Publications/Software/Resources
The PAPI Project. The Performance Application Programming Interface, portable interface to hardware counters and more.
The LLCBench Project. Parameterize your platform in terms of the performance of the memory heirarchy, floating point units and the MPI implementation.
How to call a debugger automatically from your code.
My Master's Thesis: Portable High Performance Active Messaging for PVM Otherwise known by it's other title Why Philip Mucci lives in Knoxville, Tennessee?
Slides from my talk at the National Center for Atmospheric Research about a the perils of benchmarking some related tools.
Slides from my talk at the Army Research Laboratory's annual User's Group Meeting about my research and involvement in the PET program.
Slides from my 3 day course on optimization for the Origin 2000. This course focuses on performance analysis, compiler flags and single processor optimization, with a brief discussion of MPI and Shared Memory
performance issues.
Slides about how I spent my summer.
Presentation: Slides from my talk at Rice. I'm trying to motivate them to integrate the hardware performance information into a preprocessor for F90 source files with a feedback loop and run-time library. Performance Analysis and Optimization through Run-time Simulation and Statistics
Presentation: Slides from my SOAPBOX at the recent Second International Workshop on Software Engineering and Code Design in Parallel Meteorological and Oceanographic Applications. Tell us Phil, what is the future of HPC hardware and performance?
Presentation: Tutorials on Performance Optimization
Collaboration: with NASA Ames on modelling the Information Power Grid.
Presentation: Performance Data Standard and API Early information regarding our efforts to standardize a set of definitions and a simple, machine-independent API to obtain hardware performance statistics to be used by tool writers and application engineers. The vendors agree...
Presentation: on the Performance Data Standard API
Presentation: Cache Simulator Early information on simulating caches to obtain better application performance and understanding.
Paper: Architecture Characterization of DOD MSRC Platforms This paper will appear in the DOD HPUG Proceedings.
The following three are availables as a tech report number ut-cs-98-394.
Paper: The CacheBench Report
Paper: The BLASBench Report
Paper: The MPBench Report
Software: Load Status for our Labs
Presentation: Tutorial on Parallel System Performance Evaluation.
Research: Explanation on the performance of C=A'*B on the IBM Power 2 590 processor. This was done for Cleve M. of MathWorks. Ever heard of MATLAB?
My Fast Cached Hashing package for rapid lookup based on
contiguous bits of integer keys is here.
The 2.1 Release of the ParkBench Benchmark Suite.
pvmtest12.tar.gz Version 1.2 of a test suite for PVM version 3. This package comes with a GUI that runs under Tcl/Tk. This package is severely dated and no longer supported. But seriously, why are you still using PVM, Harness or Snipe when you can use MPI?
School Address:
Ayres Hall Room 101
Department of Computer Science
University of Tennessee
Knoxville, TN 37996-1301
(423) 974 - 8493 (voice)
(423) 974 - 8296 (fax)
Home Address:
154 Taylor Rd.
Stow, MA 01775
(978) 897 - 8864
