Cypress CSC-1200T Podręcznik Użytkownika Strona 95

  • Pobierz
  • Dodaj do moich podręczników
  • Drukuj
  • Strona
    / 124
  • Spis treści
  • BOOKMARKI
  • Oceniono. / 5. Na podstawie oceny klientów
Przeglądanie stron 94
Chapter 9. Programming tools 95
error, and it can be used to determine the cause of the problem.
The TotalView debugger can be used to examine core files. Start the
debugger with the executable name (prog.x in the example) and the
core file name:
totalview prog.x core
After this the debugger shows where each process has been stopped
and you can examine the values of the variables. Some of the processes
may have been stopped in an assembly routine. To get back to a user
subroutine, select the subroutine name in the call sequence pane located
at the upper left corner.
More information on the TotalView debugger can be obtained with the
command man totalview, from the help system of the TotalView pro-
gram itself or from the manual Introducing the Cray TotalView Debug-
ger [Crae], available online at the WWW address http://www.csc.fi:
8080.
9.4 Obtaining performance information
You can use the MPP Apprentice software to inspect possible perfor-
mance bottlenecks in your program. There is also a text-based command
appview for viewing the same profiling data. The Performance Analysis
Tool (PAT) is a low-overhead profiling tool available on the T3E.
9.4.1 MPP Apprentice
MPP Apprentice is a postexecution performance analysis tool that helps
you to locate bottlenecks on Fortran 90, C, C++ and HPF programs on the
Cray T3E. It can be applied to both single processor and multiprocessor
programs. MPP Apprentice reports, for example, time statistics summed
across all processing elements for the whole program, as well as each
DO loop, conditional clause or other statement groups in a program.
It shows the total execution time, synchronization and communication
time, the time to execute a subroutine and the number of instructions
executed. It does not record time-stamped events, but collects summary
information of the program execution.
MPP Apprentice works for both optimized and non-optimized code. It
offers suggestions for improving performance and how to get rid of the
bottlenecks. Apprentice works under the X Window System.
You have to use specific compiler options to generate a compiler infor-
mation file (CIF) at compile time and a run-time information file (RIF)
Przeglądanie stron 94
1 2 ... 90 91 92 93 94 95 96 97 98 99 100 ... 123 124

Komentarze do niniejszej Instrukcji

Brak uwag