Rashid Jamil

Intel Trace Analyzer and Collector v7.2.1.009

Intel Trace Analyzer and Collector v7.2.1.009 | 65,71 Mb Tool Intel Trace Analyzer and Collector provide opportunities analysis and graphical representation of complex data profiling. Excellent scalability in time and with increasing number of processors. Analyze, optimize and execute high-performance applications in cluster systems based on Intel ®. Tools Intel ® Trace Analyzer and Intel ® Trace Collector will give you all the necessary information so that you can analyze and optimize the performance of MPI-cluster and quickly detect bottlenecks in the communication channels. Version 7.2 tool supports the comparison of data files, tracking, display counters and library check the accuracy of MPI-programs. Graphical user interface is available for Windows and Linux. Comparing the results of several runs the program. Wide range of options, filtering the results to facilitate analysis. Intel ® Trace Analyzer and Collector 7.2 is a powerful tool for understanding MPI application behavior, quickly finding bottlenecks, and achieving high performance for parallel cluster applications. It supports Intel ® architecture-based cluster systems, features a high degree of compatibility with current standards, and now includes trace file comparison, counter data displays, and an MPI correctness checking library. Analyze MPI performance, speed up parallel application runs, locate hotspots and bottlenecks, and compare trace files with graphics providing extensively detailed analysis and aligned timelines-all with Intel ® Trace Analyzer and Collector. Specifications: Advantages tools Intel ® Trace Analyzer and Intel ® Trace Collector Edition 7.2 Performance analysis of MPI-applications. Accelerating the performance of parallel applications. Identification of hot spots and bottlenecks code. Introduction of trace files in the form of graphs, suitable for comparison and detailed analysis of run-time flows. Support for operating systems, Linux * and Microsoft * Windows * (Windows Compute Cluster Server 2003 *, Windows XP * and Windows Server 2003 *) Intuitive graphical user interface, providing various options for viewing the trace files The high scalability and low system cost and efficient use of memory Ease of downloading executable components; rapid equipping executable MPI-application to measure the Definition of error communication exchange by using the check on the MPI-code Integrated online support Easy installation. Full documentation Full tracing of applications and / or simplified the collection of statistical data What's new? In a GUI tool for Intel Trace Analyzer added records of verification of the correct Added support for: Intel ® Compiler version 11.0 OS Microsoft * Windows Vista * and HPC Server 2008 A wide range of functions, the possibility of additional settings, increased productivity Binary means of measurements based on the PIN Analysis of behavior of the functions, processes, flows, run-time applications, clusters, nodes Various possibilities of filtering and combining the functions, processes, messages Display the performance counters in a timeline Caching files tracing reduces the system cost and memory consumption Tracing-based events for multithreaded applications developed using MPI, and without it Fault-tolerant routing Support for MPI-1, SHMEM, MPI-IO, ROMIO Checking the memory (when plugged in the library checking the correctness of MPI) Tools Intel ® Trace Collector Automated trace MPI-applications and verification of the correctness of MPI The basic routing of distributed processes (applications without the MPI-interface) and the tracing of individual processes Tracing at the level of flow, even with the completion of the emergency applications Data collection HPM (PAPI, rusage, counters, CA) Ability to adjust the settings in the trace file Disabling / enabling of certain functions Preferences Checking the memory with the help of Valgrind * Equipment binary runtime means of measurement Equipment to measure the compiler Icc / ifort / icpc-tcollect Gcc / g + +-finstrument-functions API: equipping the source of measurement (meters, functions, registered messaging and collaboration) Tools Intel ® Trace Analyzer Derivation of time scales: event, quantitative, qualitative scale counters Flexible profiles communications and collaboration Profile functions (schedule calls, call tree, the alignment in the load-load and operating) A detailed comparison of the two trassirovok Multilevel concludes the source code with a text browser Adaptive multi tegirovanie and filtering of events Hierarchical grouping and consolidation of functions or processed data Many configuration options to customize the charts Export data profiling as text and diagrams, printing Command Line Interface Checking MPI In the tools Intel Trace Analyzer and Intel Trace Collector feature checking MPI-code to detect mutual locks, data corruption or errors in the MPI-parameters, data types, buffer exchange of information, communications with the two-junction and in joint work. Since the verification is done with the program, and error messages are displayed at the time of their discovery, the process of debugging the code much faster. This function allows you to check to make a detailed analysis of the points stop using the debugger, but is a little tool for the job runs the application. The main advantage of the functions of verification tools implemented in the Intel Trace Analyzer and Intel Trace Collector, is the ability to zoom in large computer systems and detecting errors in the program with a very large number of processes. The test can be configured in such a way that it finds a mutual lock, regardless of the communication network. Using tracking data types and the completion of MPI-calls in the Intel Trace Collector tool, you can re-use requests, and communicators. (The library functions checks compiled from source code library functions of collecting data on productivity). Using the Intel Trace Analyzer tool, you can quickly deploy a stack of calls, and use debugging information to associate the command address and a string of source code, how to use pointers frames or without them. Batch processing and interactive debugging of both available from the command line or from the graphical user interface. With the help of a diagram showing the challenges of functions, and interactions between processes, identifies bottlenecks and application errors associated with incorrect order of code execution. Various screen shots to demonstrate the measurement of performance, timing diagrams and graphics performance of the parallel streams are represented here. Equipment code means for measuring and tracing Toolbox Intel Trace Analyzer and Intel Trace Collector allowed the means to equip the binary measure, virtually without changing its structure. Equipment to measure, available for both static and dynamic binary executable files, allows for the automatic verification of MPI-traffic, as well as points of start and completion functions. In addition, the possibility of tracking and recording performance parameters of parallel flows in applications developed in the languages C, C + + and Fortran. Toolbox Intel Trace Analyzer and Intel Trace Collector are suitable for working with MPI-applications, and distributed applications developed in the languages C, C + + and Fortran, without the use of technology MPI. For applications using the library of Intel ® MPI Library, is supported by tracing internal states MPI. More detail here OS: Windows 98, Me, 2000, XP and Windows Vista

Rashid Jamil

About Rashid Jamil -

Subscribe to this Blog via Email :

Thanks for comments.