CPU cache management through software

[Will Sargent]

Figured some people may be interested in this: moving data in a CPU cache closer to the core executing it, through software.  It got nominated for a best paper award.

PR release: http://newsoffice.mit.edu/2015/smarter-multicore-chips-0218

Abstract: Cache hierarchies are increasingly non-uniform, so

for systems to scale efficiently, data must be close to the threads

that use it. Moreover, cache capacity is limited and contended

among threads, introducing complex capacity/latency tradeoffs.

Prior NUCA schemes have focused on managing data to reduce

access latency, but have ignored thread placement; and applying

prior NUMA thread placement schemes to NUCA is inefficient, as

capacity, not bandwidth, is the main constraint.

We present CDCS, a technique to jointly place threads and

data in multicores with distributed shared caches. We develop

novel monitoring hardware that enables fine-grained space allocation

on large caches, and data movement support to allow

frequent full-chip reconfigurations. On a 64-core system, CDCS

outperforms an S-NUCA LLC by 46% on average (up to 76%)

in weighted speedup and saves 36% of system energy. CDCS

also outperforms state-of-the-art NUCA schemes under different

thread scheduling policies.

Paper: http://people.csail.mit.edu/sanchez/papers/2015.cdcs.hpca.pdf

Slides: http://people.csail.mit.edu/sanchez/papers/2015.cdcs.hpca.slides.pdf

Some earlier work: http://people.csail.mit.edu/beckmann/publications/presentations/2013.jigsaw.pact.slides.pdf