Fujitsu has introduced its new Primergy CX600 server M1 a massively parallel computing platform suitable for engineering, scientific computing, big data and artificial intelligence.Β The team Primergy CX600 M1 includes Intel Xeon Phi processor designed for areas of scientific computing, engineering, big data and business analytics.

With up to eight compute nodes in 2U unit and having the 2nd generation Intel Xeon Phi processors, this server provides computing density and unprecedented performance.

“Based on this high-density server, could hold in one standard rack up to 168 compute nodes, more than 12,000 cores and 64 terabytes of memory, providing unprecedented performance of more than 500 teraflops,” says Juan Antonio Garcia, director of business development at Fujitsu HPC and Analytics.

The new server is characterized by its high density, with up to eight nodes in two units of rack, which involves six-fold increase in performance compared to standard servers. It is also compatible with existing software, for incorporation no surprises in ICT infrastructure.

According to Fujitsu, the team is optimal for markets such as universities and research centers, industry, scientific and any segment that requires the use of applications and big data analytics.

“It is important to note that in Japan already have projects underway with this platform. The server Primergy CX600 M1 is the essential element in the supercomputer of 25 Petaflop “Oakforest-PACS” Fujitsu recently awarded by the universities of Tokyo and Tsukuba in Japan. With 8,208 compute nodes, this system is expected to be operational in December this year and will become one of the world’s most powerful supercomputers, “adds Garcia.

In terms of efficiency, the new server shared infrastructure reduces power consumption thanks to central fans and power supply connection integrated into the chassis hot. This reduces consumption and facilitates the involvement of individual nodes without affecting the rest.

In addition, customers can choose a cooling direct-to-chip hot water at 45 Β° C based on the Cool-Central Liquid Cooling technology, which removes heat directly from the CPU module in a much more efficient way than through a column air. Thus, a reduction of up to half the costs of data center cooling is obtained, as indicated from Fujitsu, which entails the power to have a higher density of integration.


Please enter your comment!
Please enter your name here