Mathematica 7 compares to other languages

Jon Harrop jon at ffconsultancy.com
Tue Dec 9 01:56:28 CET 2008


Xah Lee wrote:
> A moron, wrote:
> > You failed the challenge that you were given.
> 
> you didn't give me a challenge.

Thomas gave you the challenge:

  "What I want in return is you to execute and time Dr. Harrop's original
code, posting the results to this thread... By Dr. Harrop's original code,
I specifically mean the code he posted to this thread. I've pasted it below
for clarity.".

Thomas even quoted my code verbatim to make his requirements totally
unambiguous. Note the parameters [9, 512, 4] in the last line that he and I
both gave:

  AbsoluteTiming[Export["image.pgm", Graphics at Raster@Main[9, 512, 4]]]

You have not posted timings of that, let alone optimized it. So you failed.

> I gave you. I asked for $5 sincerity 
> wage of mutal payment or money back guarantee, so that we can show
> real code instead of verbal fight. You didn't take it and do nothing
> but continue petty quarrel on words.

Then where did you post timings of that exact code as Thomas requested?

>
http://www.gossamer-threads.com/lists/python/python/698196?do=post_view_threaded#698196 
> ” ) You refuse to acknowledge it, and continue babbling, emphasizing that
> my code should be some hundred times faster make valid argument.

That is not my code! Look at the last line where you define the scene:

  Timing[Export["image.pgm",Graphics[at]Raster at Main[2,100,4.]]]

Those are not the parameters I gave you. Your program is running faster
because you changed the scene from over 80,000 spheres to only 5 spheres.
Look at your output image: it is completely wrong!

> As i said, now pay me $300, i will then make your Mathematica code in
> the same level of speed as your OCmal. If it does not, money back
> guaranteed.

Your money back guarantee is worthless if you cannot even tell when you have
failed.

> Show me your OCmal code that will compile on my machine (PPC Mac, OSX
> 10.4.x).

The code is still on our site:

  http://www.ffconsultancy.com/languages/ray_tracer/

OCaml, C++ and Scheme all take ~4s to ray trace the same scene.

> I'll make your Mathematica code in the same speed level as 
> your OCmal code. (you claimed Mathematica is roughly 700 thousand
> times slower to your OCmal code. I claim, i can make it, no more than
> 10 times slower than the given OCmal code.)

You have not even made it 10% faster, let alone 70,000x faster. Either
provide the goods or swallow the fact that you have been wrong all along.

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/?u



More information about the Python-list mailing list