The Appro Xtreme-X Supercomputer is a hybrid system featuring a combination of CPUs and GPUs in each compute node to achieve higher performance than traditional processor-only systems.
Each compute node offers 2 CPUs based on the not-yet-released Intel Xeon E5 CPUs based on the Sandy Bridge-E architecture and 4 GPUs from Nvidia?s Tesla M2090 Series.
Each node provides 2.99 TeraFlops of computing power, which according to Appro is the world's fastest GPU accelerated node for massively clustered systems. The entire supercomputer achieves a peak performance of 802 teraflops (802 trillion times per second).
"We are happy to complete installation and deployment of the Appro Xtreme-X Supercomputer based at the University of Tsukuba,? said Daniel Kim, CEO of Appro.
?This cutting-edge hybrid supercomputer system will deliver performance, scalability and manageability to help lead a new beginning of scientific research," concluded the company?s rep.
HA-PACS is a 3-year project at the CCS, University of Tsukuba, to develop a new generation of large scale GPU clusters to be used as a test bed study to develop "tightly coupled parallel computing acceleration mechanism," to allow direct communication among the GPUs.
The plan is to develop exascale computing technology to support future computational science.
"The Center for Computational Sciences at University of Tsukuba is pleased to have access to the Appro Xtreme-X Supercomputer to support the HA-PACS project.
?This deployment will put us on the fast path towards Exascale computing," said Taisuke Boku, deputy director of Center for Computational Sciences at University of Tsukuba. The new supercomputer went live in early February.
detroit weather imessage imessage sukkot sukkot aziz ansari aziz ansari
No comments:
Post a Comment