70 likes | 200 Views
Other Tools. HPC Code Development Tools July 29, 2010 Sue Kelly smkelly@sandia.gov. Unclassified Unlimited Release R&A 5280854.
E N D
Other Tools HPC Code Development Tools July 29, 2010 Sue Kelly smkelly@sandia.gov Unclassified Unlimited Release R&A 5280854 Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company,for the United States Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
Graphical Instrumentation, Visualization and Analysis PeekPerf StandardizedData Collection Libraries HPM, … Symbolic Binary Instrumentation pSigma “Action Point” Binary Instrumentation HPC Toolkit from IBM • source code and binary instrumentation feature • a user-interface that brings all the sophisticated capabilities of binary instrumentation to the user at a high level of abstraction • the functionality to collect different dimensions of performance data • slated for sequoia and blue waters likely as HPCST
HPCToolkit from Rice • Works by sampling an execution of a multithreaded and/or multiprocess program using hardware performance counters, unwinding thread call stacks, and attributing the metric value associated with a sample event in a thread to the calling context of the thread/process in which the event occurred. • HPCToolkit supports performance analysis of fully optimized code; it even measures and attributes performance metrics to shared libraries that are dynamically loaded at run time. • deployed on BG/P (Intrepid), XT4/5 (Jaguar, Franklin) and Sun TACC (Ranger) • http://hpctoolkit.org/
Paraver • Parallel Visualization and Events Representation by Xavier Martorell at UPC Center for Parallelism of Barcelona (CEPBA) http://www.bsc.es/paraver • Designed to visualize and analyze parallel event trace files, such as those created by MPItrace for • Communication and load balance • Combined OpenMP and MPI • Hardware performance counters • Usage • Compile programs with special libraries • Run programs to produce trace files • View and analyze traces • Has been ported to BG/L, Cray XT
DDT • Distributed Debugging tool • http://www.allinea.com/?page=48 • Has been ported to Ranger (large Sun cluster) at TACC and Franklin (Cray XT4) at NERSC • supports up to 1024 processors (hello world on 220,000 MPI processes on Cray XT5 (Jaguar) • apps must compile with –g • similar functionality and look and feel as TotalView
Javelina • http://javelina-cc.sourceforge.net/ • Code coverage tool that uses dynamic instrumentation. Once a section of code has been executed, the instrumentation for that code is removed • Obtains its line number information from the compiler-generated debugging information, so it ports easily
ThreadSpotter • Acumem ThreadSpotter for optimizing single- and multi-threaded apps works with OpenMP and MPI apps • Tutorial-like documentation • http://www.acumen.com • Companion product SlowSpotter analyzes memory cache performance