On Sat, Aug 02, 2008 at 08:54:07AM -0700, Alan Irwin wrote: > On 2008-08-02 10:09+0100 Andrew Ross wrote: > > >C and f95 versions of example 21 give identical results here for me. > >Either this is a 64-bit issue, or it is a difference between different > >versions of gfortran (I'm using 4.2.3). Can you test further or narrow > >down where the differences are? > > Hi Andrew: > > The numerical differences are limited, but where they occur the differences > are substantial as you can see from the attached diff file which I generated > with > > diff -au x21c.psc x21f95.psc > /tmp/x21.diff > > It appears some PostScript operations are done in entirely different order > between the f95 and C version. So I lean toward your second hypothesis > (gfortran version or maybe even C version) to explain these differences. > The f77 example version is identical to the f95 example version. (The f77 > and f95 interfaces and examples were all compiled by gfortran). My gfortran > and gcc version (4.3.1-2) are substantially later than yours. > > My knowledge of PostScript code is insufficient to evaluate the practical > effect of these substantial numerical and ordering differences, but maybe > most of it is only a formatting change? Visual inspection of the results > with the gv viewer shows only minor differences (a center contour coming and > going for some of the subplots of page 3 was all I could spot by eye). I am > also attaching the complete C and f95 compressed results since it is > possible that both of my C and f95 results might be different than yours. > In any case, that wider PostScript context given by the whole file might be > able to help you narrow down where the problem is occurring.
Alan, A first initial look at the postscript shows differences even on the first page, suggesting perhaps a slightly different random distribution of numbers to start with. I've made a few of the constants in the example explicitly of kind plflt, including the max / min values for the range. This might make a difference. Can you check? If I can elimiate these differences it might make a difference to the subsequent pages. If not, at least I have narrowed down the options. Andrew ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
