On Friday 12 December 2003 03:59, Camm Maguire wrote:
<snipped>
>2) If you can be a bit more specific about which tests, and even which
> routines or subsystems, are behind the "1.1-2" times slower figure
> you cite, it would be most helpful. Even try my latest gprof
> patch, run a test which shows significant slowdown vs. CMUCL, and
> profile it. I'd like to get this figure to about "0.8" or so :-).
I have one example which shows that maxima -gcl wins factor 4 behind maxima
-cmucl. I don't know wether it is a "real" effect or result of the amateur
approach to the maxima codding. The example is about direct verification of
the Cartan identity - the Lie derivative of the exterior form must commutate
with exterior derivative of one. If you are interesting in I can send you the
code. Actually I posted it oneday at the list but nobody was interested.
Here is result of computation for the 3 -d and 4-th -order forms with cmucl
#########
(C6) DECSYM(q, 3, 0, [ANTI(ALL)], [])
Evaluation took 0.04 seconds (0.04 elapsed) using 487.156 KB.
(D6) DONE
(C7) SHOW(q([I1, i2, i3]) @ i4 @L [v, i5] - q([I1, i2, i3]) @L [v, i4] @ i5)
(D7) 0
Evaluation took 16.68 seconds (16.99 elapsed) using 69557.297 KB.
(C8) DECSYM(p, 4, 0, [ANTI(ALL)], [])
Evaluation took 0.02 seconds (0.01 elapsed) using 144.500 KB.
(D8) DONE
(C9) SHOW(p([I1, i2, i3, i4]) @ i5 @L [v, i6]
- p([I1, i2, i3, i4]) @L [v, i5] @
i6)
(D9) 0
Evaluation took 139.25 seconds (149.45 elapsed) using 544732.039 KB.
##########
And with gcl
###########
(C6) DECSYM(q, 3, 0, [ANTI(ALL)], [])
Evaluation took 0.00 seconds (00.40 elapsed)
(D6) DONE
(C7) SHOW(q([I1, i2, i3]) @ i4 @L [v, i5] - q([I1, i2, i3]) @L [v, i4] @ i5)
(D7) 0
Evaluation took 2.16 seconds (2.45 elapsed)
(C8) DECSYM(p, 4, 0, [ANTI(ALL)], [])
Evaluation took 0.00 seconds (0.00 elapsed)
(D8) DONE
(C9) SHOW(p([I1, i2, i3, i4]) @ i5 @L [v, i6]
- p([I1, i2, i3, i4]) @L [v, i5] @
i6)
(D9) 0
Evaluation took 29.93 seconds (30.70 elapsed)
Evaluation took 32.28 seconds (33.51 elapsed)
(D10) car_iden.demo
##############