Re: Octave 3.6.4 VS2010 and the C++ API

[Mike Puglia]

Thanks again for all the help.  I tried all of the libraries and the problem seems to be limited to OpenBLAS.  I've tried this on three machines now (running XP or Windows Server 2008), and so it seems to be universal. The other three BLAS libraries in the distribution (generic, Intel SSE3 and SSE3 multi-threaded) all calculated correctly, but with a hit to performance.  On the Intel libraries the performance hit was about 25% and the generic was 2x on my machine.  Most of my work involves running things that take on the order of 1 or 2 days, so this is pretty costly.

On a somewhat related note, I'm also having problems running OpenBLAS on Amazon EC2.  It seems that the virtual machines there don't support AVX, and so OpenBLAS is defaulting to a backup (Nehalem?) that can run without AVX, which causes a performance hit of about 2x.

Do you see any alternatives to using OpenBLAS here (ATLAS maybe?) or can you suggest a fix?  Any advice you have would be greatly appreciated. 

---

From: Michael Goffioul <address@hidden>
To: Mike Puglia <address@hidden>
Cc: "address@hidden" <address@hidden>
Sent: Thursday, September 26, 2013 12:05 AM
Subject: Re: Octave 3.6.4 VS2010 and the C++ API

On Wed, Sep 25, 2013 at 10:55 AM, Mike Puglia <address@hidden> wrote:

Thanks for the help.  It is much appreciated.  I installed OpenBlas with both builds, which seemed to be recommended.  I'll try some of the other libraries out and see if they work.  It seems I might need to reinstall though, as only one library can be chosen during installation.  Do you know of an easier way?

You need to reinstall.

 

Regarding the m-files, I don't think it is working at the command line.  I get the same behavior as with the oct file.  2x2 matrix multiplication works but the special case does not.  The following snippet works on the MINGW build, but not VS2010:

e0 = complex(0,0)

e1 = complex(1,1)

a = [e0 e1; e0 e1]

b = [e0 e0; e1 e1]

c = a*b

d = a(1,:)*b(:,1)

The octave output on the VS2010 build is:

c =

   0 + 2i   0 + 2i

   0 + 2i   0 + 2i

d =  0.0000e+000 + 5.7929e-195i

OK, so at least the error is consistent.

Michael.