From: Mitch Watrous <email@example.com>
Subject: Intel's Fortran Compiler
Date: 2 Jul 1998 02:05:09 -0400
I would like to address some questions to anyone who is using Intel's Fortran compiler with NONMEM.
Myself, I am currently installing NONMEM on a Pentium II running Windows NT.
As background, it should be noted that the NONMEM subroutine FLU is used to flush the buffer of I/O units, and this subroutine must be modified at installation time so that it calls the appropriate function for your operating system. With Digital Visual Fortran, for example, the FLU subroutine should call the run-time funtion COMMITQQ to flush the buffer.
My questions, then, are these:
(1.) Does Intel's compiler provide a function which performs the same task?
(2.) If not, what is the best way to flush buffers?
(3.) Has anyone tried calling the REWIND function?
(4.) Are there any other tips or hints I need to successfully install NONMEM on my machine?
Mitchell Watrous, Ph. D.
Resource Facility for Population Kinetics
Department of Bioengineering
University of Washington
Seattle, WA 98195-2255
Phone: (206) 543-3016
Fax: (206) 543-3081
From: firstname.lastname@example.org.EDU (ABoeckmann)
Subject: Re: Intel's Fortran Compiler
Date: 2 Jul 1998 13:49:19 -0400
Mitch Watrous asked about Intel's Fortran compiler. This is a new one to me. I looked at www.intel.com and could not find any information about it. Some clarification as to the exact name and vendor of this compiler would be helpful.
I'd also like to clarify Mitch's remarks in case anyone is wondering if he or she should have done something about FLU during installation of NONMEM.
Users Guide III, Chapt. 3, page 23-24 discusses FLU.
As distributed, it contains code that works for the Microsoft Fortran PowerStation and DIGITAL Visual Fortran compilers (CALL COMMITQQ(I)). The UNIX SETUP utility changes this to code that works on Sun Fortran compilers (CALL FLUSH(I)).
So users of these compilers need not worry about FLU.
We give suggestions for other compilers in the guide. We also say that the code in FLU may be "commented out" if nothing works for a given platform.
FLU is not necessary. It is intended to force the NONMEM report and the intermediate output from the estimation step to be written to disk in a more timely manner, so that people can monitor the output of a long NONMEM run more closely.
However, buffering by the operating system (as opposed to buffering by the FORTRAN I/O library routines) may still occur, and FLU might not be effective on a given platform.
Once the NONMEM run terminates, the buffers are written automatically, so the complete report will always be available eventually.
-- Alison Boeckmann