Re: Famous Mathematica 5.0 not working with Mathematica 6?
Szabolcs
> Hello. I am performing an evaluation of Mathematica 6 with the demo
> version on both Windows XP 32 bits and OS X 10.4.9. However both crash
> when evaluating the famous Karl's Mathematica benchmark:
> http://www2.staff.fh-vorarlberg.ac.at/~ku/karl/timings50.html
>
> Did anyone experience the same thing? Have you got any suggestion?
> Thank you.
>
>
I guess the plotting routines run out of memory ...
DrMajorBob
"No more memory available.
Mathematica kernel has shut down.
Try quitting other applications and then retry."
on an AMD 3200+, Mathematica 6, WinXP SP-2, with 2GB RAM.
Version 5.2 runs the same code, no problems, with this output:
Version = \[InvisibleSpace]5.2 for Microsoft Windows (June 20, 2005)
Times = \[InvisibleSpace]{4.047,3.359,4.734,7.375,7.125,4.844,3.672,
5.594,18.359,8.203,30.5,8.391,5.187,6.922,10.875}
Total Kernel Time = \[InvisibleSpace]129.187\[InvisibleSpace], Total
Time =
= \
\[InvisibleSpace]158.1
Benchmark = \[InvisibleSpace]3.65397
Bobby
On Wed, 30 May 2007 04:35:42 -0500, <eric.l...@gmail.com> wrote:
> Hello. I am performing an evaluation of Mathematica 6 with the demo
> version on both Windows XP 32 bits and OS X 10.4.9. However both crash
> when evaluating the famous Karl's Mathematica benchmark:
> http://www2.staff.fh-vorarlberg.ac.at/~ku/karl/timings50.html
>
> Did anyone experience the same thing? Have you got any suggestion?
> Thank you.
>
>
>
--
Jens-Peer Kuska
probably that is the reason for the package:
Needs["Benchmarking`"]
and the function
Benchmark[]
regards
Jens
Jean-Marc Gulliet
> Hello. I am performing an evaluation of Mathematica 6 with the demo
> version on both Windows XP 32 bits and OS X 10.4.9. However both crash
> when evaluating the famous Karl's Mathematica benchmark:
> http://www2.staff.fh-vorarlberg.ac.at/~ku/karl/timings50.html
>
> Did anyone experience the same thing? Have you got any suggestion?
> Thank you.
Hi Eric,
I have just made some tests on my system (Windows XP SP2, Pentium IV
HT2.6 GHz, 1 Go RAM) with 5.2 ans 6.0. As expected, Karl's notebook
works fine with 5.2 (using less than 100 Mo of RAM); however, testing
6.0 results in consuming all of the available memory (I killed the
kernel after it had eating up 1.8 Go).
Regards,
Jean-Marc
Jean-Marc Gulliet
> Hello. I am performing an evaluation of Mathematica 6 with the demo
> version on both Windows XP 32 bits and OS X 10.4.9. However both crash
> when evaluating the famous Karl's Mathematica benchmark:
> http://www2.staff.fh-vorarlberg.ac.at/~ku/karl/timings50.html
>
> Did anyone experience the same thing? Have you got any suggestion?
> Thank you.
Hi,
I have done some more testing, and the culprit seems to be the graphic
engine. Having commented out the following two expressions from Karl's
original notebook,
Timing[Plot3D[Sin[x y], {x, 0, 2 \[Pi]}, {y, 0, 2 \[Pi]},
PlotPoints -> 4000, DisplayFunction -> Identity]][[1]],
Timing[Plot3D[Evaluate[q4[x, t]], {x, -40, 40}, {t, -40, 40},
PlotPoints -> 2000, DisplayFunction -> Identity]][[1]]
Mathematica did not consume too much memory.
Moreover, if one loads the legacy graphic engine
In[1]:= << Version5`Graphics`
then evaluating the following expression
In[2]:= Plot3D[Sin[x y], {x, 0, 2 \[Pi]}, {y, 0, 2 \[Pi]},
PlotPoints -> 4000, DisplayFunction -> Identity]
Out[2]= --SurfaceGraphics--
works fine. However, evaluating the same expression with the new graphic
engine eats up all the memory available.
In[3]:= << Version6`Graphics`
In[4]:= Plot3D[Sin[x y], {x, 0, 2 \[Pi]}, {y, 0, 2 \[Pi]},
PlotPoints -> 4000, DisplayFunction -> Identity]
Regards,
Jean-Marc
ulises
Although the V6 and V5 plotting engines are quite different, a more
fair comparison would be:
Timing[Plot3D[Sin[x*y], {x, 0, 2*Pi}, {y, 0, 2*Pi},
PlotPoints -> 4000, PerformanceGoal -> "Speed",
NormalsFunction -> None];][[1]]
On my machine the speed difference is still about 9x (120 sec vs 14
sec). Note that the ouputs are completely different, in V5 you had a
SurfaceGraphics object which only had the z values, equivalent to
Table[]. In V6, you have all the {x,y,z} coordinates, plus all the
Polygon[] entries for the all mesh consisting of triangles, all
enclosed in a GraphicsComplex[] object.
In V6, the emphasis is on adaptive refinement and quality, which means
that smaller values of PlotPoints should be enough to generate a
smoother surface.
For this benchmark test files, as mentioned before, using the legacy
<<Version5`Graphics` is the only way of making a fair comparison.
Ulises Cervantes
WRI
Igor C. Antonio
points in Mathematica 6.0 due to the adaptive sampling feature of
Plot3D.
(http://www.wolfram.com/products/mathematica/newin6/content/HighImpactAdaptiveVisualization/)
You should try setting the option MaxRecursion->0 in Plot3D to disable
that adaptive sampling.
--
Igor C. Antonio
Wolfram Research, Inc.
To email me personally, remove the dash.
Murray Eisenberg
package. But in any case, prudence would suggest using some third-part
benchmark program!
Jens-Peer Kuska wrote:
> Hi,
>
> probably that is the reason for the package:
>
> Needs["Benchmarking`"]
>
> and the function
>
> Benchmark[]
>
> regards
> Jens
>
>
> eric.l...@gmail.com wrote:
>> Hello. I am performing an evaluation of Mathematica 6 with the demo
>> version on both Windows XP 32 bits and OS X 10.4.9. However both crash
>> when evaluating the famous Karl's Mathematica benchmark:
>> http://www2.staff.fh-vorarlberg.ac.at/~ku/karl/timings50.html
>>
>> Did anyone experience the same thing? Have you got any suggestion?
>> Thank you.
>>
>>
>
--
Murray Eisenberg mur...@math.umass.edu
Mathematics & Statistics Dept.
Lederle Graduate Research Tower phone 413 549-1020 (H)
University of Massachusetts 413 545-2859 (W)
710 North Pleasant Street fax 413 545-1801
Amherst, MA 01003-9305