Mai 2006 - Gambit-list - mailman.iro.umontreal.ca

About tables and equal? performance
by Christian 14 Mai '06

14 Mai '06

Hello I've tested the performance of the Gambit (hash-)tables. When loaded with many values, they beat perl hashes (I've tried to only measure the actual hash manipulation in perl, which means I've subtracted the time needed for loops. The following timings are a bit unfair to perl though, since perl is treating integers as strings in hash keys.) My timings with big tables (1 mio keys) and random integers as keys (on a Athlon (32bit) 2Ghz, Debian, perl 5.8.7): One table-ref requires about 1500 cpu cycles. Table-set! with previously non-existing keys called on a table which has not been created with a big enough initial size (e.g. resizing is necessary) requires about 3000-5000 cycles. Perl requires about 5500 cycles per hash write (with this many keys). (When pre-allocating gambit tables with a big enough size, writing time drops to about 1330 cycles.) BTW, the purely functional wt-tree implementation found in mit-scheme or slib, adapted to gambit by me (ask me if you want the code) requires about 5000 cycles per read. Adding a new value takes about 15000 cycles. I guess Gambit is so fast because it uses open-addressing (afaik, perl is using (or at least has been in the past) linked lists for the buckets), and because I'm using real integers in Gambit, whereas perl treats everything as strings when being used as keys. But surprisingly, accessing or updating small hashes in perl is very fast. With something like perl -we 'for(1..10000000){$z{"funun"}++}; print keys(%z),"\n"' each hash update (after subtraction of the loop time) only takes about 300 cycles. (I don't know whether they are reverting to a different data structure in this case or if it's just that fast.) I've written a table implementation for small tables, which can be found at http://scheme.mine.nu/gambit/scratch/quicktable/. (The implementation using vectors is a bit faster than the one (also included there) using alists.) This implementation takes about 200 cycles for one quicktable-ref/add1/quicktable-set! operation series, or about 160 cycles for one quicktable-update! operation (both when using eq? for key comparison), whereas Gambit's builtin table implementation takes about 1200 cycles for one table-ref/add1/table-set! operation series. Maybe my approach (just scanning the vector linearly, instead of calculating a hash value first to be used for addressing, if the table size is small) could be integrated into the gambit table implementation easily? There is one issue about this though: I must warn that the scan in quicktables is only fast when using eq?. Gambit's equal? function is very slow. Just comparing about 2-5 keys using equal? increases the time to about the same 1200 cycles as taken by the current Gambit hash tables. Now I'm not sure if that's only because equal? is not inlined (as I've mentioned in a previous mail, such function calls cost about 200 cycles), or if its implementation is slow. (BTW I'm also a bit surprised how costly structures are in safe mode. Doing manual safety checks and declaring (not safe) is *much* faster.) Cheers Christian.

1 0

Using Emacs with Gambit-c
by rcaronr＠videotron.ca 13 Mai '06

13 Mai '06

Hello, I am a new user of this list and also of the Scheme programming language. Gambit-c was chosen because of its origin (U de Montreal). The text book I am using is Simply Scheme from Brian Harvey and Matthew Wright of MIT, I have also access to other textbooks on the net. There are also some tutorial and notes at Concordia U., and at various other places on the Net which I am using to learn this Language. Dr.Scheme has also software on the subject, however it is PLT Scheme. My question is : What to use in my .emacs configuration file to use Emacs and where should I place the gambit.el file. I have tried the pllab.el and quack.el files configurations ( with some modifications for gambiit-c. has they are PLT specifics ) but for some reason it tells me that it cannot load my gambit.el file. May be I should stick to PLT Scheme or use Bigloo what do you think ? I am using Debian Sarge for my operation system. I do not have nor do I intend to have Windows. I can use Gambit-c in a terminal but it would be preferable to use Emacs, the configurations found on the Concordia pages for AI 472, either do not work or I am doing it wrong because of lack of experience and knowledge of Emacs. Learning both Emacs and Scheme has quite a learning curve. I would appreciate the help of some knowledgeable people because frankly I am at lost. Thanks for any amount of help and time you can bring me. P.S I am willing to learn but do not have the university background which you have, my maths stop at college level and I am no Einstein. I am retired and have all the time in the world however. For you guys I probably am grandpa. Maybe I should have chosen this name instead of Tournesol, but I have used Tournesol for the last 13 years. rcaronr(a)videotron.ca Richard

2 1

How to define c-lambda for this routine.
by Bradley Lucier 12 Mai '06

12 Mai '06

Marc: What do I put in my scm code to interface with this routine: ;; void rdft(long, long, double *, long *, double *); And what type of scheme vectors can I pass to this? Brad

2 5

c-define-type & tag
by Stephane Le Cornec 11 Mai '06

11 Mai '06

Assuming the following C code: struct Point {int x; int y;}; struct PointEx {int x; int y; int z;}; PointEx* getPointEx(); int length(Point*); And its Scheme-interface: (c-define-type Point "Point") (c-define-type Point* (pointer Point)) (c-define-type PointEx "PointEx") (c-define-type PointEx* (pointer PointEx)) (define get-PointEx (c-lambda () PointEx* "getPointEx")) (define Point-length (c-lambda (Point*) int "length")) How do you use type tags properly? Doing (c-define-type PointEx* (pointer PointEx Point*)) still generates a warning in the case below. How do you write these so the C macro casts into Point* and not PointEx*? (Point-length (get-PointEx)) The workaround is to write: (define Point-length (c-lambda (Point*) int "___result = length((Point*)___arg1);")) but I still want to clarify my understanding of tags and release functions. BTW, I get many more warnings. I think it is because "int(x) = 3;" isn't a correct lvalue anymore. It does work because "int" is POD, but in fact this is a constructor call for any non-POD. -- Stephane! coleste(a)videotron.ca (Stephane Le Cornec) +------------------- Made with recycled electrons. --------------------+ | #include <disclaimer.h> Kebekkujin desu. | +---------<http://www.starfiredesign.com/starfire/index.html>----------+

1 0

eval order, define vs define-macro
by Stephane Le Cornec 11 Mai '06

11 Mai '06

(define (foo x) x) (define-macro (bar x) `(+ ,x ,(foo x))) (define (baz x) (bar x)) > (load "test") *** ERROR IN #<procedure #2>, "test.scm"@2.1 -- Unbound variable: foo 1> I was a little surprised of the result. OTOH: (define (baz2 x) (bar2 x)) (define-macro (bar2 x) `(+ ,x 2)) fails on (baz2 2) as expected while the reverse works properly. I don't understand why the eval order does not work for define-macro. Is the behavior an implementation choice? And if so why is it wanted? -- Stephane! coleste(a)videotron.ca (Stephane Le Cornec) +------------------- Made with recycled electrons. --------------------+ | #include <disclaimer.h> Kebekkujin desu. | +---------<http://www.starfiredesign.com/starfire/index.html>----------+

2 3

undefined symbol environ when creating a shared object on mac osx 10.3.9
by Eric Merritt 09 Mai '06

09 Mai '06

I have a work around so it's not a big deal, but I thought I should report it. When compiling a loadable library on mac osx 10.3.9 using the following command-line gcc -bundle -o example-program foo.o _example-program.o -L/usr/local/Gambit-C/lib -lgambc The result is ld: Undefined symbols: _environ The work around is to change the command line options gcc -bundle -flat_namespace -undefined suppress -o example-program foo.o _example-program.o -L/usr/local/Gambit-C/lib -lgambc Which compiles just fine, but doesn't quite feel right. After much research I happened across http://wiki.opendarwin.org/index.php/Fink:Porting_Notes#The__environ_symbol which details the actual problem is and how to go about fixing it. I may poke around in the gambit source and see what I can see.

1 0

define-type and id
by Eric Merritt 08 Mai '06

08 Mai '06

Guys, I ask this out of curiosity more then anything else. In the define-type statements I see have what looks like a generated id field. How does this id field get generated? If its not generated how does it get calculated? Thanks, Eric

2 2

(c-define-type array-int "std::vector<int>")
by Stephane Le Cornec 05 Mai '06

05 Mai '06

Gambit does not tolerate :: or <> in this form. Is there any reason why this wouldn't work, other than Gambit accepting the string? I could typedef std::vector<int> into std_vector_int for that, but I lose genericity. The goal here is to use macro to generate interface to native C struct and handle arrays of structs. I seems much nicer to use lambda than to define the 8 conversion defines. Has anyone used in Gambit native C calls that accept/fill struct arrays? How did you decide to handle this and why did you come to that conclusion? -- Stephane! coleste(a)videotron.ca (Stephane Le Cornec) +------------------- Made with recycled electrons. --------------------+ | #include <disclaimer.h> Kebekkujin desu. | +---------<http://www.starfiredesign.com/starfire/index.html>----------+

2 2

Fwd: [gambit-list] Path for ##include
by Marc Feeley 05 Mai '06

05 Mai '06

Begin forwarded message: > From: Marc Feeley <feeley(a)iro.umontreal.ca> > Date: May 5, 2006 10:43:06 PM GMT+02:00 > To: Eric Merritt <cyberlync(a)gmail.com> > Subject: Re: [gambit-list] Path for ##include > > On 5-May-06, at 10:30 PM, Eric Merritt wrote: > >> Is there a way to specify a set of directories that ##include will >> use to find files. After rooting around in the sources for an hour or >> so I was unable to find anything. I have been trying various >> project/library layouts and have happened across a couple that should >> work well. However, they won't work if all of there dependent headers >> must be in the same directory as the source or the gambit install >> directory. > > Do you have any suggestions? I am currently implementing a > "require" form, which has the concept of modules and I'm pondering > what would be the best way to locate the modules. I dislike search > paths (because the meaning of a program will depend on an > environment variable, which seems insecure), but maybe there is no > better way. > > Marc >

2 1

Path for ##include
by Eric Merritt 05 Mai '06

05 Mai '06

Is there a way to specify a set of directories that ##include will use to find files. After rooting around in the sources for an hour or so I was unable to find anything. I have been trying various project/library layouts and have happened across a couple that should work well. However, they won't work if all of there dependent headers must be in the same directory as the source or the gambit install directory. Thanks for your help, Eric

1 0