Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Improving Hash Join Performance through Prefetching Shimin Chen Anastassia Ailamaki Phillip B. Gibbons
 

Summary: Improving Hash Join Performance through Prefetching
Shimin Chen Anastassia Ailamaki Phillip B. Gibbons

Todd C. Mowry
Carnegie Mellon University

Intel Research Pittsburgh¡
chensm,natassa,tcm¢ @cs.cmu.edu phillip.b.gibbons@intel.com
Abstract
Hash join algorithms suffer from extensive CPU cache
stalls. This paper shows that the standard hash join algo-
rithm for disk-oriented databases (i.e. GRACE) spends over
73% of its user time stalled on CPU cache misses, and ex-
plores the use of prefetching to improve its cache perfor-
mance. Applying prefetching to hash joins is complicated
by the data dependencies, multiple code paths, and inher-
ent randomness of hashing. We present two techniques,
group prefetching and software-pipelined prefetching, that
overcome these complications. These schemes achieve 2.0­
2.9X speedups for the join phase and 1.4­2.6X speedups for

  

Source: Ailamaki, Anastassia - School of Computer Science, Carnegie Mellon University
Mowry, Todd C. - School of Computer Science, Carnegie Mellon University

 

Collections: Computer Technologies and Information Sciences