Compiler technology for blue gene systems

Authors:
Stefan Kral;Markus Triska;Christoph W. Ueberhuber
Affiliations:
Institute for Analysis and Scientific Computing, Vienna University of Technology, Wien, Austria;Institute for Analysis and Scientific Computing, Vienna University of Technology, Wien, Austria;Institute for Analysis and Scientific Computing, Vienna University of Technology, Wien, Austria
Venue:
Euro-Par'06 Proceedings of the 12th international conference on Parallel Processing
Year:
2006

Citing 12
Cited 0

Constraint satisfaction in logic programming

Constraint satisfaction in logic programming
Supercompilers for parallel and vector computers

Supercompilers for parallel and vector computers
Advanced compiler design and implementation

Advanced compiler design and implementation
A fast Fourier transform compiler

Proceedings of the ACM SIGPLAN 1999 conference on Programming language design and implementation
Exploiting superword level parallelism with multimedia instruction sets

PLDI '00 Proceedings of the ACM SIGPLAN 2000 conference on Programming language design and implementation
Graph-based code selection techniques for embedded processors

ACM Transactions on Design Automation of Electronic Systems (TODAES)
Compiling for SIMD Within a Register

LCPC '98 Proceedings of the 11th International Workshop on Languages and Compilers for Parallel Computing
Large-Scale First-Principles Molecular Dynamics simulations on the BlueGene/L Platform using the Qbox code

SC '05 Proceedings of the 2005 ACM/IEEE conference on Supercomputing
A study of replacement algorithms for a virtual-storage computer

IBM Systems Journal
Blue Gene/L programming and operating environment

IBM Journal of Research and Development
Design and exploitation of a high-performance SIMD floating-point unit for Blue Gene/L

IBM Journal of Research and Development
Vectorization techniques for the Blue Gene/L double FPU

IBM Journal of Research and Development

Quantified Score

Hi-index	0.00

Visualization

Abstract

Standard compilers are incapable of fully harnessing the enormous performance potential of Blue Gene systems. To reach the leading position in the Top500 supercomputing list, IBM had to put considerable effort into coding and tuning a limited range of low-level numerical kernel routines by hand. In this paper the Vienna MAP compiler is presented, which particularly targets signal transform codes ubiquitous in compute-intensive scientific applications. Compiling Fftw code, MAP reaches as much as 80% of the optimum performance of Blue Gene systems. In an application code MAP enabled a sustained performance of 60 Tflop/s to be reached on BlueGene/L.