⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 readme.orig

📁 fortran并行计算包
💻 ORIG
字号:
README File for cff_profCreated by Christopher F. Falzone <falzone@mcs.anl.gov>	   Rusty Lusk <lusk@mcs.anl.gov>	   Argonne National Laboratories------------------------------------------------------------------------------------Table of Contents1.  --------------------------------------------------  What is it?2.  --------------------------------------------------  How does it work?3.  --------------------------------------------------  How do I use it?4.  --------------------------------------------------  Sources.------------------------------------------------------------------------------------1. What is it?	This is a profiling library (see [1], [2]) designed to use with any MPI 	implementation.  The goal of this library is to validate collective call 	sematics.  Collective call semantics (argument correctness) has long been	an issue for new MPI programmers.  Errors are easy to make and hard to 	detect.  Often errors in semantics resault in deadlock, incorrect values, 	or worse!  The idea of this library is to intercept collective MPI calls 	and check for these kinds of errors and provide verbosity to what error	has been made.  This is to be done in a way that is protable, removeable,	and non-implentation-specific.------------------------------------------------------------------------------------2.  How does it work?	For every MPI call (i.e. MPI_Gather) there is an exact copy of the function	named PMPI (i.e. PMPI_Gather).  This allows the user to make his own 	MPI calls and call the PMPI call.  Allowing him to set up logging and so 	forth (see [1], [2]).  Using this idea I have created a library that can be	linked with any mpi program that intercepts the MPI calls, validates the 	semantics and makes the PMPI call with your supplied arguments or  prints 	an error message and calls the error handler.------------------------------------------------------------------------------------3.  How do I use it?	To use the profiling library with any mpi program all you have to do is link	the profile in with your programs object file at link time. (on my machine I 	use:		% mpicc myprog.c -c		% mpicc myprog.o -L/path/to/library -lcff_prof -o myprog		it may be diferent on your specific machine. consult your compiler/linker's 	documentation).	When the library is no longer needed simply compile the program as you 	normally would.------------------------------------------------------------------------------------4.  Sources.	[1] - Marc Snir, et. al. (1999) "MPI --The Complete Reference. Volume 1, The		MPI Core. Second Edition." The MIT Press, Massachusets. pp 387-394.			[2] - Willam Gropp, et. al. (1998) "MPI --The Complete Reference. Volume 2,		The MPI Extensions." The MIT Press, Massachusetts. pp 287-290.

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -