Mail Archives: djgpp/1997/02/06/11:09:12
Errors-To: postmaster AT ns1
Date: Thu, 6 Feb 1997 17:47:21 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
Cc: myskin AT inp DOT nsk DOT su, djgpp AT delorie DOT com
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Content-Length: 1127
On Thu, 6 Feb 1997 kagel AT quasar DOT bloomberg DOT com wrote:
> Are you sure? Did you run that small test program with and without + 0.5?
> If not, please do. I'm pretty sure it will help.
>
> Same exact results. BTW I am testing on a DG using GCC I am not home to test
> DJGPP on my Cyrix 686 so this may be the difference in internal precision
> between the x87 and M88100's MMU.
x87 is weird, and its DJGPP setup is not necessarily the most correct
one. The code should be run on a PC to be sure.
> I'm just saying we do not now enough yet to
> dismiss this guy's problem as sloppy floating point source code.
I didn't say ``sloppy''. I just wanted to convey my general experience
No value judgement intended.
that being defensive about FP computations is much more productive than
digging deep into the mysteries of x87 trying to understand who is to
blame (although there was a time when I did my share of such digging).
Right, but I do not think it will help in this case. I WILL try it at home
tonight, however, if it is an FPU or emulator issue rather than a code
generator issue or even a red herring, I may get different results since my
Cyrix 686 MAY return different results from Intel 387, 486, P5, PPro; AMD 486,
586, 686; Cyrix 486, 586 or the emulator. Just as the M88100 MMU on my DG here
returns the expected results, unlike myskin's setup.
> (BTW I agree with your recommendation in the general case, though, I
> usually use 0.505 as an error correction value.)
That's for second-graders ;-). First, people need to be convinced to add
any value at all.
Agreed.
--
Art S. Kagel, kagel AT quasar DOT bloomberg DOT com
A proverb is no proverb to you 'till life has illustrated it. -- John Keats
- Raw text -