I did few more tests with fricas0. One thing to note is that
fricas0 is incompatible with Poplog: Poplog Lisp will _not_
add '.lisp' extension to file name without extention, so
I had to add explicit file extentions. Second, we do not
get useful command line from Poplog ("user" command line
skips executable, and if we dig deeper we get flags
internal to Poplog, but paths point to Poplog installation).
Also, fricas0 skipped a few Poplog specific hunks. However,
hacking around problems I was able to run fricas0 with
Poplog. Note that Poplog compiles files on load. There
are no compilation to .fasl. As an alternative Poplog
offers "stacked images", that is one can dump what
is new in given image but reuse previous (say system)
image. With Poplog time to run integration
test is:
real 0m12.149s
user 0m6.460s
sys 0m0.120s
For comparison, without extra tweaks time for sbcl
in intrpret mode (with fasteval) is 22 sec, time
for Clozure CL 11 is 25 sec. In this test for
ECL I got:
real 0m15.486s
user 0m22.764s
sys 0m0.096s
Anyway, both Poplog and Clozure CL compile files on
load and both have fast compilers, but Poplog compiler
is significantly faster (Clozure CL optimizes better).
Interestingly, ECL has CPU time significantly higher
than real time, while Poplog has real time significantly
higher than CPU time.
Compiling first files given by Qian with sbcl I
get compile+load+execute time 6.955. So, this
is fastest, but require knowledge of critical
files.
More on Poplog: as you can see from above, Poplog
has many quirks and you may wonder why to bother?
Well, one reason is compile time: using sbcl significant
part of compile time is Lisp compile time. Similarly
with ECL and GCL (here we have mostly C compile
time) but worse since _all_ times are longer.
So to get significant improvement of compile time
we need both to improve Spad compile time _and_
Lisp compile time. Somewhat related, Poplog has
also more direct compiler interface, so it gives
hope for non-Lisp compilation mode.
--
Waldek Hebisch
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/fricas-devel/20220731133515.GA681%40fricas.math.uni.wroc.pl.
