Reducing Time to Value: Defining Massively Parallel Computing

Reducing Time to Value: Defining Massively Parallel Computing

  • August 2018 •
  • 8 pages •
  • Report ID: 5495917 •
  • Format: PDF
This IDC Perspective introduces IDC's concept of massively parallel compute (MPC) that is being applied as an infrastructure approach of choice across three major and fast-growing use case categories: modeling and simulation, AI, and big data and analytics. MPC occurs within accelerators, within nodes, and within clusters of nodes. As a compute paradigm, it offers the most powerful and efficient way to execute very complex instructions or to perform a relatively simple instruction on massive amounts of data."We are seeing a compute paradigm emerge due to the penetration of AI, big data and analytics, and modeling and simulation into the datacenter that heavily leverages massive parallelization within and between accelerators as well as in the form of compute clusters," said Peter Rutten, research manager at IDC's Core and Edge Infrastructure Group. "We have labeled this new paradigm 'massively parallel compute' and believe that it offers an opportunity for vendors to approach these use cases in a more streamlined and targeted fashion."
Loading...

We are very sorry, but an error occurred.
Please contact support@reportbuyer.com if the problem remains.