1 / 5

Efficient Cache Coherency for Multicore Processors: The Scalable Commutativity Rule

This paper discusses a scalable approach to managing pages of varying sizes in a VM system using superpages to overcome TLB limitations. It introduces the MESI Coherency protocol with four cache line states and common-sense transitions. The text explores snoop-based cache behaviors and optimizations with store buffers and invalidation queues. It highlights the use of consistency requirements like store barriers and read barriers to ensure efficient cache coherence.

waddellr
Download Presentation

Efficient Cache Coherency for Multicore Processors: The Scalable Commutativity Rule

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. The Scalable Commutativity Rule:Designing Scalable Software for Multicore Processors A. T. Clements, M. F. Kaashoek, N.Zeldovich,R. T. Morris, and E. Kohler† MIT CSAIL and †Harvard

  2. Paper Highlights • Presents an general efficient mechanism to manage pages of different sizes in a VM system • Superpages • Objective is to address the limitations of extant translation lookaside buffers (TLB).

  3. MESI Coherency protocol (I) • Write-through cache and memory consistency protocol • Name refers to four possible states for cache lines • Modified (locked and dirty) • Exclusive (locked and clean) • Shared (unlocked and clean) • Invalid • Common-sense transitions between states

  4. Snooping • Any cache that hold a line in the modified statemust intercept and service all attempted read accesses to that line from other caches • Any cache that holds a line in the shared state must act on invalidate or request-for-ownership broadcasts from other caches • A cache that holds a line in the Exclusive state must snoop all read requests to that line from all other caches, and move the line to shared state if needed

  5. Optimizations • Store buffers allow caches to write into invalid lines • Invalidation queues contain invalidation requests that have not yet been acted upon • Weaken consistency and require barriers • Store barriers flush the store buffers • Read barrier flush the invalidation queue

More Related