AI has really transformed the scenario for many fields including supercomputing. This fact was recently verified by the tough competition amidst HPCs (High Performing Computing Systems) viewed at the International Supercomputing Conference (ISC) 2018 held in Frankfurt Germany. The ISC 2018 took off with the launch of Top500 list that showcased the best supercomputing platforms most of which are still relying on Intel® Xeon® processors making it one of the most preferred options in supercomputers domains. The usage of same has increased by 2.4 percent since last year. Currently, over 97% of all latest systems mentioned in Top500 list have Intel Xeon processors.
Intel Xeon processors are known for offering needed performance and flexibility to take care of most stringent HPC workloads along with the largest range of workloads over any scale needed by industry and science. For both traditional as well as supercomputing HPC clusters, Intel has been innovating and delivering high-speed interconnect technologies that permit a very cost-effective HPC systems deployment. At the 2018 ISC, Intel shared its next-generation the Omni-Path Architecture that will hit the markets in 2019.
This Omni-Path Architecture will be providing data rate speeds till 200 GB/s, which doubles up the performance of the previous generation. This gen-next fabric will be compatible and interoperable with the present day Intel OPA. Intel OPA200’s high-performance abilities, as well as low-latency over the scale, will offer system architects with an ability to scale up to thousands of nodes while taking advantage of the improved total cost of ownership.
Here the key element is visualization that opens up ways for delivery of greater insights with much rapid turnaround while making use of highly scaled data sets. Intel also announced the Intel Select Solution for Professional Visualization. It is an easily deployable system embedded with Intel’s optimized system reference framework that is built on purpose to meet the most complex present-day data requirements and explosion challenges. The select Solution also uses the onboard memory of platform for graphical rendering of larger sized datasets in real time.