Thanks Kristoffer - I didn't use your repo, but got an example working using ccall.
On Tuesday, March 29, 2016 at 6:37:43 PM UTC-4, Kristoffer Carlsson wrote: > > I meant https://github.com/JuliaSparse/MKLSparse.jl/pull/7 > > On Tuesday, March 29, 2016 at 10:22:58 PM UTC+2, Kristoffer Carlsson wrote: >> >> Look at https://github.com/JuliaSparse/MKLSparse.jl >> >> It is quite unfinished but it has cscmv! wrapped for example. You might >> need https://github.com/JuliaSparse/MKLSparse.jl/issues/8 >> >> On Tuesday, March 29, 2016 at 10:11:41 PM UTC+2, Serban Andrei Stan wrote: >>> >>> I'm trying to get Julia working with the MKL library for fast sparse >>> matrix multiplications. >>> >>> I managed to compile a version of Julia-0.4.3 on an Ubuntu 15.10 64 bit >>> machine with MKL (Base.blas_vendor() gives :mkl, and Base.libblas_name >>> gives libmkl_rt). >>> >>> However, I can't seem to find the routines for sparse matrix >>> multiplication. I haven't found any documentation online about these, and >>> in some older posts >>> <https://groups.google.com/forum/#!topic/julia-users/oBokPh6019A> I saw >>> the syntax to be Base.LinAlg.SparseBLAS.cscmv!. However, my Base.LinAlg >>> library has no module SparseBlas. >>> >>> Did I maybe miss something during installation? I installed MKL via the >>> intel parallel studio with all custom options, and my Make.user file looks >>> like this: >>> >>>> USEICC = 1 >>>> >>>> USEIFC = 1 >>>> >>>> USE_INTEL_MKL = 1 >>>> >>>> USE_INTEL_MKL_FFT = 1 >>>> >>>> USE_INTEL_LIBM = 1 >>>> >>>> >>>> MKLROOT = /opt/intel/mkl >>>> >>> >>> >>> >>
