Hello,
I think it was already mentioned on the mailing list that
at least under Ecl Maxima becomes very inefficient after
running for some time, typically by a factor of 10 slower
than it should be, so one frequently needs to restart Maxima.
Until now I thought that would only happen for bigger computations.
However look at the following (three fresh re-starts):
Maxima 5.25.1 http://maxima.sourceforge.net
using Lisp ECL 11.1.1
(%i1) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 12.9080 seconds (12.9520 elapsed)
(%o1) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
(%i2) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 121.3250 seconds (121.5850 elapsed)
(%o2) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
(%i3) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 26.7130 seconds (26.7670 elapsed)
(%o3) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
(%i4) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 64.2590 seconds (64.3860 elapsed)
(%o4) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
Maxima 5.25.1 http://maxima.sourceforge.net
using Lisp ECL 11.1.1
(%i1) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 62.0850 seconds (62.2240 elapsed)
(%o1) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
(%i2) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 48.7650 seconds (48.8580 elapsed)
(%o2) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
(%i3) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 115.2030 seconds (115.4290 elapsed)
(%o3) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
Maxima 5.25.1 http://maxima.sourceforge.net
using Lisp ECL 11.1.1
(%i1) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 6.0920 seconds (6.1730 elapsed)
(%o1) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
(%i2) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 70.7340 seconds (70.9190 elapsed)
(%o2) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
(%i3) ifactors(1000000000000000000000000000000000000000000000000000000000007-1);
Evaluation took 103.8050 seconds (104.0320 elapsed)
(%o3) [[2,1],[7,1],[107,1],[1679641,1],[8255453,1],[152778774688461206737,1],
[315114064590027073770947,1]]
One sees that it should compute the results in 6 seconds, but right from the start
in can take 20 times that much.
According to my experience, there are certain functions (like ifactors) which
show such a behaviour, while otherwise you "only" get the slower increase in
run-time, after every computation it takes a bit more.
I hope somebody can look into it. Quite a lot of run-time is wasted in this
way, and it creates also a bad feeling.
We have experienced many examples of such behaviour, but I hope the above
is convincing (and reproducible; if it really is Ecl-specific, then one
should convince the Ecl-guys to fix it ...).
Thanks for your attention.
Oliver