Cray, one of the most well known names in the supercomputing industry have decided to move into the GPU sector with the release of the XK6 machine, a successor to the XE6 AMD Opteron based machine which was announced a year ago.
Cray say that the XK6 machines will be available in Q3-Q4 of this year and Barry Bolding, vice president of Cray’s product division said it was unique in the GPU computing sector.
HPC wire said that the XK6 is not different from their CPU machines, being based around a blade configuration with four of the eight AMD Opteron sockets and four Nvidia Tesla GPU modules. Each of the four node blades is made up from two Gemini interconnect processors, four Opteron CPU’s and four Nvidia Tesla 20 series.
Cray have used the X2090 Tesla which is a compact version of the more common M2090 module. It has a 665 gigaflop GPU (double precision), 6GB of GDDR5 and 178GB/s of memory bandwidth. A single blade machine will deliver 70 teraflops.
An X2090 is said to be four to five times as expensive as a top of the line Opteron but since a 20-series Tesla delivers about 10 times the raw floating point performance at only about twice the power consumption, a GPU solution makes sense as long as the codes can extract those extra FLOPS.
The XK6 will use the XE6’s software with GPU specific libraries and tools such as Nvidia’s CUDA SDK. Cray are also developing their own OpenMP based compiler which is going to be targeted for GPU acceleration, which is a higher level programming model than CUDA.
Kitguru says: enough horsepower for you?