Michel Talon wrote:
> ?iga Lenar?i? wrote:
>
>
>
>> I think Allegro
>>
>
> What if you don't care about Allegro? does it run maxima faster?
>
I have not done extensive timings. However, some experiments that
Barton Willis and I have conducted which involve using hashtables to
make the
simplifier run much faster, vastly improve the speed on Allegro, but
less so on other lisps. Although the results are not written out in a
satisfactory form,
I would guess that Barton would be as happy as I would be, if you wished
to experiment further, perhaps with your extensive experience with other
computers,
operating systems, lisps, etc.
> Answer, no.
Could you share your timing results with us?
> And sbcl runs twice slower than cmucl on the same programs.
>
I am surprised by this result. Can you share this data? My impression
is that they should be quite similar.
My own experience with CMUCL on SUN workstations was that it could
produce fast-running code, but the compiler was slow and annoyingly
verbose.
SBCL would rate higher in my estimation if it did not have some apparent
difficulty in running on Windows, in some intermittent unpredictable
manner which I think has to do with memory management.
>
>> has some custom bindings to Win32 API to enable
>>
>
> who cares the Win32 API ? You perhaps, me not a single instant.
>
Your opinion, valuable as it may be to you, is probably not so widely
shared.
>
>> So cl-opengl bindings (which are written with CFFI) work on all lisps
>> and on all operating systems. OpenGL library allows you to talk
>> directly to graphics card hardware, so that means you have to basicly
>>
>
> What if your operating system doesn't support OpenGL on your graphic card?
>
I have no idea whether OpenGL is supported on any of the graphics cards
in any of the computers and operating systems I use, so my answer is,
"nothing"
> Then you emulate it on software and loose a good bit of performance, once
> more.
>
>
Do you think this would be an issue in the overall performance of Maxima?