www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.announce - D is up on the "Computer Language Shootout" site.

reply "Dave" <Dave_member pathlink.com> writes:
http://shootout.alioth.debian.org/

Presumably because it is a 'new' language, the results for D are not part of 
the primary list on the home page yet.

The listing that includes D can be found here:

http://shootout.alioth.debian.org/great/index.php?sort=fullcpu

And the D results compared to other languages are here:

http://shootout.alioth.debian.org/great/benchmark.php?test=all&lang=all&sort=fullcpu

If you want to submit a faster/better/more elegant solution for any of the 
tests, read the FAQ first. And make sure to follow the 'rules' for each test 
(for example, a number of the tests specify line I/O as opposed to buffering 
the entire input).

The goal of the original code is to not only run fast, but be reasonably 
elegant and easy to follow as well, and show at least some features of the D 
language and runtime as opposed to copy and paste C/++, even if the code may 
run a tad slower from time to time.

- Dave
Mar 18 2005
next sibling parent reply "Charles" <cee-lo green.com> writes:
Umm,
http://shootout.alioth.debian.org/great/benchmark.php?test=all&lang=all&sort
=fullcpu ,
shows D as the default best!!!

Awesome!!

Thanks to whoever put this up ,this will get alot of face-time for D!

Charlie

"Dave" <Dave_member pathlink.com> wrote in message
news:d1fii1$281u$1 digitaldaemon.com...
 http://shootout.alioth.debian.org/

 Presumably because it is a 'new' language, the results for D are not part

 the primary list on the home page yet.

 The listing that includes D can be found here:

 http://shootout.alioth.debian.org/great/index.php?sort=fullcpu

 And the D results compared to other languages are here:

=fullcpu
 If you want to submit a faster/better/more elegant solution for any of the
 tests, read the FAQ first. And make sure to follow the 'rules' for each

 (for example, a number of the tests specify line I/O as opposed to

 the entire input).

 The goal of the original code is to not only run fast, but be reasonably
 elegant and easy to follow as well, and show at least some features of the

 language and runtime as opposed to copy and paste C/++, even if the code

 run a tad slower from time to time.

 - Dave

Mar 18 2005
parent reply "Charles" <cee-lo green.com> writes:
Thanks to Dave Flabedo :D


"Charles" <cee-lo green.com> wrote in message
news:d1fjsf$29i9$1 digitaldaemon.com...
 Umm,

 =fullcpu ,
 shows D as the default best!!!

 Awesome!!

 Thanks to whoever put this up ,this will get alot of face-time for D!

 Charlie

 "Dave" <Dave_member pathlink.com> wrote in message
 news:d1fii1$281u$1 digitaldaemon.com...
 http://shootout.alioth.debian.org/

 Presumably because it is a 'new' language, the results for D are not


 of
 the primary list on the home page yet.

 The listing that includes D can be found here:

 http://shootout.alioth.debian.org/great/index.php?sort=fullcpu

 And the D results compared to other languages are here:


 =fullcpu
 If you want to submit a faster/better/more elegant solution for any of


 tests, read the FAQ first. And make sure to follow the 'rules' for each

 (for example, a number of the tests specify line I/O as opposed to

 the entire input).

 The goal of the original code is to not only run fast, but be reasonably
 elegant and easy to follow as well, and show at least some features of


 D
 language and runtime as opposed to copy and paste C/++, even if the code

 run a tad slower from time to time.

 - Dave


Mar 18 2005
parent "Dave" <Dave_member pathlink.com> writes:
"Charles" <cee-lo green.com> wrote in message 
news:d1fka4$2a24$1 digitaldaemon.com...
 Thanks to Dave Flabedo :D


 "Charles" <cee-lo green.com> wrote in message
 news:d1fjsf$29i9$1 digitaldaemon.com...
 Umm,

 =fullcpu ,
 shows D as the default best!!!

 Awesome!!

 Thanks to whoever put this up ,this will get alot of face-time for D!

 Charlie


Almost all of it based and reliant on the work of others <g> I hope this does get D some good face-time - I think the Shootout site is pretty well done as far as Benchmarks go.
Mar 18 2005
prev sibling next sibling parent reply clayasaurus <clayasaurus gmail.com> writes:
Dave wrote:
 http://shootout.alioth.debian.org/
 
 Presumably because it is a 'new' language, the results for D are not part of 
 the primary list on the home page yet.
 
 The listing that includes D can be found here:
 
 http://shootout.alioth.debian.org/great/index.php?sort=fullcpu
 
 And the D results compared to other languages are here:
 
 http://shootout.alioth.debian.org/great/benchmark.php?test=all&lang=all&sort=fullcpu
 
 If you want to submit a faster/better/more elegant solution for any of the 
 tests, read the FAQ first. And make sure to follow the 'rules' for each test 
 (for example, a number of the tests specify line I/O as opposed to buffering 
 the entire input).
 
 The goal of the original code is to not only run fast, but be reasonably 
 elegant and easy to follow as well, and show at least some features of the D 
 language and runtime as opposed to copy and paste C/++, even if the code may 
 run a tad slower from time to time.
 
 - Dave
 
 

That is some amazing work you've done there, Dave : )
Mar 18 2005
parent reply "Dave" <Dave_member pathlink.com> writes:
"clayasaurus" <clayasaurus gmail.com> wrote in message 
news:d1fmco$2c54$1 digitaldaemon.com...
 Dave wrote:
 http://shootout.alioth.debian.org/

 Presumably because it is a 'new' language, the results for D are not part 
 of the primary list on the home page yet.

 The listing that includes D can be found here:

 http://shootout.alioth.debian.org/great/index.php?sort=fullcpu

 And the D results compared to other languages are here:

 http://shootout.alioth.debian.org/great/benchmark.php?test=all&lang=all&sort=fullcpu

 If you want to submit a faster/better/more elegant solution for any of 
 the tests, read the FAQ first. And make sure to follow the 'rules' for 
 each test (for example, a number of the tests specify line I/O as opposed 
 to buffering the entire input).

 The goal of the original code is to not only run fast, but be reasonably 
 elegant and easy to follow as well, and show at least some features of 
 the D language and runtime as opposed to copy and paste C/++, even if the 
 code may run a tad slower from time to time.

 - Dave

That is some amazing work you've done there, Dave : )

The people who developed D and the libraries are amazing, it was just fun for me ;) Plus, I basically just 'ported' a lot of the implementations that were already there for other languages; if it happens to be elegant than it's not my doing <g> Hopefully the results stick when other languages catch-up with the implementations of all of the tests. Just offhand, if you look at what C is still missing - D does pretty well in those - so the final looks to maybe end up with DMD 2nd, mid-way between Intel C and GCC (with everything else remaining roughly the same). Not bad.
Mar 18 2005
parent Brad Anderson <brad dsource.dot.org> writes:
Dave wrote:
the final looks to maybe
 end up with DMD 2nd, mid-way between Intel C and GCC (with everything else 
 remaining roughly the same). Not bad.

Don't feel bad coming in second to them. All the tests are running on *their* chips...
Mar 18 2005
prev sibling parent reply Dejan Lekic <leka entropy.tmok.com> writes:
If you read all those pages carefully You would see that D is the only
language with ALL test programs done 100% - that is the main reason why
it's first. If you check every single test separately You won't see D on
first place - more likely somewhere around 10th, what is OK...

-- 
...........
Dejan Lekic
  http://dejan.lekic.org
  
Mar 19 2005
parent reply "Dave" <Dave_member pathlink.com> writes:
"Dejan Lekic" <leka entropy.tmok.com> wrote in message 
news:d1i4kn$1uab$1 digitaldaemon.com...
 If you read all those pages carefully You would see that D is the only
 language with ALL test programs done 100% - that is the main reason why
 it's first. If you check every single test separately You won't see D on
 first place - more likely somewhere around 10th, what is OK...

If you look at the other 'leaders' like C and C++ you'll see that it is the same for them (they aren't tops in every test either). A simple avg. of the top 3 by dividing the total score with the # of completed tests has D in 2nd. Right now, GCC is missing 9 tests and DMD is missing one. If you use the 'shootout calculator' to zero-out those missing tests D still comes in a close 3rd. D should do pretty well in all of the 10 missing tests, so the trend looks to be in DMD's favor for ending up 2nd or maybe 3rd overall with this set of tests and these compilers. Certainly a lot better than 10th...
 -- 
 ...........
 Dejan Lekic
  http://dejan.lekic.org
 

Mar 19 2005
next sibling parent reply "Charles" <cee-lo green.com> writes:
I'll be exteremely happy to see D in the top 5 , considering intel and GCC
( and most C++ compilers ) have had decades to develop , and teams of
programmers.

And for me the big winner is the development time for D ( much shorter than
C and C++ ) , which is really the most expensive part of any software
project.

Again big-ups to Dave for doing this, and doing it well :).

Charlie


"Dave" <Dave_member pathlink.com> wrote in message
news:d1i780$21ai$1 digitaldaemon.com...
 "Dejan Lekic" <leka entropy.tmok.com> wrote in message
 news:d1i4kn$1uab$1 digitaldaemon.com...
 If you read all those pages carefully You would see that D is the only
 language with ALL test programs done 100% - that is the main reason why
 it's first. If you check every single test separately You won't see D on
 first place - more likely somewhere around 10th, what is OK...

If you look at the other 'leaders' like C and C++ you'll see that it is

 same for them (they aren't tops in every test either). A simple avg. of

 top 3 by dividing the total score with the # of completed tests has D in
 2nd.

 Right now, GCC is missing 9 tests and DMD is missing one. If you use the
 'shootout calculator' to zero-out those missing tests D still comes in a
 close 3rd.

 D should do pretty well in all of the 10 missing tests, so the trend looks
 to be in DMD's favor for ending up 2nd or maybe 3rd overall with this set

 tests and these compilers.

 Certainly a lot better than 10th...

 --
 ...........
 Dejan Lekic
  http://dejan.lekic.org


Mar 19 2005
parent reply "Charles" <cee-lo green.com> writes:
And big-ups to Walter of course for making it all possible!


"Charles" <cee-lo green.com> wrote in message
news:d1i8b1$22ci$1 digitaldaemon.com...
 I'll be exteremely happy to see D in the top 5 , considering intel and GCC
 ( and most C++ compilers ) have had decades to develop , and teams of
 programmers.

 And for me the big winner is the development time for D ( much shorter

 C and C++ ) , which is really the most expensive part of any software
 project.

 Again big-ups to Dave for doing this, and doing it well :).

 Charlie


 "Dave" <Dave_member pathlink.com> wrote in message
 news:d1i780$21ai$1 digitaldaemon.com...
 "Dejan Lekic" <leka entropy.tmok.com> wrote in message
 news:d1i4kn$1uab$1 digitaldaemon.com...
 If you read all those pages carefully You would see that D is the only
 language with ALL test programs done 100% - that is the main reason



 it's first. If you check every single test separately You won't see D



 first place - more likely somewhere around 10th, what is OK...

If you look at the other 'leaders' like C and C++ you'll see that it is

 same for them (they aren't tops in every test either). A simple avg. of

 top 3 by dividing the total score with the # of completed tests has D in
 2nd.

 Right now, GCC is missing 9 tests and DMD is missing one. If you use the
 'shootout calculator' to zero-out those missing tests D still comes in a
 close 3rd.

 D should do pretty well in all of the 10 missing tests, so the trend


 to be in DMD's favor for ending up 2nd or maybe 3rd overall with this


 of
 tests and these compilers.

 Certainly a lot better than 10th...

 --
 ...........
 Dejan Lekic
  http://dejan.lekic.org



Mar 19 2005
parent "Walter" <newshound digitalmars.com> writes:
You're welcome! What Dave's been doing for D is very valuable.


"Charles" <cee-lo green.com> wrote in message
news:d1i8d7$22du$1 digitaldaemon.com...
 And big-ups to Walter of course for making it all possible!


 "Charles" <cee-lo green.com> wrote in message
 news:d1i8b1$22ci$1 digitaldaemon.com...
 I'll be exteremely happy to see D in the top 5 , considering intel and


 ( and most C++ compilers ) have had decades to develop , and teams of
 programmers.

 And for me the big winner is the development time for D ( much shorter

 C and C++ ) , which is really the most expensive part of any software
 project.

 Again big-ups to Dave for doing this, and doing it well :).

 Charlie


 "Dave" <Dave_member pathlink.com> wrote in message
 news:d1i780$21ai$1 digitaldaemon.com...
 "Dejan Lekic" <leka entropy.tmok.com> wrote in message
 news:d1i4kn$1uab$1 digitaldaemon.com...
 If you read all those pages carefully You would see that D is the




 language with ALL test programs done 100% - that is the main reason



 it's first. If you check every single test separately You won't see




 on
 first place - more likely somewhere around 10th, what is OK...

If you look at the other 'leaders' like C and C++ you'll see that it



 the
 same for them (they aren't tops in every test either). A simple avg.



 the
 top 3 by dividing the total score with the # of completed tests has D



 2nd.

 Right now, GCC is missing 9 tests and DMD is missing one. If you use



 'shootout calculator' to zero-out those missing tests D still comes in



 close 3rd.

 D should do pretty well in all of the 10 missing tests, so the trend


 to be in DMD's favor for ending up 2nd or maybe 3rd overall with this


 of
 tests and these compilers.

 Certainly a lot better than 10th...

 --
 ...........
 Dejan Lekic
  http://dejan.lekic.org




Mar 19 2005
prev sibling parent reply Dejan Lekic <leka entropy.tmok.com> writes:
Shootout doesn't proof anything anyway so this discussion is with no
purpose. Thanks for sending D sources to them.

-- 
...........
Dejan Lekic
  http://dejan.lekic.org
  
Mar 19 2005
parent reply Derek Parnell <derek psych.ward> writes:
On Sun, 20 Mar 2005 06:14:37 +0100, Dejan Lekic wrote:

 Shootout doesn't proof anything anyway so this discussion is with no
 purpose. 

Of course, that's just your opinion. Other people may believe differently.
Thanks for sending D sources to them.

Yes. Thanks for this effort, David. To me, including D into the 'shoot out' shows a number of things... ** D is nowhere near the slowest of languages. ** D is capable of many different jobs. ** Somebody is keen enough to do the work and submit it. ** Somebody else is interested enough to add them to the site. ** It does bring D into a little more of the spotlight. ** It brings an opportunity for people to improve the submitted sources, and thus further showcasing the potential of D. -- Derek Parnell Melbourne, Australia 20/03/2005 5:13:46 PM
Mar 19 2005
parent "Dave" <Dave_member pathlink.com> writes:
"Derek Parnell" <derek psych.ward> wrote in message 
news:czjsgs3ypkqh.16v4o782ha49f.dlg 40tude.net...
 On Sun, 20 Mar 2005 06:14:37 +0100, Dejan Lekic wrote:

 Shootout doesn't proof anything anyway so this discussion is with no
 purpose.

Of course, that's just your opinion. Other people may believe differently.
Thanks for sending D sources to them.

Yes. Thanks for this effort, David.

You're welcome - like I mentioned, it's been fun! Funny thing is, I think I've spent more time tuning some C programs than it took to port those tests to D (note to self, array slicing and BufferedStream rule!).
 To me, including D into the 'shoot out' shows a number of things...

 ** D is nowhere near the slowest of languages.
 ** D is capable of many different jobs.
 ** Somebody is keen enough to do the work and submit it.
 ** Somebody else is interested enough to add them to the site.
 ** It does bring D into a little more of the spotlight.
 ** It brings an opportunity for people to improve the submitted sources,
 and thus further showcasing the potential of D.

Those are great points. If you check the site again, D is still holding it's own in the performance dept. after several more submissions for C and others. According to the alioth stats., this has been the busiest month for the Shootout in 1/2 year. It must have gotten /.'d in October of last year or something <g> Also, at: http://shootout.alioth.debian.org/great/benchmark.php?test=all&lang=all&sort=fullcpu you can play with the CPU, Memory and Code Line multipliers. If you zero out CPU and Code Lines and set memory to "1", D is doing well there (unlike most of the other GC'd implementations - all of the D tests I sent in use only the GC). But, more to your points above, if you do likewise with the 'Code Lines' - set it to 1 and zero the rest - D ranks near the top there. Considering that it is competing against functional and scripting languages, and probably has one of the lesser mature libraries, I think this is outstanding! In an effort to make the code reasonably clear, I was probably more 'verbose' than needed as well. All but one of the tests (pidigits) was done with the base runtime library only. I think Java and Mono C# may be the only ones able to do all of the tests w/o a 3rd party lib. involved (but I haven't made a major study of this). To me, this is all great, great news! And of course this doesn't even really showcase all of the other important features of D either. I find myself asking.. How can D /not/ really take off once more people start to try it? Walter and others who've helped with libraries, bug tracking, etc. have done an outstanding job with D, IMHO. - Dave
 -- 
 Derek Parnell
 Melbourne, Australia
 20/03/2005 5:13:46 PM 

Mar 25 2005