Microsoft Azure clients now can run workloads on Cray supercomputers - Techies Updates

Breaking News

Wednesday, October 25, 2017

Microsoft Azure clients now can run workloads on Cray supercomputers

Microsoft and Cray are joining to give Azure clients with information intentsive HPC, AI workloads access to Cray supercomputers running in select Microsoft datacenters.

The match reported on October 23 that clients in select Microsoft datacenters will have the choice to utilize Cray XC and Cray CS supercomputers joined to Cray ClusterStor stockpiling frameworks. The frameworks will be associated straightforwardly to Azure and will coordinate with Azure virtual machines, Azure Data Lake stockpiling, and additionally with Microsoft's manmade brainpower (AI) and Machine Learning administrations. 

Microsoft and Cray are focusing on clients who need to run elite registering, manmade brainpower, demonstrating reproductions, and progressed examination workloads "at supercomputing scale." This new Azure choice is for clients who might be not able buy and keep up their own particular on-premises Cray frameworks. 

This isn't the first run through Microsoft and Cray have cooperated. Last December, Microsoft and Cray joined on scaling the Microsoft Cognitive Tookit so it could keep running on Cray XC50 supercomputers. In 2005, Cray prime supporter Burton Smith joined Microsoft and remains a Technical Fellow with the organization, taking a shot at equipment, gadgets, and quantum registering. 

Microsoft likewise is chipping away at giving Azure clients access to field-programmable entryway cluster (FPGA) handling power for information concentrated undertakings like profound neural-systems administration. Microsoft is bringing its Brainwave AI structure on running on FPGAs to outside designers so as to push the limit of the sorts of AI-impacted administrations conceivable to convey in the cloud, including PC vision, regular dialect preparing, and discourse. Microsoft is expecting to make this innovation financially accessible to Azure clients in 2018.

No comments:

Post a Comment