Xeon Phi first commercial Stampede supercomputer online

Last year, Intel officially walked into the Coprocessor field, launching its own accelerated computing product, Xeon Phi, to compete directly with Nvidia. Compared to Nvidia, the Xeon Phi solution is much more similar to the X86 processor, so it is easier to develop and debug in the Accelerated computing world, which is not an advantage for Nvidia Cuda.

Xeon Phi is launched at the same time, the TACC Center also specializes in the use of the accessories http://www.aliyun.com/zixun/aggregation/13840.html "> Super Computer Stampede. The entire stampede consists of 6,400 Dell PowerEdge c8220x and C8220 server cabinets, each containing 2 8 core Xeon E5 processors, 32GB of RAM, and a specially customized Xeon Phi accelerator card.

According to the latest news, Stampede has been formally launched yesterday, the current configuration contains 12800 Intel Xeon E5 processors and 6880 Intel Xeon Phi Coprocessor, the total computing power close to the 10pflops,stampede will be used for earthquake prediction, Research projects such as sea level change and brain tumor growth rate.

From an official point of view, if the computing power is really close to 10PFlops, it can be ranked third or fourth in the TOP500.

Stampede Xeon Phi is a specially customized version, compared to the General edition 60 core, 1.053GHz frequency, the set plate has 61 cores, the frequency has been upgraded to 1.1GHz, and with 8GB GDDR5 memory. Interested friends can refer to the article "Computing power is not an absolute advantage Intel Xeon phi parsing."