multithreading Quantlib?

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

multithreading Quantlib?

Ilya Mirman

Hello,

 

I am wondering - is there demand for / interest in multithreading Quantlib?

 

We at Cilk Arts just released Cilk++ (3 keywords and a runtime system that extends compute-intensive C++ apps into the multicore realm), and there's an edition available for open source projects.

 

http://www.cilk.com/multicore-blog/bid/7775/Cilk-for-Linux-a-great-stocking-stuffer

 

Thank you,

 

ilya

 

 

Ilya Mirman

Cilk Arts, Inc.

55 Cambridge Street | Burlington | MA | 01803 | USA

Tel: 781-725-2455 x709 | Mobile: 978-460-1002 | Fax: 781-253-0280 |

www.cilk.com

 

Cilk++ for Linux is NOW AVAILABLE


------------------------------------------------------------------------------

_______________________________________________
QuantLib-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-dev
Reply | Threaded
Open this post in threaded view
|

Re: multithreading Quantlib?

Luigi Ballabio
On Mon, 2009-01-05 at 19:40 -0500, Ilya Mirman wrote:
> I am wondering - is there demand for / interest in multithreading
> Quantlib?

Hi Ilya,
        there is interest, but not the necessary expertise among the main
developers.  It would be good to have the option, though, in case any
developer with the skills were to come onboard.


> We at Cilk Arts just released Cilk++ (3 keywords and a runtime system
> that extends compute-intensive C++ apps into the multicore realm), and
> there's an edition available for open source projects.

It does sounds interesting, but on your site
(<http://www.cilk.com/home/get-cilk>) I read that "If you are building
open source software and sharing your derivative works with the
community, the CAPL operates like the GPL."  We'd have a problem with
that, as the QuantLib license is a modified BSD license and it's not
compatible with the GPL.  Did you happen to have this problem with other
projects? Did you solve it, and if so, how?

Thanks,
        Luigi


--

Anyone who says he can see through women is missing a lot.
-- Groucho Marx



------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
QuantLib-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-dev
Reply | Threaded
Open this post in threaded view
|

Re: multithreading Quantlib?

Ilya Mirman
Hi Luigi,

Regarding the expertise: Cilk++ is pretty easy to learn.  We've had high
school students learn it in a couple days.  We'd be glad to put together a
web-based live class for the Quantlib developers if that's of interest.
And, if/when you start applying Cilk++ to the library, we can help along the
way a bit as well.

Regarding the licensing:  The essence of the CAPL license is that for open
source projects that are willing to share their code, it's free and open.
If the end user is not willing to share the resulting code, then they can
buy a commercial license.

Cheers,
ilya

-----Original Message-----
From: Luigi Ballabio [mailto:[hidden email]]
Sent: Tuesday, January 13, 2009 4:17 AM
To: Ilya Mirman
Cc: [hidden email]
Subject: Re: [Quantlib-dev] multithreading Quantlib?

On Mon, 2009-01-05 at 19:40 -0500, Ilya Mirman wrote:
> I am wondering - is there demand for / interest in multithreading
> Quantlib?

Hi Ilya,
        there is interest, but not the necessary expertise among the main
developers.  It would be good to have the option, though, in case any
developer with the skills were to come onboard.


> We at Cilk Arts just released Cilk++ (3 keywords and a runtime system
> that extends compute-intensive C++ apps into the multicore realm), and
> there's an edition available for open source projects.

It does sounds interesting, but on your site
(<http://www.cilk.com/home/get-cilk>) I read that "If you are building
open source software and sharing your derivative works with the
community, the CAPL operates like the GPL."  We'd have a problem with
that, as the QuantLib license is a modified BSD license and it's not
compatible with the GPL.  Did you happen to have this problem with other
projects? Did you solve it, and if so, how?

Thanks,
        Luigi


--

Anyone who says he can see through women is missing a lot.
-- Groucho Marx



------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
QuantLib-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-dev
Reply | Threaded
Open this post in threaded view
|

Re: multithreading Quantlib?

Luigi Ballabio
Hi Ilya,

On Tue, 2009-01-13 at 08:14 -0500, Ilya Mirman wrote:
> Regarding the expertise: Cilk++ is pretty easy to learn.  We've had high
> school students learn it in a couple days.

Yes, I'm sure of that---if we're talking about syntax, that is.
My problem is that I still don't have the mindset for writing
multithreaded code (nor, unfortunately, the time to learn it right now.)
But I'd be happy if some other developers wanted to have a go at it.


> Regarding the licensing:  The essence of the CAPL license is that for open
> source projects that are willing to share their code, it's free and open.
> If the end user is not willing to share the resulting code, then they can
> buy a commercial license.

But this would change the QuantLib license.  As of now, the end user can
use or modify QuantLib without having to share the resulting code.
Including Cilk++ code would prevent him from doing so.  It's the same
problem that forces us not to use, say, the GNU multi-precision library.
Am I correct?

Luigi


--

Of course, if people could just act consistently, they could keep
their desks clean, avoid cavities, lose weight, give up smoking,
play a musical instrument, and possibly even produce software on
a regular and timely basis.
-- Alistair Cockburn



------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
QuantLib-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-dev
Reply | Threaded
Open this post in threaded view
|

Re: multithreading Quantlib?

Ilya Mirman
Hi Luigi,

With Cilk++, you maintain the serial semantics of your code.  Among other
things, this means that you can go between the multithreaded version and the
serial original with a couple keystrokes.  So if someone's not interested in
the multithreaded version of Quantlib deployed under CAPL or the commercial
license, just compile it with GCC instead of the Cilk++ compiler (which
consists of our additions to the GCC 4.2 compiler).

Cheers,
ilya


-----Original Message-----
From: Luigi Ballabio [mailto:[hidden email]]
Sent: Tuesday, January 13, 2009 9:08 AM
To: Ilya Mirman
Cc: [hidden email]
Subject: RE: [Quantlib-dev] multithreading Quantlib?

Hi Ilya,

On Tue, 2009-01-13 at 08:14 -0500, Ilya Mirman wrote:
> Regarding the expertise: Cilk++ is pretty easy to learn.  We've had high
> school students learn it in a couple days.

Yes, I'm sure of that---if we're talking about syntax, that is.
My problem is that I still don't have the mindset for writing
multithreaded code (nor, unfortunately, the time to learn it right now.)
But I'd be happy if some other developers wanted to have a go at it.


> Regarding the licensing:  The essence of the CAPL license is that for open
> source projects that are willing to share their code, it's free and open.
> If the end user is not willing to share the resulting code, then they can
> buy a commercial license.

But this would change the QuantLib license.  As of now, the end user can
use or modify QuantLib without having to share the resulting code.
Including Cilk++ code would prevent him from doing so.  It's the same
problem that forces us not to use, say, the GNU multi-precision library.
Am I correct?

Luigi


--

Of course, if people could just act consistently, they could keep
their desks clean, avoid cavities, lose weight, give up smoking,
play a musical instrument, and possibly even produce software on
a regular and timely basis.
-- Alistair Cockburn



------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
QuantLib-dev mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-dev
Reply | Threaded
Open this post in threaded view
|

Re: multithreading Quantlib?

hari001
In reply to this post by Ilya Mirman
Hi,

As newly registered user i only wanted to say hi to everyone else who uses this forum. I would like to know more about it.

Thanks
Regards
Jimmy


Equity Loans