TCL Speed vs PERL 
Author Message
 TCL Speed vs PERL

Been playing around with TCL and comparing it somewhat to perl.  In some
informal timing comparisons I've noticed that PERL 5 is at least 2 to 3
times faster than TCL 8. This comes from some simple script tests like:

for  i = 1 to MAXi
   for j = 1 to MAXj
    Y = expression with i and J

and varying the max values and/or expressions. Granted if I were using
some unique feature of each language I could expect some differences in
speed as each language might have different parts optimized.  But when
using simple "for" loops and expressions I would have expected similiar
speeds.   Anyone know about any plans to speed up TCL to make it comprable
in execution speed to PERL?  



Tue, 07 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Quote:

>> Been playing around with TCL and comparing it somewhat to perl.  In some
>> informal timing comparisons I've noticed that PERL 5 is at least 2 to 3
>> times faster than TCL 8. This comes from some simple script tests like:

>> for  i = 1 to MAXi
>>    for j = 1 to MAXj
>>     Y = expression with i and J

>> and varying the max values and/or expressions. Granted if I were using
>> some unique feature of each language I could expect some differences in
>> speed as each language might have different parts optimized.  But when
>> using simple "for" loops and expressions I would have expected similiar
>> speeds.   Anyone know about any plans to speed up TCL to make it comprable
>> in execution speed to PERL?

>For a more extensive set of benchmarks comparing the performance of
>Tcl with Perl, MetaTalk, and dtksh (the two other major GUI-aware
>scripting language tools for UNIX systems), see the benchmarks section
>of the white paper comparing the languages on the MetaCard WWW site
>http://www.metacard.com/

>The punch line is that Perl and MetaTalk have comparable speed and
>both are *much* faster than Tcl or dtksh.  Unfortunately, much of the
>performance difference is due to the design of the two latter
>languages: it is often impossible to precompile expressions because of
>ambiguities about the types of the tokens in a string.  This means
>that it's unlikely that these languages will ever achieve the
>performance of languages that have a syntax (and function call
>architecture) more conducive to optimizing preprocessors.
>  Regards,
>    Scott

I checked out this page.  Although there may be design constraints in the
languages  that will effect the final outcome it is clear that you used a
very old copy of TCL to do your benchmarking. You used a 7.4  version and
compared it to perl 5 as well as your own product. That's hardly a fair
fight.  TCL 8.0 in my own informal benchmarks is about 10x faster in the
looping tests than 7.6.  Don't know how much faster than 7.4.  I'm afraid I
find your benchmarks outdated and invalid  for this reason alone.  Your
metacard product may infact be faster than TCL 8.0.  But I doubt to the
extent that you imply.  Perhaps these benchmarks were done before TLC 8.0
was available.  (But I doubt before TCL 7.6) In any case I suggest you
update your page with more accurate timings.  


Wed, 08 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Quote:

> Been playing around with TCL and comparing it somewhat to perl.  In some
> informal timing comparisons I've noticed that PERL 5 is at least 2 to 3
> times faster than TCL 8. This comes from some simple script tests like:

> for  i = 1 to MAXi
>    for j = 1 to MAXj
>     Y = expression with i and J

> and varying the max values and/or expressions. Granted if I were using
> some unique feature of each language I could expect some differences in
> speed as each language might have different parts optimized.  But when
> using simple "for" loops and expressions I would have expected similiar
> speeds.   Anyone know about any plans to speed up TCL to make it comprable
> in execution speed to PERL?

For a more extensive set of benchmarks comparing the performance of
Tcl with Perl, MetaTalk, and dtksh (the two other major GUI-aware
scripting language tools for UNIX systems), see the benchmarks section
of the white paper comparing the languages on the MetaCard WWW site
http://www.metacard.com/

The punch line is that Perl and MetaTalk have comparable speed and
both are *much* faster than Tcl or dtksh.  Unfortunately, much of the
performance difference is due to the design of the two latter
languages: it is often impossible to precompile expressions because of
ambiguities about the types of the tokens in a string.  This means
that it's unlikely that these languages will ever achieve the
performance of languages that have a syntax (and function call
architecture) more conducive to optimizing preprocessors.
  Regards,
    Scott



Wed, 08 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL



Quote:

>For a more extensive set of benchmarks comparing the performance of
>Tcl with Perl, MetaTalk, and dtksh (the two other major GUI-aware
>scripting language tools for UNIX systems), see the benchmarks section
>of the white paper comparing the languages on the MetaCard WWW site
>http://www.metacard.com/

>The punch line is that Perl and MetaTalk have comparable speed and
>both are *much* faster than Tcl or dtksh...

These MetaTalk people have been bragging for awhile in this newsgroup.
I'm sure they have a great product but they are not comparing the
latest Tcl in there charts.  It was some version of Tcl7.4.  They
should stop trying to pull people away from Tcl with mis-information.

Here are a few comparisons of Perl5.003 and Tcl8.0 using there very
own benchmarks.  Using Linux on a 200mhs PPro

-----------------------------------             Tcl8.0  Perl5.003
-----------------------------------------------------------------
1000000 repeats in                            3.694873     1
10000 iterative factorial(100) in             6.331929     3
10000 iterative factorial(100) with 'if' in   8.009346     4

Perl is little faster at some things but nowhere near what that
white paper claims..  Tcl is much better now.  I wonder if the Perl
script is truncating it's times? e.g. 3.99999 -> 3.  I didn't even
look at the code to see if it was written very badly.

Bill
--
##############################################################################
# Bill Thorson                # Research Coordinator                         #
# Dept of Atmospheric Science # Tropical Meteorology Project                 #
# Colorado State University   #                                              #



Wed, 08 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Quote:

> The punch line is that Perl and MetaTalk have comparable speed and
> both are *much* faster than Tcl or dtksh.

if raw speed were the only thing about which i cared, i might listen.
i happen to like readable (few obsure cladiddletwiddle characters with
semantics vastly different from the obscure cladiddletwiddle characters
i've already learned) nonproprietary (ie not often costing a fortune,
not likely to die with the company, not subject to the whims of any
one company; IMO Sun's excerised good self-control here) nonspecialised
(ie not tied into one tool, extensible) dynamic (ie introspective,
capable of self-modifying code) languages available in source.

to get all that, a little speed is worth the sacrifice.  hell, it's
unavoidable.

Quote:
>   Regards,
>     Scott

--
Hume dot Smith at Dess dot Tallships dot iStar dot CA


Wed, 08 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Hume Smith wrote in article

:
:> The punch line is that Perl and MetaTalk have comparable speed and
:> both are *much* faster than Tcl or dtksh.
:
:if raw speed were the only thing about which i cared, i might listen.
:i happen to like readable (few obsure cladiddletwiddle characters with
:semantics vastly different from the obscure cladiddletwiddle characters
:i've already learned) nonproprietary (ie not often costing a fortune,
:not likely to die with the company, not subject to the whims of any
:one company; IMO Sun's excerised good self-control here) nonspecialised
:(ie not tied into one tool, extensible) dynamic (ie introspective,
:capable of self-modifying code) languages available in source.

am i to take that PERL is [now? or was it always] proprietary, and that PERL
is 'specialized'/'tied to one tool'?

confused and curious...

jdm

;; jean-david marrow



Thu, 09 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Quote:

> I believe Smith was taking a swipe at the previous poster, who
> has an interest in Metatalk, which apparently *is* proprietary.
> Any apparent reference to Perl was just sloppy quoting.

no, PERL falls under my "unreadable" brush, as does C++.  i dread the day
i may be forced to seriously modify someone else's perl.  (i've done a
couple of minor changes, and those were bad enough.)

Quote:
>                    who uses both perl and tcl

--
Hume dot Smith at Dess dot Tallships dot iStar dot CA


Fri, 10 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Perl is certainly not proprietary nor likely to become so; in fact,
I venture you'd be a dead man now if you'd said that in the hearing
of Larry Wall or any of the other principal perpetrators of Perl.

I believe Smith was taking a swipe at the previous poster, who
has an interest in Metatalk, which apparently *is* proprietary.
Any apparent reference to Perl was just sloppy quoting.

                        regards, tom lane
                        who uses both perl and tcl



Fri, 10 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Quote:

> >For a more extensive set of benchmarks comparing the performance of
> >Tcl with Perl, MetaTalk, and dtksh (the two other major GUI-aware
> >scripting language tools for UNIX systems), see the benchmarks section
> >of the white paper comparing the languages on the MetaCard WWW site
> >http://www.metacard.com/

> >The punch line is that Perl and MetaTalk have comparable speed and
> >both are *much* faster than Tcl or dtksh...

(snip)

Quote:
> Here are a few comparisons of Perl5.003 and Tcl8.0 using there very
> own benchmarks.  Using Linux on a 200mhs PPro
> -----------------------------------             Tcl8.0  Perl5.003
> -----------------------------------------------------------------
> 1000000 repeats in                            3.694873     1
> 10000 iterative factorial(100) in             6.331929     3
> 10000 iterative factorial(100) with 'if' in   8.009346     4
> Perl is little faster at some things but nowhere near what that
> white paper claims..  Tcl is much better now.  I wonder if the Perl
> script is truncating it's times? e.g. 3.99999 -> 3.  I didn't even
> look at the code to see if it was written very badly.

The Tcl 8 benchmarks are available on the WWW site, though the article
was originally written when Tcl 7.4 was the current release.  Tcl 8
does show dramatic speedups over 7,4 for some tests, but it still
falls far short of Perl and MetaTalk (which the you undoubtedly
know since the benchmarks all all run in a batch, and you
just selected the few where Tcl looks the best).  There are tests
where MetaTalk and Perl are *12* times as fast as Tcl rather than just
the usual 2 or 3 times as fast (as the person who started this
thread reported).
  Regards,
    Scott
Quote:
> Bill
> --
> ##############################################################################
> # Bill Thorson                # Research Coordinator                         #
> # Dept of Atmospheric Science # Tropical Meteorology Project                 #
> # Colorado State University   #                                              #



Fri, 10 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL

Quote:

> Perl is certainly not proprietary nor likely to become so; in fact,
> I venture you'd be a dead man now if you'd said that in the hearing
> of Larry Wall or any of the other principal perpetrators of Perl.

Sound like dangerous guys.  I hope not to run into them in a
dark alley.

Quote:
> I believe Smith was taking a swipe at the previous poster, who
> has an interest in Metatalk, which apparently *is* proprietary.
> Any apparent reference to Perl was just sloppy quoting.

Actually, it's not.  It's based on the xTalk language used in
HyperCard, SuperCard, Oracle Media Objects, HyperSense and other
products.  As I recall, one of the primary definitions of "open"
vs "proprietary" was that multiple implementations of the
technology exist.  This is true for MetaTalk.  How about Tcl?
  Regards,
    Scott
Quote:
>                         regards, tom lane
>                         who uses both perl and tcl



Fri, 10 Mar 2000 03:00:00 GMT  
 TCL Speed vs PERL


: > I believe Smith was taking a swipe at the previous poster, who
: > has an interest in Metatalk, which apparently *is* proprietary.
: > Any apparent reference to Perl was just sloppy quoting.

: no, PERL falls under my "unreadable" brush, as does C++.  i dread the day
: i may be forced to seriously modify someone else's perl.  (i've done a
: couple of minor changes, and those were bad enough.)

I guess it depends on what you are used to.  Despite the fact that
I have been using tcl for while I still haven't gotten the hang of
it - IOW it just doesn't feel comfortable.  Reminds me of Lisp with
square brackets...

I'll report back when I finish my current project (Expect, Tcl and
Perl) and we'll see if I've gotten the hang of it by then :-)  Could
be I'm just Tcl impaired.

cheers,
Bryan Miller



Sat, 11 Mar 2000 03:00:00 GMT  
 
 [ 25 post ]  Go to page: [1] [2]

 Relevant Pages 

1. Perl vs TCL (was: Execution speed of Perl?)

2. Perl speed vs. Python speed

3. Java speed vs Tcl speed

4. Speed of Python vs. Perl

5. Speed problems with Python vs. Perl

6. Yet another Python vs. Perl speed issue/question

7. Speed of Python vs. Perl

8. python vs. perl speed comparisons

9. Python vs Perl: benchmarking for speed?

10. PYTHON VS. PERL VS. TCL

11. Python Binding [Was: Re: PYTHON VS. PERL VS. TCL ]

12. jredford's flames and criticism (was: PYTHON VS. PERL VS. TCL )

 

 
Powered by phpBB® Forum Software