Date: Mon, 22 Feb 1999 19:34:56 +0000 (GMT) From: Daniel Barker To: djgpp AT delorie DOT com Subject: Re: Does Windows 98 lose DMPI descriptors? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Reply-To: djgpp AT delorie DOT com On Sun, 21 Feb 1999, Eli Zaretskii wrote: > On Thu, 18 Feb 1999, Daniel Barker wrote: > > > The BASH script ran slowly and caused many > > page faults, presumably since it involves many program launches > > What do you mean ``presumably''? Why should it crash because of > involving ``many program launches''? Isn't this already a sign of a > clear trouble? There were no crashes during this experiment. I was not talking about invalid page faults, access violations or any kind of illegal operation. I meant page faults as monitored by Windows 98's System Monitor. I assume these occur, harmlessly but with a speed penalty, when memory is accessed at widely differing addresses. It is my experience on other systems that such page faults occur at a high rate on launching a new process. Since the shell script in question was doing little other than launch "expr" once per iteration, the rate of program launches was high. I assume this caused the high rate of occurrence of page faults. I was surprised at the high rate of page faults, 50 to 100 times the rate when a single, fairly small binary is running continuously. But I did (and do) not regard it as a sign of trouble, merely a point of general interest. > > BASH, EXPR and TRUE all were DJGPP ports of GNU programs, recently > > downloaded. > > Correction: `true' is not a program, it's a shell script. So invoking > `true' does NOT involve running a subsidiary program. Thanks for pointing this out. This means the script in fact did just over 45000 launches of DJGPP binaries, each binary being the program "expr". Daniel Barker, Institute of Cell and Molecular Biology, Swann Building, King's Buildings, Mayfield Road, Edinburgh EH9 3JR UK