[Date Index]
[Thread Index]
[Author Index]
Re: Slow Show/Graphics in v6.0
*To*: mathgroup at smc.vnet.net
*Subject*: [mg81101] Re: Slow Show/Graphics in v6.0
*From*: Alex Shvydky <ashv at lle.rochester.edu>
*Date*: Wed, 12 Sep 2007 03:50:02 -0400 (EDT)
*References*: <200709050653.CAA26889@smc.vnet.net> <fc4iac$eh5$1@smc.vnet.net>
Thanks for the reply, Mark.
Here is my side of this whole issue.
If you are/were in business of scientific software
you /would know of the concept of a legacy code.
We, here at the lab, have our own legacy code and
we do not expect to re-optimize it every time Intel
releases their new C/FORTRAN compiler, since it
works on the same code usually faster.
We do not have either time or desire to rewrite
and reoptimize our Mathematica routines that have
been working acceptably fast for the past several years.
Now, for a student that uses Mathematica for his/her
homework (as I did myself in the past),
I would understand that it really does not matter
too much if the new "improved" version is so slow that
you should go through literally each and every line of code
trying to optimize it to make it barely useful,
since any new couple-of-lines program can be rewritten
on the hourly basic.
BTW, the Graphics optimization that you mentioned
("wrapping a large number of Points or Lines in a single head") also
speeds up the V5.2 code, such that V5.2 code is still an order of
magnitude faster than the V6.0 code:
Timing[Show[Graphics[{Hue[0.7], AbsoluteThickness[0.1], Line[
Flatten[Table[{{xar[[it, ir]], yar[[it, ir]]}, {
xar[[it, ir + 1]], yar[[it,
ir + 1]]}, {
xar[[it + 1, ir + 1]], yar[[it + 1,
ir + 1]]}, {xar[[it + 1, ir]],
yar[[it + 1, ir]]}}, {ir, 1, mr - 1}, {
it, 1, mt - 1}], 2]]}], PlotRange -> {{-1, 1}, {0,
1}}, AspectRatio -> 1/2, Axes -> True, ImageSize ->
800]]
It takes 31.5 sec in V6.0 and 1.5 sec in V5.2.
Now, here is another trivial example:
Timing[
mr = 500;
mt = 500;
xar = Table[((ir - 1.)/(mr - 1.))*Cos[Pi*((it - 1.)/(mt - 1.))], {it, 1,
mt}, {ir, 1, mr}];
yar = Table[((ir - 1.)/(mr - 1.))*Sin[Pi*((it - 1.)/(mt - 1.))], {it, 1,
mt}, {ir, 1, mr}];
]
It takes 0.56 seconds in V5.2 and 4.703 seconds in V6.0 (Why???? What
now, is it the "front-end" again?)
But, hey, I guess it is OK, isn't it? It is not like we
have anything useful to do at the moment and can readily
to spend a couple of weeks rewriting the code.
Alex
mcmcclure at unca.edu wrote:
> On Sep 8, 4:05 am, Alex Shvydky <a... at lle.rochester.edu> wrote:
>> My intention was to get the confirmation (and hopefully
>> draw WR team's attention) that THE SAME CODE runs on
>> v6.0 considerably slower than in v5.2.
>
> And from your original post:
>> It was not my intention to discuss here what should be
>> the fastest algorithm..., but rather to compare the
>> execution time for an IDENTICAL code on v6.0 and v5.2.
>
> I think you're missing a major point here. While
> executing code written for V5 in V6 might be a
> reasonable starting place for speed comparisons, it
> is not reasonable to compare the results without
> considering possible improvements that could be made
> to the V6 code.
>
> In Graphics, for example, wrapping a large number of
> Points or Lines in a single head is an important step
> to get V6 Graphics to render efficiently. This was not
> an option in V5 since it was unimportant for the V5
> engine.
>
>> Frankly, I am quite surprised that some of the forum
>> members treat these horrible slow downs as something
>> that should be expected and accept it rather than
>> voice their concern.
>
> Clearly these are things that are worth discussing and
> in fact there are some slowdowns. I'm pretty confident
> that Wolfram is working on them, as they've already
> released V6.0.1.
>
> Mark
>
>
Prev by Date:
**Re: Anomolous behaviour of Penrose Triangle Demonstration**
Next by Date:
**Re: DisplayForm Problem**
Previous by thread:
**Re: Slow Show/Graphics in v6.0**
Next by thread:
** Re: Slow Show/Graphics in v6.0**
| |