Philip Mucci is a computer scientist and consultant. He is
currently working as a consultant for SiCortex in addition to his permanent positions
as a Research Scientist for the Center
for Parallel Computers (PDC) at the Royal Institute of Technology
in Stockholm, Sweden and as a Research Consultant for the Innovative Computing
Laboratory led by Dr. Jack Dongarra at the University of Tennessee
. His research is focused on the performance and optimization of high-end technical compute platforms. Philip
earned his B.A. in Computer Science from The Johns Hopkins University in
1993 and he earned his M.S. in Computer Science from
University of Tennessee in 1998. Extended Curriculum Vitae
Active Projects and Research
- SiCortex, the future of high efficiency Linux-based technical compute clusters.
- Perfmon2, next generation support in the Linux kernel for advanced performance analysis.
- PerfMiner, an automated hardware performance collection system based on PAPI.
- OMII-Europe, Open Middleware Infrastructure Institute, Performance and Benchmarking (JRA4) architect.
- PAPI: Inventor and Technical Lead of this R & D 100 Award winning project.
- PapiEx/Monitor/LLCBench: Head developer and Project Lead.
- Department of Energy's SciDAC Performance
Engineering Research Institute Team Member.
Interests
-
Next-generation performance analysis for terascale systems.
- Benchmarking, analysis and optimization of parallel numerical applications and their platforms.
- End-to-end system software design for large parallel systems.
- High performance, low-latency communication substrates for technical compute clusters.
- Operating system performance: 100% resource availability, copy avoidance schemes, intelligent resource placement and noise reduction on machines of scale.
- Development of performance analysis hardware on next generation microprocessors.
-
Dynamic, adaptive, continuous, feedback-directed instrumentation systems.
Software
- PapiEx: PAPI Execute: a command line PAPI performance analysis tool, no source code required!
- Monitor: A library for trapping/catching library, process and thread initialization/creation/destruction.
- PAPI: Performance Application Programming Interface.
- Work on the Linux kernel, MPICH and MPICH2 to improve NUMA performance on the AMD Opteron Microprocessor.
- Some scripts to restore a badly corrupted EXT3 directory/partition/hard disk.
- PROCMON: A tool for tracking memory usage of a process. (papiex can do this too, but this is a bit more portable.)
- IOtrack: A library for trapping/catching/tracing file and socket I/O for performance analysis.
- LLCBench: Low-Level Architecture Characterization Benchmarks,
CacheBench, MPBench and BLASBench.
- DynaProf: a dynamic, interactive, performance analysis and instrumentation tool, no source code required!
- PVMAM: A portable and robust Active Messaging Implementation for PVM. (1997)
- FCH: A highly optimized and highly portable, hash table implementation. (1996)
- VHPM: The original hardware performance counter library, tool and kernel patch for the
IBM Power 2 series. (1996)
- PVM: The precursor to MPI. (1994-1997)
- PVMDBF: PVM Distributed Battle Framework, a discrete event simulator for time-step driven battle simulation. (1996)
- PVMTEST: PVM test suite and Tcl/Tk GUI. (1995)
Upcoming and Recent Professional Activities
- The 3rd High Performance Computing in the Arctic Conference in Tromso, Norway, December, 2007
- Supercomputing 2007 in Reno, NV, November, 2007
- International Supercomputing Conference in Dresden, Germany, June, 2007
- Workshop of the Functionality of Hardware Performance Counters at Micro-39 in Orlando, Florida, December, 2006
- Supercomputing 2006 in Tampa, Florida, November, 2006
Still can't find what you're looking for? Check my old web page.
How to get in touch
|
