Re: [boost] GSOC 2013

Hi David and Bjorn, David -That will really be nice to refactor. Considering its use in a lot of places like boost Geometry operations, it would really help. Bjorn - Agreed. What I was thinking is if any one out here has a similar implementation with boost.asio support and license. Apache QPID is there, but I feel it is kind of little heavy weight. For small stuff it may be a overkill. ~Shakti On Fri, Mar 22, 2013 at 7:28 PM, David Bellot wrote:

Hi,

for uBlas, the main project is indeed to improve the current framework to allow the integration of SSE, GPU, multi-core etc... and the make writing new algorithms easy. I want also a unified representation of vector and matrices. So far, we are like in R, a vector is a array in the C-style, while a matrix is really a class representing thanks to a 2-dimensional array, the notion of a mathematical matrix. In other libraries they only have one base representation for all of this.

We have a lot of duplicate code and I don't see the difference between a 1-row or 1-column matrix and a vector (at least not in my math textbook).

The advantage to factorize code is obviously to be able to apply in a uniform way and with less work all these techniques like SSE, etc...

Cheers, David

On Fri, Mar 22, 2013 at 7:59 AM, Shakti Misra wrote:

...

Hi All, Nice to see so many stuff going on. I have a few questions, if any effort is going on in this area then it should be highlighted. 1. For uBlas and some other libraries. Is there any enhancements planned for using SSE etc? Like Eigen has. 2. I read there is interest for bench-marking framework. Can there be some projects for performance counters. Especially using hardware performance counters? 3. Some transportation layer like ZeroMQ on top of boost Asio. ~Thanks Shakti

On Thu, Mar 21, 2013 at 7:12 PM, Gottlob Frege

...

wrote:

...

On Mon, Mar 4, 2013 at 10:42 AM, Mateusz Loskot wrote:

...

...

The best thing is that you guys write a nice description on the web

On 4 March 2013 15:28, David Bellot wrote: page for

...

students to know about it. So far, we already 3 of them. Great !

It may look overwhelming for students too. As it has been pointed already, few projects that qualify to Boost are actually feasible within GSoC time frame.

Something feasible could possibly be done with moving pieces of Adobe's ASL into Boost. http://stlab.adobe.com/index.html

Sean Parent has always encouraged Boost to take whatever they wanted from ASL.

Tony

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost