Richard Fateman wrote:
> Could you share your timing results with us?
>> And sbcl runs twice slower than cmucl on the same programs.
Last week i have run maxima computations which took around 10 hours
each. For the sake of comparison i have run the same job under sbcl on
another machine, exactly similar. It was running more than twice slower
until it bombed out.
>>
>> who cares the Win32 API ? You perhaps, me not a single instant.
>>
> Your opinion, valuable as it may be to you,
This is why i was careful to mention myself as not caring about Win32 :)
> is probably not so widely shared.
I am certainly not at Berkeley university but you could be surprised to see
how little Windows is used at Paris 6. In our lab there are around 150
Linux machines, and exactly one Windows machine, in the improbable event
somebody needs this crapware, plus some Macs. If you look at the statistics
you will see that Windows has been bought 150 times, and Linux never.
Maxima for Linux has been downloaded once, even if it may run on 100
machines, furthermore a Windows version may have been downloaded several
times for testing purposes, say on laptops. This to say how little these
statistics are reliable ...
> Do you think this would be an issue in the overall performance of Maxima?
Not at all. But i know for sure that OpenGL drivers are problematic on Linux
(only closed source for Linux for nVidia, big problems for Ati), and even
more on FreeBSD (no AMD64 version on nVidia, etc.) so i see a big problem
when relying on OpenGL for a program which works perfectly at the command
line. Personnally i hate the "modern" graphical interface of Maple, based
on Java, which is orders of magnitude worse and slower than the old Motif
one. So when i hear that it is a good idea to base a GUI on Eclipse i have
some doubts. I find the GUI in wxmaxima perfectly adequate.
--
Michel Talon