Stavros Macrakis wrote:
> On 5/22/07, ahmet alper parker <aaparker at gmail.com> wrote:
>
>>
>> As a recommendation I would like to say that can the newer builds for
>> windows be compiled with clisp for not to confront the same ram
>> problems.
>> I am sure it would be easy for the developers to compile with msys-mingw
>> environment with clisp.
>
>
Maxima supports having several builds with different LISP
implementations simultaneously,
maxima --list-avail
shows currently available ones.
It's a bit unfortunate that Windows binaries come with single
implementation.
For some (unknown to me, but may be unavoidable) reason Windows maxima
comes with GCL which comes with GCC, both stripped versions
IMO packaging into windows Maxima binaries for CLISP *and* GCL without
GCL itself will provide more flexibility.
>
> Memory usage is not the only criterion. I'll let others address the
> system
> issues, but clisp appears to be much slower in some cases. For example
> (1GHz, 512Mb, W2k):
>
> clisp GCL
> 42 sec 5 sec genmatrix(a,6,6)^^-1
> 13 sec 3 sec makelist(i,i,1,100000)
> 19 sec 5 sec makelist(factor(x^2-i),i,1,10000)
> 104 sec 20 sec makelist(factor(x^i-1),i,1,1000)
>
> These factors of 4-8 are not trivial.
run_testsuite() time difference is factor 2, so, I believe factor 4-8
isn't common
Best regards,
Vadim.