Pritam,

I your problem should be sent to the main R list [1] or even the R-SIG-HPC
[2] since it is more related to the mmap package than the spatial aspect of
rasters. So you might have more chances of receiving an answer there.

1 : https://stat.ethz.ch/mailman/listinfo/r-help
2 : https://stat.ethz.ch/mailman/listinfo/r-sig-hpc

Etienne

2012/9/24 Pritam Sukumar <[email protected]>

> Hello,
>
> I am a graduate student at the University of Illinois working on processing
> raster files in R.
>
> We are trying to parallelize the process and are working towards enabling
> asynchronous writes. To this end, we are using the 'mmap' package created
> by you. I had an issue I was hoping you could help me with:
>
> We are using the 'snow' package and using 'mmap' to map chunks of the image
> in different clusters. The relevant snippet of the code is below:
>
> *clusterMap(cl,function(fun,i,args,x,tr,filename,outbands) *
> *{*
> *  # Stuff to split raster into independent chunks*
> *  out <- mmap(filename,mode=real64())*
> *  # Process chunk of file*
> *  munmap(out)*
> *  return(NULL)*
> *
>
> },i,MoreArgs=list(fun=fun,x=x,tr=tr,args=args,filename=filename,outbands=outbands))
> *
> *
> *
> What it's meant to do is send different chunks to different nodes of a
> cluster (or to different cores of a single CPU, which is what I'm presently
> doing) runs perfectly fine on linux, so I don't think it's a problem with
> the way the chunks are created since they are designed to be completely
> independent of each other. The error message I get is:
>
> *Error in checkForRemoteErrors(val) : *
> *  2 nodes produced errors; first error: invalid mmap pointer*
>
> The error does not occur if I only 'mmap' and 'munmap' and do not do any
> operations on the file. Strangely enough, the code works on Linux, but not
> on Windows. From the source for 'mmap' I found that the linux backed in the
> Unix 'mmap' while on Windows, the backend is 'MapViewOfFile', which leads
> me to believe that the issue stems from a difference in the underlying OS.
> My adventures with R are only beginning, so I was hoping someone could shed
> more light on this matter.
>
> Thank you very much!
>
> -Pritam
>
>         [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> [email protected]
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-Geo mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/r-sig-geo

Reply via email to