The family of GPU accelerators Tesla K20 are the fastest and more energy efficient GPU accelerators ever built. The Tesla K20X accelerator is one out of the family powering the highest-performance supercomputer – Oak Ridge Lab’s Titan supercomputer – the new champion for DOE labs beating the former supercomputer named Lawrence Livermore’s Sequoia.The Tesla K20X is based on NVIDIA Kepler compute architecture.
The new NVIDIA K20X-equipped XK7 blades from Cray is delivering 7.6 petaflops on Linpack for Oak Ridge Lab’s Titan supercomputer. K20X features 250 GB/sec of Memory bandwidth (ECC off) and 6 GB of Memory size (GDDR5). It also offers 2688 CUDA cores. Tesla K20X has the ability to provide highest computing performance in a single processor. It offers 3.95 teraflops single-precision and 1.31 teraflops double-precision peak floating point performance.
The Tesla K20X GPU accelerator is energy efficient – can deliver three times higher energy efficiency than previous-generation GPU accelerators. Because of this feature, Oak Ridge Lab’s Titan supercomputer is delivering 2,120 megaflops/watt.