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

Floating-point codes for testing register allocation
by Bradley Lucier 03 Avr '07

03 Avr '07

Marc: I've included below two files that I hope will eventually replace the fft bignum multiplication routines in _num.scm. (The advantage of the new approach is that it is provably correct; I'm waiting to see if there are any disadvantages.) The main loops in direct-fft-recursive and inverse-fft-recursive are scheduled to take 122 and 96 cycles, respectively, on my G5 using CC='gcc -mcpu=970 -m64' with the official Apple gcc 4.0.1; most of the time is taken in constantly reloading W and a from the stack. (You can find this information by passing -dR to gcc. BTW, why does b21 erase the .c file after compiling something with -dynamic? It's almost as if you're embarrassed by the C code that gsc generates ;-) If the number of registers are increased so that W and a are kept in registers, then these two loops are each scheduled in about 31 cycles, the same as a hand rewrite of the code in C. Please see whether some type of SSA form for GVM variables used in a basic block improves performance the same way that generating SSA- type code for F64 variables in each basic block helped the floating- point code. Some trivial timing comparisons with the existing code are given below. I hope that when I'm done the new code will be faster than the old. (For example, the new code is more "cache-oblivious" than the old.) Brad > [descartes:~/programs/gambc40b17/fftmul] lucier% gsc -:m100000 > Gambit Version 4.0 beta 21 > > > (load "fftmul-fast") > "/Users/lucier/programs/gambc40b17/fftmul/fftmul-fast.o2" > > (define a (expt 3 10000000)) > > (define b (expt 3 10000000)) > > (define c (time (* a a))) > (time (* a a)) > 2193 ms real time > 2090 ms cpu time (1948 user, 142 system) > 1 collection accounting for 22 ms real time (3 user, 19 system) > 110229552 bytes allocated > no minor faults > no major faults > > (define d (time (fft-mul a a))) > (time (fft-mul a a)) > 2621 ms real time > 2431 ms cpu time (2203 user, 228 system) > 1 collection accounting for 44 ms real time (4 user, 40 system) > 112555696 bytes allocated > no minor faults > no major faults > > (define e (time (* a b))) > (time (* a b)) > 3659 ms real time > 3002 ms cpu time (2777 user, 225 system) > 1 collection accounting for 77 ms real time (3 user, 74 system) > 171743360 bytes allocated > no minor faults > no major faults > > (define f (time (fft-mul a b))) > (time (fft-mul a b)) > 4007 ms real time > 3585 ms cpu time (3219 user, 366 system) > 1 collection accounting for 64 ms real time (3 user, 61 system) > 183596720 bytes allocated > no minor faults > no major faults > > (= c d) > #t > > (= e f) > #t

1 0

Speeding up rational arithmetic
by Bradley Lucier 03 Avr '07

03 Avr '07

Marc: I thought that I might speed up rational arithmetic a bit in the common case where the numerators and denominators are small integers, so I tried some new declarations: (define-prim (##ratnum.* x y) (##declare (run-time-bindings) (mostly-fixnum)) (let ((p (macro-ratnum-numerator x)) (q (macro-ratnum-denominator x)) (r (macro-ratnum-numerator y)) (s (macro-ratnum-denominator y))) (if (##eq? x y) (macro-ratnum-make (* p p) (* q q)) ; already in lowest form (let* ((gcd-ps (gcd p s)) (gcd-rq (gcd r q)) (num (* (quotient p gcd-ps) (quotient r gcd-rq))) (den (* (quotient q gcd-rq) (quotient s gcd-ps)))) (if (##eq? den 1) num (macro-ratnum-make num den)))))) but this isn't quite right, since, e.g., for the form (* p p), if p is not a fixnum, then gsc generates calls to (* p p) instead of (##* p p) even though (declare (standard-bindings) is in effect. So a user could screw up rational arithmetic simply by redefining * (which is allowed, after all). Perhaps the code generator should be changed to call (##* p p) directly in this case. The full expansion is below. Brad (define ##ratnum.* (lambda (x y) (let ((s ('#<procedure #7 ##vector-ref> y 1)) (r ('#<procedure #7 ##vector-ref> y 0)) (q ('#<procedure #7 ##vector-ref> x 1)) (p ('#<procedure #7 ##vector-ref> x 0))) (if ('#<procedure #27 ##eq?> x y) ('#<procedure #75 ##subtype-set!> ('#<procedure #76 ##vector> (if (and ('#<procedure #14 ##fixnum?> p) ('#<procedure #14 ##fixnum?> p)) (if ('#<procedure #298 ##eqv?> p 0) 0 (let ((temp.236 (if ('#<procedure #298 ##eqv?> p -1) ('#<procedure #207 ##fx-?> p) ('#<procedure #208 ##fx*?> p p)))) (if temp.236 temp.236 (* p p)))) (* p p)) (if (and ('#<procedure #14 ##fixnum?> q) ('#<procedure #14 ##fixnum?> q)) (if ('#<procedure #298 ##eqv?> q 0) 0 (let ((temp.239 (if ('#<procedure #298 ##eqv?> q -1) ('#<procedure #207 ##fx-?> q) ('#<procedure #208 ##fx*?> q q)))) (if temp.239 temp.239 (* q q)))) (* q q))) 2) (let ((gcd-ps ('#<procedure #94 gcd> p s))) (let ((gcd-rq ('#<procedure #94 gcd> r q))) (let ((num (let ((temp.245 (if (and ('#<procedure #14 ##fixnum? > gcd-rq) (and ('#<procedure #14 ##fixnum?> r) ('#<procedure #21 ##not> ('#<procedure #298 ##eqv?> gcd-rq 0)))) (if ('#<procedure #298 ##eqv?> gcd-rq -1) (or ('#<procedure #207 ##fx-? > r) (quotient r gcd-rq)) ('#<procedure #216 ##fxquotient> r gcd-rq)) (quotient r gcd-rq))) (temp.244 (if (and ('#<procedure #14 ##fixnum? > gcd-ps) (and ('#<procedure #14 ##fixnum?> p) ('#<procedure #21 ##not> ('#<procedure #298 ##eqv?> gcd-ps 0)))) (if ('#<procedure #298 ##eqv?> gcd-ps -1) (or ('#<procedure #207 ##fx-? > p) (quotient p gcd-ps)) ('#<procedure #216 ##fxquotient> p gcd-ps)) (quotient p gcd-ps)))) (if (and ('#<procedure #14 ##fixnum?> temp.245) ('#<procedure #14 ##fixnum?> temp.244)) (if ('#<procedure #298 ##eqv?> temp. 245 0) 0 (let ((temp.246 (if ('#<procedure #298 ##eqv?> temp.245 -1) ('#<procedure #207 ##fx-?> temp.244) ('#<procedure #208 ##fx*?> temp.244 temp.245)))) (if temp.246 temp.246 (* temp.244 temp.245)))) (* temp.244 temp.245))))) (let ((den (let ((temp.252 (if (and ('#<procedure #14 ##fixnum?> gcd-ps) (and ('#<procedure #14 ##fixnum?> s) ('#<procedure #21 ##not> ('#<procedure #298 ##eqv?> gcd-ps 0)))) (if ('#<procedure #298 ##eqv?> gcd-ps -1) (or ('#<procedure #207 ##fx-?> s) (quotient s gcd-ps)) ('#<procedure #216 ##fxquotient> s gcd-ps)) (quotient s gcd-ps))) (temp.251 (if (and ('#<procedure #14 ##fixnum?> gcd-rq) (and ('#<procedure #14 ##fixnum?> q) ('#<procedure #21 ##not> ('#<procedure #298 ##eqv?> gcd-rq 0)))) (if ('#<procedure #298 ##eqv?> gcd-rq -1) (or ('#<procedure #207 ##fx-?> q) (quotient q gcd-rq)) ('#<procedure #216 ##fxquotient> q gcd-rq)) (quotient q gcd-rq)))) (if (and ('#<procedure #14 ##fixnum?> temp.252) ('#<procedure #14 ##fixnum?> temp.251)) (if ('#<procedure #298 ##eqv?> temp. 252 0) 0 (let ((temp.253 (if ('#<procedure #298 ##eqv?> temp.252 -1) ('#<procedure #207 ##fx-?> temp.251) ('#<procedure #208 ##fx*?> temp.251 temp.252)))) (if temp.253 temp.253 (* temp.251 temp.252)))) (* temp.251 temp.252))))) (if ('#<procedure #27 ##eq?> den 1) num ('#<procedure #75 ##subtype-set!> ('#<procedure #76 ##vector> num den) 2))))))))))

1 1

Gambit-C submitted to MacPorts
by Arto Bendiken 02 Avr '07

02 Avr '07

I've submitted Gambit as a proposed new port [1] for inclusion in the MacPorts packaging system [2]. Once the port is accepted, installing Gambit on a Mac OS X system will be as simple as: # sudo port install gambit-c I had to do quite a bit of patching [3] of the installation paths in the makefiles to comply with the MacPorts directory structure. Perhaps some of the changes, like prefixing installation paths with $(DESTDIR), could be integrated into future Gambit versions to ensure easier interoperability with various packaging systems including MacPorts. Cheers, Arto [1] http://trac.macports.org/projects/macports/ticket/11688 [2] http://en.wikipedia.org/wiki/MacPorts [3] http://trac.macports.org/projects/macports/attachment/ticket/11688/Portfile -- Arto Bendiken | http://bendiken.net/

2 2

Fwd: gaps in the mesh
by Oana Petrache 02 Avr '07

02 Avr '07

Note: forwarded message attached. --------------------------------- The fish are biting. Get more visitors on your site using Yahoo! Search Marketing.

1 0

Appending 32bit integers to a file
by Phil Dawes 30 Mar '07

30 Mar '07

Hi Gambit List, What's the simplest way to append 32bit unsigned integers to a file in gambit? (not bothered about maintaining endian portability or anything) I'm currently using a C hack (see below), but it feels like there ought to be something less clunky. Thanks, Phil (c-declare "#include <stdio.h>") (define fopen (c-lambda (nonnull-char-string nonnull-char-string) (pointer "FILE") "fopen")) (define fclose (c-lambda ((pointer "FILE")) int32 "fclose")) (define write-u32vector-to-file ; args: port, u32-vector, len (c-lambda ((pointer "FILE") scheme-object unsigned-int32) int32 "___result = fwrite((void*)(___BODY_AS(___arg2,___tSUBTYPED)), 4, ___arg3, ___arg1);")) (define read-u32vector-from-file ; args: port, u32-vector, len (c-lambda ((pointer "FILE") scheme-object unsigned-int32) int32 "___result = fread((void*)(___BODY_AS(___arg2,___tSUBTYPED)), 4, ___arg3, ___arg1);"))

1 0

gambit's list displaying is not very pretty
by dillo gimp 28 Mar '07

28 Mar '07

(display (list (list 1 2 ) 3)) 123 how do I properly display ((1 2) 3)? thanks

3 2

ssax-sxml for gambc-4.0b21
by Aycan iRiCAN 28 Mar '07

28 Mar '07

Here is the proper Makefile of ssax-sxml-gambit-060529.tgz for gambit 4.0 beta 21. You should change the prefix variable before using this Makefile. Best Regards, -- Aycan iRiCAN KOR Bilişim Teknolojileri http://www.core.gen.tr/

3 4

ssax-sxml for gambc-4.0b21
by Aycan iRiCAN 28 Mar '07

28 Mar '07

This a slightly modified Makefile that creates a shared-library and a loadable library for ssax-sxml-gambit-060529.tgz. Here is an example transcript for loadable ssax-sxml library. ssax-sxml $ gsi Gambit Version 4.0 beta 21 > (load "ssax-sxml") "/tmp/ssax-sxml/ssax-sxml.o1" > (ssax:xml->sxml (open-input-string "<example>Peace at home, peace in the world.</example>") '()) (*TOP* (example "Peace at home, peace in the world.")) > (sxml:document "http://library.readscheme.org" '()) (*TOP* (html (head (title "Bibliography of Scheme-related Research") (meta (@ (name "generator") (content "SchemeBib and Site-ML"))) (meta (@ (name "description") (content "Online bibliography of Scheme research"))) (meta (@ (name "keywords") (content "Scheme, bibliography, research"))) ... ... Best Regards, -- Aycan iRiCAN KOR Bilişim Teknolojileri http://www.core.gen.tr/

1 0

Re: [gambit-list] Conflict bewteen #<<c-declare-end and syntax-case?
by Raffael Cavallaro 25 Mar '07

25 Mar '07

On Feb 7, 2007, at 8:06 PM, Marc Feeley wrote: > It's probably easy to fix for the implementors of syntax-case > (Dybvig et al), but I shudder at the thought of figuring out which > of the over 40 usages of 'quote in syntax-case.scm has to be made > conditional on the object being quoted (the quote should be dropped > for strings, numbers, booleans, and characters). Given that you can't fix syntax-case.scm, would it be possible to modify c-declare, etc. so that they take both strings and quoted strings? regards, Ralph Raffael Cavallaro, Ph.D. raffaelcavallaro(a)mac.com

2 2

Commercial support?
by a Human Being 25 Mar '07

25 Mar '07

Hi guys We are looking for commercial support (in Europe) for the Gambit-C compiler and runtime system (and Termite too). Any ideas ? Chris

2 1