[Rd] loading RBlas.dll

Prof Brian Ripley ripley at stats.ox.ac.uk
Mon Jan 26 09:10:46 CET 2009


So you don't need this: you just need to replace Rblas.dll.

DLLs can contain initialization code: that is standard Windows' 
practice so please read up on it.

On Sat, 24 Jan 2009, sachin1234 wrote:

>
> hi Antonio,
>
> The name's Sachin :) . I need this because im not implementing the ATLAS,
> fast BLAS or anything like that. What im trying to do is to make a GPU
> version of blas. I have implemented this so far but extremely inefficiently.
> What i need to do is to do things like initialisation of the GPU hardware as
> soon as Rblas.dll gets loaded.
>
> Whats happening so far is something like this:
>
> GPU_DGEMM (arguments)}/*this is a blas function*/
>    initialise_GPU;
>    /*rest of code*/
>    return;
> }
>
> sorry for not being clear
> sachin
>
>
> Antonio, Fabio Di Narzo wrote:
>>
>> hi, whoever you are (I've seen no signature),
>>
>> 2009/1/22 sachin1234 <Sachinthaka.Abeywardana at csiro.au>:
>>>
>>> hi everyone,
>>>
>>> Im implementing a new RBlas dll but just for optimising sake i need to
>>> know
>>> how and WHERE (which source code) RBlas gets loaded. there is a call to
>>> loadlibrary() in dynload.c BUT it seems this is not the function that is
>>> loading Rblas.
>>
>> Why do you need this? The R-admin manual explains how to replace the R
>> built-in version of Blas with a custom, optimized one. Is this what
>> are you looking for? One of the methods mentioned here doesn't even
>> require re-compilation of R.
>>
>> Bests,
>> f.
>>
>>>
>>> how do i know this? well i added the code :(just after the loadlibrary()
>>> call)
>>>        /*********************************************/
>>>        /*this section writes into files which libraries were loaded*/
>>>        file=fopen("libraries.txt","a");
>>>        fputs(path,file);
>>>        fputs("\n",file);
>>>        fclose(file);
>>>        /******************************************/
>>> so that it prints out the path names of which ever libraries are loaded
>>> and
>>> the output is:
>>> C:/DOCUME~1/abe044/MYDOCU~1/PROJEC~1/GPU_BLAS/R-28~1.1/library/methods/libs/methods.dll
>>> C:/DOCUME~1/abe044/MYDOCU~1/PROJEC~1/GPU_BLAS/R-28~1.1/library/grDevices/libs/grDevices.dll
>>> C:/DOCUME~1/abe044/MYDOCU~1/PROJEC~1/GPU_BLAS/R-28~1.1/library/stats/libs/stats.dll
>>>
>>> and i think i got tools.dll to load as well. I cant find where R.dll,
>>> Rblas.dll etc gets loaded. My main focus is Rblas.dll
>>>
>>> now heres what i expect. Whenever you open R it runs one of its script
>>> files
>>> which contains dyn.load and loads it. HOWEVER as far as i understand
>>> dyn.load must be using dynload.c. Hence it should spit loading blas into
>>> libraries.txt which it doesnt do.
>>>
>>> I am stumped. I've spend nearly 2 days on this. any help would be
>>> appreciated
>>>
>>>
>>> --
>>> View this message in context:
>>> http://www.nabble.com/loading-RBlas.dll-tp21617712p21617712.html
>>> Sent from the R devel mailing list archive at Nabble.com.
>>>
>>> ______________________________________________
>>> R-devel at r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>>
>>
>>
>>
>> --
>> Antonio, Fabio Di Narzo
>> Ph.D. student at
>> Department of Statistical Sciences
>> University of Bologna, Italy
>>
>> ______________________________________________
>> R-devel at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>
>>
>
> --
> View this message in context: http://www.nabble.com/loading-RBlas.dll-tp21617712p21646457.html
> Sent from the R devel mailing list archive at Nabble.com.
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595



More information about the R-devel mailing list