nVidia Tesla K20 GPU

Take a GPU that is designed for computing weather modelling, seismic processing, computational physics, satellite imaging and huge data analytics. Put it in a server tower with 2 Xeon CPUs. What do you have? Your own personal supercomputer.

Based on the the world’s fastest and most efficient high performance computing architecture, nVidia Kepler, the nVIDIA Tesla GPUs are used some of the world’s fastest computers. 

The K20X model, for instance, is used in the Titan, the world’s fastest computer – as we wrote in our blog post:

Twice a year, TOP500 lists the world’s fastest computer. In November 2012, a Cray XK7 system, called Titan, achieved 17.59 Petaflop/s (quadrillions of calculations per second). Installed at Oak Ridge National Laborotary in Tennessee  Titan has 560,640 processors, including 261,632 NVIDIA K20x accelerator cores.

Essential IT Solutions builds server with nVidia Tesla K20

We are building a Tesla K20 into a server for one of our clients. It is a thing of beauty!

nVidia Tesla K20 GPU

You’re welcome to call us for more info about the Tesla K20 – ph 021 853 4790. And, yes, we just knew that you’d like one too and would like to know the price. At the current exchange rate, the nVidia Tesla K20 costs R49 999.

And this is what you pay for:

nVidia model nVidia Tesla K20
Number and Type of GPU 1 Kepler GK110
Peak double precision floating point performance 1.17 Tflops
Peak single precision floating point performance 3.52 Tflops
Memory bandwidth (ECC off) 208 GB/sec
Memory size (GDDR5) 5 GB
CUDA cores 2496

The Tesla M2070Q will run on Windows 7 but the others, including the K20, need more, shall we say, robust platforms to run on: Windows Server 2012, 2008 R2 and 2008 64-bit (all editions) or Linux 32-bit and 64-bit (RHEL 5.4 Server, Ubuntu 9.10 Server, RHEL 4.8 Server, SLES 11).


Date: 07 Mar 2013