1 / 4

Blackcomb: Hardware-Software Co-design for Non-Volatile Memory in Exascale Systems

Blackcomb: Hardware-Software Co-design for Non-Volatile Memory in Exascale Systems. Principal Investigators Jeffrey Vetter, ORNL Robert Schreiber, HP Labs Trevor Mudge, University of Michigan Yuan Xie, Penn State University. Overarching Exascale Motivation.

jirair
Download Presentation

Blackcomb: Hardware-Software Co-design for Non-Volatile Memory in Exascale Systems

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Blackcomb: Hardware-Software Co-design for Non-Volatile Memory in Exascale Systems Principal InvestigatorsJeffrey Vetter, ORNLRobert Schreiber, HP LabsTrevor Mudge, University of Michigan Yuan Xie, Penn State University

  2. Overarching Exascale Motivation • Small memory capacity has profound impact on other features • ‘Feeding the beast’ • Poor efficiencies • Small messages, I/O

  3. Blackcomb Strategy • Energy efficiency: NV memories have zero standby power • Reliability: MRAM/PCRAM are resilient to soft errors • Performance: NVM will be much faster than magnetic disk • Identify and evaluate the most promising non-volatile memory (NVM) technologies • Explore assembly of NVM technologies into a storage and memory stack • Propose an exascale HPC system architecture that builds on our new memory architecture • Build the abstractions and interfaces that allow software to exploit NVM to its best advantage • Programming support for application-unaware resilience • Characterize key DOE applications and investigate how they can benefit from these new technologies

  4. Opportunities go far beyond a plugin replacement for disk drives… • New distributed computer architectures that address exascale resilience, energy, and performance requirements • replace mechanical-disk-based data-stores with energy-efficient non-volatile memories • explore opportunities for NVM memory, from plug-compatible replacement (like the NV DIMM, below) to radical, new data-centric compute hierarchy (nanostores) • place low power compute cores close to the data store • reduce number of levels in the memory hierarchy • Adapt existing software systems to exploit this new capabilities

More Related