blas

问题 I installed Intel MKL and other libraries for a customized numpy. Here is my ~/.numpy-site.cfg : [DEFAULT] library_dirs = /usr/lib:/usr/local/lib include_dirs = /usr/include:/usr/local/include [mkl] library_dirs = /opt/intel/mkl/lib/intel64/ include_dirs = /opt/intel/mkl/include/ mkl_libs = mkl_rt lapack_libs = [amd] amd_libs = amd [umfpack] umfpack_libs = umfpack [djbfft] include_dirs = /usr/local/djbfft/include library_dirs = /usr/local/djbfft/lib This configuration file seems OK during the

问题 I run (k)Ubuntu 12.04.2 and R 3.0.1 I wrote a bunch of code that used to run in parallel, but now it no longer does. Not even this runs in parallel any more: library(doMC) registerDoMC(4) Results = foreach (i = 1:1e6, .combine = "c") %dopar% { sqrt(i) } And that definitely should. What I think broke it is either the R 3.0.1 update or a -dev, -devel BLAS package I installed. (openBLAS I think) I've tried system(sprintf("taskset -p 0xffffffff %d", Sys.getpid())) as suggested elsewhere, and get

问题 I'm currently using BLAS function DSYMM to compute Y = AX and then DGEMM for YA' , but I'm wondering is there some more efficient way of computing the matrix product AXA T , where A is an arbitrary n×n matrix and X is a symmetric n×n matrix? 来源： https://stackoverflow.com/questions/13292325/m-x-s-x-tm-multiplication-with-blas

问题 I have recently discovered that the performance of a cblas_sgemm call for matrix multiplication dramatically improves if the matrices have a "large" number of zeros in them. It improves to the point that it beats its cublas cousin by around 100 times. This could be most probably attributed to some automatic detection of sparsity and suitable format conversion by cblas_sgemm function. Unfortunately, no such behavior is exhibited by its cuda counterpart i.e. cublasSgemm. So, the question is,

问题 I am having trouble writing a MEX file in MATLAB that can perform a simple linear operation such as taking the inverse of a matrix. I have successfully managed to take the inverse of a matrix using Visual Studio 2010 and have successfully created a MEX file hence the only thing I am having trouble is getting these two concepts together. I have tried to compile a MEX example code that I got from the MathWorks site but with no luck. Here is what I have tried, Saved the file (renamed it) I got

问题 I am getting an error when i test CBLAS in CLAPACK. When i run the test code ./xblat2d < dblat2.in I get the error output: "TransA must be 111, 112 or 113, but is set to -1Parameter 2 to routine cblas_dgemv was incorrect" Does anyone know what kind of problem this indicates? Details I am trying not to use reference CBLAS, and use ATLAS CBLAS instead. So, i compiled wrapper library libcblaswr.a and changed the line in make.inc to BLASLIB = ../../libcblaswr.a -L/usr/local/atlas/lib -lcblas -L

问题 GotoBLAS is an old package but some wonderful open source projects are depending upon it. I obtainded it from github.com: https://github.com/dofi/gotoBLAS/tree/master/Downloads But I encountered the following compiling error (wrapped for legibility), gcc -O2 -DEXPRECISION -m128bit-long-double -Wall -m64 -DF_INTERFACE_GFORT -fPIC -DSMP_SERVER -DMAX_CPU_NUMBER=2 -DASMNAME= -DASMFNAME=_ -DNAME=_ -DCNAME= -DCHAR_NAME=\"_\" -DCHAR_CNAME=\"\" -I.. -w -o linktest linktest.c ../libgoto2_barcelonap-r1

问题 I have been searching for decent documentation on blas, and I have found some 315 pages of dense material that ctrl-f does not work on. It provides all the information regarding what input arguments the routines take, but there are a LOT of input arguments and I could really use some example code. I am unable to locate any. I know there has to be some or no one would be able to use these libraries! Specifically, I use ATLAS installed via macports on a mac osx 10.5.8 and I use gfortran from

问题 I am calling a BLAS function in my code by including the BLAS library, and my code apparently is faulty somehow, as the compiler spits out the error: "ddot was not declared in this scope." #include <stdio.h> #include <stdlib.h> #include <blas.h> int main() { double m[10],n[10]; int i; int result; printf("Enter the elements into first vector.\n"); for(i=0;i<10;i++) scanf("%lf",&m[i]); printf("Enter the elements into second vector.\n"); for(i=0;i<10;i++) scanf("%lf",&n[i]); result = ddot(m,n);

问题 I am trying to do a Cholesky decomposition via pdpotrf() of MKL-Intel's library, which uses ScaLAPACK. I am reading the whole matrix in the master node and then distribute it like in this example. Everything works fine when the dimension of the SPD matrix is even. However, when it's odd, pdpotrf() thinks that the matrix is not positive definite. Could it be because the submatrices are not SPD? I am working with this matrix: and the submatrices are (with 4 processes and blocks of size 2x2): A