TPC-H Microbenchmarks

May 2013
Authors: Haicheng Wu, Greg Diamos, Jin Wang
Testbench authors: Jeff Young, Se Hoon Shon

This directory contains the TPC-H microbenchmarks designed by Haicheng Wu, Greg Diamos, and Jin Wang for the MICRO 2012 paper, “Kernel Weaver: Automatically Fusing Database Primitives for Efficient GPU Computation”. These microbenchmarks represent common patterns used by the TPC-H data warehousing benchmark suite and illustrate simple CPU and GPU versions of each pattern.

The original benchmarks were cleaned up and updated by Jeff and Se Hoon to allow for running via a function call rather than the original standalone test for each benchmark. This version of each microbenchmark is also used as part of the Oncilla TPC-H testbench for large input sizes.

You can find the source code to the primitives at the CASL repository. The current available version is 0.6564 (numbered by svn revision).