Fortran sleep function ???
in [Fortran]
|
From: user1 on 21 Dec 2008 19:17 Ron Shepard wrote: >> Is there a portable way to waste 30 milliseconds? > > I know of no standard way in fortran to pause a program for a > specified time. A nonstandard way might be something like > > call system( 'sleep 10' ) > > to pause for 10 seconds, for example; this does not work for > fractions of a second, which is the original question. You can > write a C function that accesses the standard usleep() function, > which uses units of microseconds, or the nanosleep() function, which > uses units of nanoseconds. If your compiler supports C > interoperability, then this approach is standard conforming (through > C and POSIX). There are some I/O and message passing libraries that > support sleep functionality (usually by using usleep() or > nanosleep() under the hood), so if you don't want to write your own > interface you could link with one of these libraries. > > The general advantage of this kind of approach is that the operating > system sets a timer rather than relying on the CPU spinning with > wasted work for the sleep period. Perhaps this has more to do with > esthetics than with practicalities, but I really don't like the idea > of spinning the CPU with do loops that accomplish nothing useful. > > $.02 -Ron Shepard There are practicalities involved. The method I posted will monopolize the CPU. A "call delay(10000)" in the code I posted will put 99% cpu usage on the windows taskmgr for 10 seconds. Fortran could sure use a portable sleep / usleep / nanosleep function.
From: Gary Scott on 21 Dec 2008 20:06 user1 wrote: > Ron Shepard wrote: > >>> Is there a portable way to waste 30 milliseconds? >> >> >> I know of no standard way in fortran to pause a program for a >> specified time. A nonstandard way might be something like >> >> call system( 'sleep 10' ) >> >> to pause for 10 seconds, for example; this does not work for fractions >> of a second, which is the original question. You can write a C >> function that accesses the standard usleep() function, which uses >> units of microseconds, or the nanosleep() function, which uses units >> of nanoseconds. If your compiler supports C interoperability, then >> this approach is standard conforming (through C and POSIX). There are >> some I/O and message passing libraries that support sleep >> functionality (usually by using usleep() or nanosleep() under the >> hood), so if you don't want to write your own interface you could link >> with one of these libraries. >> >> The general advantage of this kind of approach is that the operating >> system sets a timer rather than relying on the CPU spinning with >> wasted work for the sleep period. Perhaps this has more to do with >> esthetics than with practicalities, but I really don't like the idea >> of spinning the CPU with do loops that accomplish nothing useful. >> >> $.02 -Ron Shepard > > > There are practicalities involved. The method I posted will monopolize > the CPU. A "call delay(10000)" in the code I posted will put 99% cpu > usage on the windows taskmgr for 10 seconds. Fortran could sure use a > portable sleep / usleep / nanosleep function. > > yeah, but on Windows, the API for timers is minimum of 1 millisecond. If you want a faster clock, you have to add hardware. I recently purchased a discrete card with 3 1-usec timers. You can gang them together to get a larger signal range. Still, i can only read them about 7 microseconds per byte. So a 16-bit timer takes 14 microseconds on average (2Ghz emachines T4200). There was a thread about IVF's sleepqq having to do with the imprecision of the wakeup process. Even though the resolution is 1 millisecond, it has a fairly large tolerance factor. -- Gary Scott mailto:garylscott(a)sbcglobal dot net Fortran Library: http://www.fortranlib.com Support the Original G95 Project: http://www.g95.org -OR- Support the GNU GFortran Project: http://gcc.gnu.org/fortran/index.html If you want to do the impossible, don't hire an expert because he knows it can't be done. -- Henry Ford
|
Pages: 1 Prev: WatFor77 -- Re: Command-line argument capture Next: a simple overflow example |