nVidia Tesla K20X graphics accelerator
nVidia Kepler is the world’s fastest and most efficient high performance computing architecture. GPU accelerators based on Kepler are used in the world’s fastest computers. Think weather modelling, seismic processing, computational physics, satellite imaging and anything else where data and huge are found in the same sentence.
The nVidia Tesla K20X, for instance, drives the current record holder, a Cray XK7 system, called Titan, that achieves 17.59 Petaflop/s (quadrillions of calculations per second).
Installed at Oak Ridge National Laboratory in Tennessee, Titan has 560,640 processors, including 261,632 NVIDIA K20x accelerator cores. It’s theoretical peak performance is more than 20 petaflops; that is, more than 20,000 trillion calculations per second. It is the absolute perfect computer to calculate the wife’s shopping budget.
Or, if your missus has already laid first claim to the $97 million it cost to build a Titan you can always consider a more affordable personal computer using a Tesla K20X. nVidia says that if you put 2 Xeon CPUs into a tower and add a K20X, you basically have your own personal supercomputer.
You can pick the Tesla K20X up for $7 700. Certainly, your wife would let you have that!
nVideo Tesla K20X specifications:
| nVidia model | nVidia Tesla K20X |
| Number and Type of GPU | 1 Kepler GK110 |
| Peak double precision floating point performance | 1.31 Tflops |
| Peak single precision floating point performance | 3.95 Tflops |
| Memory bandwidth (ECC off) | 250 GB/sec |
| Memory size (GDDR5) | 6 GB |
| CUDA cores | 2688 |
To convince the wife that you should have a Tesla K20X, show her this video:
https://youtu.be/ttezjVFvKII