[R-sig-hpc] Use R on Cuda

Ei-ji Nakama nakama at ki.rim.or.jp
Tue Oct 19 05:37:44 CEST 2010


Hi,
I performed only a simple experiment.

1. wrapper for gemm
http://prs.ism.ac.jp/~nakama/cudablas_gemm/

2. make

3 LANG=C make test

$ LANG=C make test
R -q -e 'set.seed(1);N<-4e3;A<-matrix(runif(N^2),N,N);system.time(B<-A%*%A)'
| sed -e '/^>/d'
   user  system elapsed
  9.976   0.160   1.761
LD_PRELOAD=./gemm.so R -q -e
'set.seed(1);N<-4e3;A<-matrix(runif(N^2),N,N);system.time(B<-A%*%A)' |
sed -e '/^>/d'
   user  system elapsed
  0.968   0.044   1.008
R -q -e 'set.seed(1);N<-4e3;A<-matrix(as.complex(runif(N^2)),N,N);system.time(B<-A%*%A)'
| sed -e '/^>/d'
   user  system elapsed
 39.098   0.600   6.619
LD_PRELOAD=./gemm.so R -q -e
'set.seed(1);N<-4e3;A<-matrix(as.complex(runif(N^2)),N,N);system.time(B<-A%*%A)'
| sed -e '/^>/d'
   user  system elapsed
  3.388   0.068   3.454

The size that I could operate depended on RAM of GPU.


2010/10/19 Marcelo Lima <mlimagb at gmail.com>:
> Dear all,
>
> I'm trying to run R on a cuda 3X (GTX 480) using an i7 processor, however I
> haven't been successful. Any suggestions?!
>
> Thanks in advance,
>
> Marcelo
>
> --
> Marcelo Andrade de Lima
> UNIFESP - Universidade Federal de São Paulo
> Departamento de Bioquímica
> Disciplina de Biologia Molecular
> Rua Três de Maio 100, 4 andar - Vila Clementino, 04044-020
> Lab +55 11 55764438 R.1188
> Cell +55 11 92725274
> mlima at unifesp.br
>
>        [[alternative HTML version deleted]]
>
>
> _______________________________________________
> R-sig-hpc mailing list
> R-sig-hpc at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-hpc
>
>


-- 
EI-JI Nakama  <nakama (a) ki.rim.or.jp>
"\u4e2d\u9593\u6804\u6cbb"  <nakama (a) ki.rim.or.jp>



More information about the R-sig-hpc mailing list