Tuesday, June 20, 2017

HPE to release greatly adaptable systems to power AI

The new HPE Apollo 6000 Gen10, HPE SGI 8600 and HPE Apollo 10 Series frameworks will be comprehensively accessible in July 2017.


Hewlett Packard Enterprise (HPE) is preparing to discharge three new High Performance Computing frameworks upgraded for hugely parallel handling and AI-related errands. 

Gone for research and national security associations that need to crunch tremendous measures of information, the new HPE Apollo 6000 Gen10, HPE SGI 8600 and HPE Apollo 10 Series frameworks will be accessible worldwide from July 2017. 

The HPE SGI 8600 is a fluid cooled, petascale framework, in view of the inheritance SGI ICE XA design, that HPE says is worked to unravel the most complex logical, building and national security challenges. 

The HPE SGI 8600 scales to more than 10,000 hubs without the requirement for extra switches and backings varieties of fluid cooled Nvidia Tesla GPU quickening agents, connected through rapid NVLink interconnects. 

The HPE Apollo 6000 Gen10 System is the most recent adaptation of HPE's stage, updated to offer more than 300 teraflops for every rack, for higher rack-scale effectiveness and better value execution. These enhancements over its ancestors has been acknowledged by diminishing dormancy and expanding IOPs execution, and also by decreasing force utilization and cooling necessities. 

The framework likewise incorporates HPE's "silicon foundation of trust", which keep servers from booting utilizing any firmware that doesn't coordinate the unique mark incorporated with the framework's incorporated lights-out chip. 

The HPE Apollo 10 Series includes frameworks streamlined for passage level Deep Learning and other AI applications, and that are intended to be moderately easy to oversee and convey. 

Inside the arrangement, the HPE Apollo sx40 System is a 1U double attachment Intel Xeon Gen10 server with bolster for up to 4 NVIDIA Tesla SXM2 GPUs with NVLink. In the interim, the HPE Apollo pc40 System is a 1U double attachment Intel Xeon Gen10 server with bolster for up to 4 PCIe GPU cards. 

Close by the new frameworks, HPE has likewise updated its Performance Software Suite - which unites a scope of utilities for provisioning, overseeing, advancing and observing elite processing frameworks. 

HPE keeps on putting resources into memory-driven processing research, with the expectation it will support endeavor applications, and a week ago was granted a US inquire about concede to proceed with deal with building up an exascale supercomputer called The Machine.



No comments:

Post a Comment