pro September 2011

pro@common-lisp.net

6 participants
6 discussions

[pro] GPU/CUDA & common lisp
by Tamas Papp 05 Apr '12

05 Apr '12

Hi everyone, I am about to buy a new server for number crunching, and I would like to keep my options open about GPU/CUDA based computations. The most intensively parallelizable thing that I am doing is particle filtering, and currently that works fine (and is really fast) in a multi-core CPU with SBCL, but I keep hearing wonderful things about GPUs from people, and I was wondering if I could make use of them but still program in CL. I basically have two questions: - What is the current status of using CPUs for computation in CL (in particular SBCL)? All I found was CL-GPU but I haven't tried it yet (my current server does not have a GPU). - What hardware is supported (if I want to use CL)? Can I use something like Nvidia's TESCLA C line? Any help/pointers/accounts would be appreciated, I am just exploring this so I don't know much about it yet, apologies if I am not asking the right questions. Best, Tamas

4 5

[pro] Administrativia: Posting through gmame
by Hans Hübner 04 Oct '11

04 Oct '11

Hi, the post earlier today was made through gmame, and that post bounced at multiple subscriber's MTAs because it was considered spam. In order to avoid that, I'm not going to manually forward postings coming in through gmame anymore. Please consider subscribing to the mailing list directly (you can still read through gmame, but you need to post by email). Thanks, Hans

2 1

Re: [pro] help with macro
by karol skocik 26 Sep '11

26 Sep '11

Hi, thanks for help. On Mon, Sep 26, 2011 at 3:02 PM, Scott McKay <swm(a)google.com> wrote: > You're pretty much asking for type inference, but with you > being the one doing the inference. Yes, it's more or less a way to provide more type hints to SBCL. On Mon, Sep 26, 2011 at 4:28 PM, Peter Stirling <peter(a)pjstirling.plus.com> wrote: > You are misunderstanding the way that symbols work: > > First your source code is read by the reader, and symbols are looked up > by their package prefix, or interned in the current package if > unqualified. After this point the raw text that from your lisp file is > thrown away and macro-expansion and compilation is performed using only > the symbols and constants identified/produced by the reader. Since your > macro is trying to shadow symbols rather than generate new ones you only > need to worry about package prefixes at the call site of your macro, and > not in the macro itself > Thanks for explanation. I was confused thinking that reader subsequently reads more of the form as it is expanded, which is rubbish. > :use in defpackage will import all external symbols into the defined > package, and since "+" is external in :cl, and it's highly likely that > you are ":use :cl" it wouldn't need a prefix. > > > (defmacro with-typed-arguments ((name) type &body body) > (let ((func (gensym))) > `(let ((,func #',name)) > (declare (sb-ext:disable-package-locks ,name)) > (macrolet ((,name (&rest args) > (list* 'funcall ',func > (mapcar (lambda (arg) > (list 'the ',type arg)) > args)))) > (declare (sb-ext:enable-package-locks ,name)) > ,@body)))) > > > Should do the expansion you want, but it's not the "right" way in > common-lisp. > Works nicely, thank you! Why do you consider it as not "right"? Have a nice day. Karol > On Mon, 2011-09-26 at 09:31 +0200, karol skocik wrote: >> Hi Lispers, >> I need a help with a following macro I am trying to write: >> >> When the call site looks like this: >> >> (with-typed-arguments (distance) single-float >> (+ (distance a b c d) >> (distance x y z w))) >> >> I want it to expand into something like this: >> >> (+ (DISTANCE (THE SINGLE-FLOAT A) (THE SINGLE-FLOAT B) (THE >> SINGLE-FLOAT C) (THE SINGLE-FLOAT D)) >> (DISTANCE (THE SINGLE-FLOAT X) (THE SINGLE-FLOAT Y) (THE >> SINGLE-FLOAT Z) (THE SINGLE-FLOAT W)))) >> >> Now, what I have currently is: >> >> (defmacro with-typed-arguments ((&rest functions) argument-type &body body) >> (flet ((macrolet-body (fun) >> `(,fun (&rest args) >> `(funcall #',',fun ,@(iter (for arg :in args) >> (collect `(the >> ,',argument-type ,arg))))))) >> `(macrolet (,@(mapcar #'macrolet-body functions)) >> ,@body))) >> >> But it some things I don't understand: >> >> 1.) the function called with typed arguments is called using funcall - >> not a show stopper, just a style issue. Without funcall the expansion >> loops forever since the expansion does not emit the code, but expands >> deeper into infinite recursion of macrolets. Can it be fixed somehow? >> >> 2.) in the flet macrolet-body, I have a trouble understanding ,',fun >> construction next to the funcall - why the "fun" next to funcall can't >> be replaced by macrolet-body argument fun with ,,fun ? It is buried in >> `` but ,,fun there just got me (for call (macrolet-body +)) -> `(,+ >> ...). Let me show my trouble in the example: >> >> In this definition: >> (defmacro with-typed-arguments ((&rest functions) argument-type &body body) >> (flet ((macrolet-body (fun) >> `(,fun (&rest args) >> `(funcall #',,fun ,@(iter (for arg :in args) >> (collect `(the >> ,',argument-type ,arg))))))) >> `(macrolet (,@(mapcar #'macrolet-body functions)) >> ,@body))) >> >> this expansion: >> (with-typed-arguments (distance) single-float >> (+ (distance a b c d) >> (distance x y z w))) >> >> results in this: >> (MACROLET ((DISTANCE (&REST ARGS) >> `(FUNCALL #',DISTANCE >> ;; <- why is there ,DISTANCE when there is `(... `(... ,,fun >> ? >> ,@(ITER >> (FOR ARG :IN ARGS) >> (COLLECT `(THE SINGLE-FLOAT ,ARG)))))) >> (+ (DISTANCE A B C D) (DISTANCE X Y Z W))) >> >> 3.) package-locks - I want it to work with arithmetic functions >> (especially!) defined in CL package. That doesn't work, since I can't >> make a macrolet ((+ (&rest args) ...) because of package-locks. I am >> happy with SBCL specific fix, using SBCL's >> sb-ext:disable-package-locks, or sb-ext:without-package-locks, but I >> can't make it work. >> Especially, I have a trouble understanding, how do I create a symbol >> with package prefix needed in declaration like this: >> (declare (sb-ext:disable-package-locks cl::+)) >> >> Is there a way to prepend the symbol-package and '::' to the given >> symbol? How to construct such fully qualified symbol like cl::+? >> >> Thank you, >> Karol >> >> _______________________________________________ >> pro mailing list >> pro(a)common-lisp.net >> http://lists.common-lisp.net/cgi-bin/mailman/listinfo/pro > > >

1 0

[pro] help with macro
by karol skocik 26 Sep '11

26 Sep '11

Hi Lispers, I need a help with a following macro I am trying to write: When the call site looks like this: (with-typed-arguments (distance) single-float (+ (distance a b c d) (distance x y z w))) I want it to expand into something like this: (+ (DISTANCE (THE SINGLE-FLOAT A) (THE SINGLE-FLOAT B) (THE SINGLE-FLOAT C) (THE SINGLE-FLOAT D)) (DISTANCE (THE SINGLE-FLOAT X) (THE SINGLE-FLOAT Y) (THE SINGLE-FLOAT Z) (THE SINGLE-FLOAT W)))) Now, what I have currently is: (defmacro with-typed-arguments ((&rest functions) argument-type &body body) (flet ((macrolet-body (fun) `(,fun (&rest args) `(funcall #',',fun ,@(iter (for arg :in args) (collect `(the ,',argument-type ,arg))))))) `(macrolet (,@(mapcar #'macrolet-body functions)) ,@body))) But it some things I don't understand: 1.) the function called with typed arguments is called using funcall - not a show stopper, just a style issue. Without funcall the expansion loops forever since the expansion does not emit the code, but expands deeper into infinite recursion of macrolets. Can it be fixed somehow? 2.) in the flet macrolet-body, I have a trouble understanding ,',fun construction next to the funcall - why the "fun" next to funcall can't be replaced by macrolet-body argument fun with ,,fun ? It is buried in `` but ,,fun there just got me (for call (macrolet-body +)) -> `(,+ ...). Let me show my trouble in the example: In this definition: (defmacro with-typed-arguments ((&rest functions) argument-type &body body) (flet ((macrolet-body (fun) `(,fun (&rest args) `(funcall #',,fun ,@(iter (for arg :in args) (collect `(the ,',argument-type ,arg))))))) `(macrolet (,@(mapcar #'macrolet-body functions)) ,@body))) this expansion: (with-typed-arguments (distance) single-float (+ (distance a b c d) (distance x y z w))) results in this: (MACROLET ((DISTANCE (&REST ARGS) `(FUNCALL #',DISTANCE ;; <- why is there ,DISTANCE when there is `(... `(... ,,fun ? ,@(ITER (FOR ARG :IN ARGS) (COLLECT `(THE SINGLE-FLOAT ,ARG)))))) (+ (DISTANCE A B C D) (DISTANCE X Y Z W))) 3.) package-locks - I want it to work with arithmetic functions (especially!) defined in CL package. That doesn't work, since I can't make a macrolet ((+ (&rest args) ...) because of package-locks. I am happy with SBCL specific fix, using SBCL's sb-ext:disable-package-locks, or sb-ext:without-package-locks, but I can't make it work. Especially, I have a trouble understanding, how do I create a symbol with package prefix needed in declaration like this: (declare (sb-ext:disable-package-locks cl::+)) Is there a way to prepend the symbol-package and '::' to the given symbol? How to construct such fully qualified symbol like cl::+? Thank you, Karol

1 0

[pro] Quicklisp's effect on your CL habits
by Zach Beane 25 Sep '11

25 Sep '11

Hi, I've been trying to learn more about how Quicklisp has changed how Common Lisp hackers work. When I posed the question on the Quicklisp list, I got a response from Elliot Slaughter with a point that I hadn't considered before: > I think part of what makes Quicklisp successful (in my opinion) is > that both new *and old* Lisp users love and make use of it. To me, > it's really magical to be able to cd into my project, start lisp, and > quickload it, and have it just work, even if I have customized > versions of some of the dependencies. I think the extent to which > Quicklisp "just works" but also stays out of the way makes older Lisp > hackers happy, which in turn makes the community more vibrant because > library and implementation authors are using the same system that new > users are being told to use (which wasn't always true with e.g. > lisp-starter-pack or asdf-install). As I suspect the "pro" list likely has more old-timers than newcomers, I'd like to get your take on this idea. Zach

2 1

[pro] Test again: please disregard....
by Marco Antoniotti 19 Sep '11

19 Sep '11

Sorry.... -- Marco Antoniotti, Associate Professor tel. +39 - 02 64 48 79 01 DISCo, Università Milano Bicocca U14 2043 http://bimib.disco.unimib.it Viale Sarca 336 I-20126 Milan (MI) ITALY Please note that I am not checking my Spam-box anymore. Please do not forward this email without asking me first.

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

pro September 2011