Mail Archives: djgpp/2004/02/18/13:48:05
> From: CBFalconer <cbfalconer AT yahoo DOT com>
> Newsgroups: comp.os.msdos.djgpp
> Date: Wed, 18 Feb 2004 17:06:08 GMT
> >
> > I thought about something like that, but this could only have
> > effect on programs that trigger exceptions, no?
>
> Such as events triggered by the timers, giving time slices to
> other processes, etc.
I don't think so. Do you have examples?
A multitasking, multiprocessing OS should save and restore the state
of the FPU between context switches. IIRC, Windows indeed does that,
except in the exceptional conditions, when a program triggered an FP
exception in the x87. That's why I suggested, at the beginning of
this thread, to turn on the SIGFPE generation (it is usually masked
off by the DJGPP startup code). The OP replied that doing so didn't
make any difference, so it seems like this problem is not the reason
for the different behavior.
Hans-Bernhard also mentioned the effects on the FPU precision, but I
don't think this could be relevant unless the program in point fiddles
with the default precision of the FPU, which I think is not the case.
- Raw text -