1 / 7

Evaluation and Driving Applications for On-Chip Networks

Rajeev Balasubramaniam (Utah) Angelos Bilas (Crete) Jay Jayasimha (Intel) Rich Oehler (AMD) D K Panda (Ohio State) Fabrizio Petrini (Pacific National Labs) Drew Wingard (Sonics). Evaluation and Driving Applications for On-Chip Networks. Workloads.

tilden
Download Presentation

Evaluation and Driving Applications for On-Chip Networks

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. Rajeev Balasubramaniam (Utah) Angelos Bilas (Crete) Jay Jayasimha (Intel) Rich Oehler (AMD) D K Panda (Ohio State) Fabrizio Petrini (Pacific National Labs) Drew Wingard (Sonics) Evaluation and Driving Applications for On-Chip Networks

  2. Workloads • Environments: Traditional multi-core, SoC • Apps • Running on desktops/laptops: traditional, gaming • HPC: real time simulation, bioinformatics, financial, other traditional apps • Datacenter: TPC, server consolidation • Recognition, Mining, Synthesis • Healthcare: MRI, etc. • Embedded (SoC-type): handheld games, medical

  3. Architectural Characterization • Access patterns: memory, I/O, streaming • sharing: read/write, read only, etc. • Synchronization: fine/coarse grain, collective • QoS, need for real time guarantees, performance isolation, isolation for security • Programming models • Mixed mode (msg passing + coherence/shared memory) • programming issues with new memory hierarchies being exposed to programmer

  4. Network Requirements • Support • granularity of data transfers • synchronization (multicast, etc.) and collective operations • Partitioning: QoS (virtual channels, partitioning), performance isolation, isolation for security • Network interfaces for different protocols

  5. Evaluation Metrics • latency, b/w under different traffic classes • power / energy thermal constrained? • area, constrained • real time deadlines, QoS (esp for SoC) • standardization of metrics • Monitoring/counters under constraints

  6. Research Issues • New programming models • performance aware constructs and annotations • Limits to scalability of coherence protocols (enhanced support for barriers, multicast) • Network support for classes of traffic • New types of network services • isolation, security, partitioning, error recovery, power-aware decisions, reconfiguration, ... • Models for RT guarantees (for SoC) • Architectural benchmark suites & characterization

  7. Tools/Methodolgy • Tools and methodology • modeling • simulation: trace, execution, full-system simulation • emulation • prototyping

More Related