mike@ames.arc.nasa.gov (Mike Smithwick) (02/04/90)
[February : National Speedbump Awareness Month] Well, dang. After waiting what seemed like 142 years, I finally got the Azteck 5.0 update. "Hot Damn!" sez I. Know I can do all those nifty things that Lettuce claimes they can do. Also with all of this optimization stuff that oughtta really speed up my software. . . So I timed a set of 3-D and planetary calculations for the 3.6 compiled version, 31 seconds. Okay, not bad. Then I install 5.0 (a real breeze BTW), recompile after making a few tweaks to the code, and rerun the thing. 38 seconds. What the . . .??? Methinks something be wrong here. Tweak some more. Recompile. 38 seconds. I twiddle with the cc options. 38 seconds. twiddle frobnicate. . . . . Twiddle, twiddle twiddle 38 seconds!! Recompiled my touched up code under 3.6. 31 seconds. The problem is A: I am totally braindead. B: Maybe 5.0 is soooo fast the longer time is due to simple Einsteinian relativistic effects. C: I never learned how to tell time. D: 5.0 is braindead and I should ask for a refund. My compiler options are : -ff -so -wa -wr -wu -pb -wn What gives? *** mike smithwick *** "Never trust a surgeon with shaving cuts. . ." [disclaimer : nope, I don't work for NASA, I take full blame for my ideas]
johnm@spudge.UUCP (John Munsch) (02/07/90)
In article <42016@ames.arc.nasa.gov> mike@ames.arc.nasa.gov (Mike Smithwick) writes: >Well, dang. After waiting what seemed like 142 years, I finally got the >Azteck 5.0 update. "Hot Damn!" sez I. Know I can do all those nifty things >that Lettuce claimes they can do. Also with all of this optimization stuff >that oughtta really speed up my software. . . > >So I timed a set of 3-D and planetary calculations for the 3.6 >compiled version, 31 seconds. Okay, not bad. > >Then I install 5.0 (a real breeze BTW), recompile after making a few tweaks >to the code, and rerun the thing. 38 seconds. [stuff deleted] >My compiler options are : -ff -so -wa -wr -wu -pb -wn This is a really simple suggestion so please don't think that I'm trying to insult your intelligence. Anyway... Manx shifted from 16 bit compiles as the default to 32 bit with the newest version. Could it be that you want the -ps option? I was able to shift my own code over to 16 bit ints thanks to the new support for prototypes (which translates to: I used to try 16 bit ints but got to see Mr. Guru a whole lot, so I ended up with 32 bit ints). John Munsch "Buck up. That's just the poison sumac talking." - The new Pogo strip
poirier@dg-rtp.dg.com (Charles Poirier) (02/07/90)
In article <42016@ames.arc.nasa.gov> mike@ames.arc.nasa.gov (Mike Smithwick) writes:
<Well, dang....
<
<So I timed a set of 3-D and planetary calculations for the 3.6
<compiled version, 31 seconds...
<
<Then I install 5.0 (a real breeze BTW), recompile after making a few tweaks
<to the code, and rerun the thing. 38 seconds....
<
<My compiler options are : -ff -so -wa -wr -wu -pb -wn
<What gives?
I'm not sure I can parse those options, but is it possible you are
using the default size for ints in both cases? As Aztec C 5.0 has
gone over to default 32 bit ints instead of 16 bit ints, this would
easily explain the loss of speed.
Cheers,
Charles Poirier
tron1@tronsbox.UUCP (HIM) (02/08/90)
>Item: 884 by *Masked* at ames.arc.nasa.gov >Author: [Mike Smithwick] > Subj: Re: Aztec 5.0, watch out. . . > Keyw: molassas >Well, dang. After waiting what seemed like 142 years, I finally got the >Azteck 5.0 update. "Hot Damn!" sez I. Know I can do all those nifty things >that Lettuce claimes they can do. Also with all of this optimization stuff hehehe --- no comment (Lattice user here) >Recompiled my touched up code under 3.6. 31 seconds. The problem is > A: I am totally braindead. > B: Maybe 5.0 is soooo fast the longer time is due to simple > Einsteinian relativistic effects. > C: I never learned how to tell time. > D: 5.0 is braindead and I should ask for a refund. > >My compiler options are : -ff -so -wa -wr -wu -pb -wn Ok... my BELIEF is that they changed the defaul INT to 32 bits for compatibility reasons, if you are not using the flag that will re-set that default to 16bit then that would be the problem (I think).. > >What gives? > *** mike smithwick *** > Buy lattice! (grin) **************************************************************************** Everything I say is Copr. 1990, except the stuff I stole from someone else and the stuff I don't want responsibility for. Kenneth J. Jamieson: Xanadu Enterprises Inc. "Professional Amiga Software" UUCP: tron1@tronsbox.UUCP BEST PATH ---> uunet!tronsbox!tron1 Sysop, Romantic Encounters BBS - (201)759-8450 / (201)759-8568 ****************************************************************************