On Aug 24, 2009, at 8:41 PM, Raymond Toy wrote:
>>
> That might be expected if you're using a very recent version of SBCL
> (and 19f or later CMUCL). I know in cmucl 19f, complex arithmetic
> uses
> sse2 by default, and that can be significantly faster because the
> operations are vectorized. The real/imag array version will suffer
> quite a bit since the complex operations probably won't be vectorized.
Forgot to comment on that bit -
I use SBCL 1.0.30 (latest) but I don't think they have imported these
improvements from CMUCL. The new CMUCL has 2 core images - one sse
optimized and one based on standard x87 coprocessor. SBCL comes with
1 core only - therefore I think it doesn't have these optimizations.
SBCL is nice because one can compile it on any CL laying around and
has Unicode support, threads...
But since most of the Maxima users run GCL I think it's important to
also test timings on GCL in general.
Regards,
Ziga