SUPERCOMPUTING' 2011, SEATTLE, USA: Shedding light on a new paradigm in high performance computing (HPC), Texas Instruments Inc. (TI) is offering developers the industry's highest performing multicore digital signal processors (DSP) at the lowest power levels, based on its TMS320C66x DSP generation.
TI's TMS320C6678 and TMS320TCI6609 multicore DSPs are well suited for computing applications in oil and gas exploration, financial modeling and molecular dynamics, where ultra-high performance, low power and easy programmability are critical requirements. TI provides free optimized libraries for HPC that make it easier to achieve maximum performance without spending time optimizing code, and supports standard programming languages such as C and OpenMP so developers can easily migrate their applications to take advantage of the power savings and performance.
"TI's new family of multicore DSPs deliver outstanding flops per watt performance and an exceptional level of density and integration," commented Phillip J. Mucci, founder and director of business development, Samara Technology Group. "This fact, combined with a number of options for high speed, low latency, socket to socket interconnect, makes TI's DSP an ideal building block for the high performance, high efficiency HPC systems of tomorrow."
With the highest performing floating point DSP core at 16 GFLOPs/W, TI's C66x KeyStone-based multicore DSPs are changing the way HPC developers meet requirements for performance, power efficiency and ease of use. Advantech, a global manufacturer of telecom computing blades and multicore processor platforms, has developed the DSPC-8681 multimedia processing engine (MPE), a half-length PCIe card with more than 500 GFLOPs of performance at an extreme low power consumption of 50W.
In addition to the currently available PCIe card, TI and Advantech will soon have full length cards providing one and two teraflops of performance, transforming the industry with faster and more efficient solutions for HPC applications. TI's optimized math and imaging libraries, as well as standard programming model, make it easy for HPC developers to quickly achieve maximum performance.
Monday, November 14, 2011
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.