UV 100 for mid-range market based on 19-inch rackmount 3U form factor. It scales to 96 sockets and 6TB of shared memory in two racks for up to 7 Tflops of performance.
