Abstract
Performance of a computation-intensive multi-purpose CFD code PHASTA is analyzed on the NCSA Intel IA-64 Linux cluster. The capabilities of current-generation, open-source performance analysis tools available on this terascale system are demonstrated. Code profiling and hardware-performance counting tools are used to measure single-processor performance. Results pinpoint dominant but inefficient subroutines when level-3 optimization is used. Performance of these subroutines improves by compiling with level-2 optimization instead, due to reduction in total instructions. Flop rates of individual subroutines are estimated to guide further tuning. Parallel performance is addressed with performance visualization of inter-processor communication. Results reveal sporadic communication overhead in the function MPI_Waitall. This overhead constitutes about 18% of total simulation time.
Chapter PDF
Similar content being viewed by others
Keywords
- Communication Overhead
- Message Passing Interface
- Event Count
- Rensselaer Polytechnic Institute
- Performance Application Program Interface
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
References
Intel Fortran compiler user’s guide. 2002.
GNU gprof homepage. http://www.gnu.org/manual/gprof-2.9.1/gprof.html. 2003.
NCSA homepage. http://www.ncsa.uiuc.edu. 2003.
PAPI homepage. http://icl.cs.utk.edu/projects/papi.
Scientific Computation Research Center homepage. http://www.scorec.rpi.edu. 2003.
VProf homepage. http://aros.ca.sandia.gov/~cljanss/perf/vprof.
MPE libraries and Upshot homepage. http://www-unix.mcs.anl.gov/perfvis/software/viewers.
psrun homepage. http://perfsuite.ncsa.uiuc.edu/psrun.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2003 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Kwok, W.Y. (2003). Performance Analysis of PHASTA on NCSA Intel IA-64 Linux Cluster. In: Sloot, P.M.A., Abramson, D., Bogdanov, A.V., Gorbachev, Y.E., Dongarra, J.J., Zomaya, A.Y. (eds) Computational Science — ICCS 2003. ICCS 2003. Lecture Notes in Computer Science, vol 2660. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-44864-0_5
Download citation
DOI: https://doi.org/10.1007/3-540-44864-0_5
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-40197-1
Online ISBN: 978-3-540-44864-8
eBook Packages: Springer Book Archive