Intel® Math Kernel Library

Convolution - border handling

I'm using VSL to do a 2D image convolution (based on FFT). The documention doesn't seem to discuss how to handle image borders. From the convolution output, it looks like the implementation is padding the borders with zero (black pixels).

Is there a way to change its behavior to repeat the border pixels? Or to mirror the image?

If not, how can I achieve this? Do I have to manually pad my input data? Or does the FFT library in MKL provide a more efficient way for me to do this?

Thanks.

MKL New User questions

I am a newbie to MKL and am trying out the 10.0.011 FFT routines with gcc as my compiler. My PC is a Intel Core 2 PC, and indeed MKL detects that max threads can be 2. My test code is not threaded.

I've run FFTs ranging from 8192 points to 262144 points. When the batch size is 1, and I use mkl_set_num_threads to change the possible thread number, I do not see any performance change. I've tried 1,2 and 4 thread settings.

FFT routines and threading

Hello,

I am a newbie to MKL and am trying out the 10.0.011 FFT routines with gcc as my compiler. My PC is a Intel Core 2 PC, and indeed MKL detects that max threads can be 2. My test code is not threaded.

I've run FFTs ranging from 8192 points to 262144 points. When the batch size is 1, and I use mkl_set_num_threads to change the possible thread number, I do not see any performance change. I've tried 1,2 and 4 thread settings.

gerfs

In linking to mkl_lapack95.lib this shows:

Error1 Error: There is no matching specific subroutine for this generic subroutine call. [GERFS]

The preceeding getrf and getrs are resolved. What's with gerfs?

Thanks,

Gerry

How to use dgeevx?

Dear all,

I am a beginer of LAPACK routines. Now I am trying to use dgeevx for computing the eigenvalues and left and right eigenvectors of a real nonsymmetric matrix on Linux. I set the parameters for dgeevx as follows:

balanc='B'

jobvl='V'

jobvr='V'

sense='B'

Moreover, I set the other parameters like lda, ldvl, ldvr, lwork following the description of the Math Kernel Library Reference Manual. The size of the matrix is 500*500.

If I run the program, my computer returns the following error:

-------------

Error in building custom DLL

I am trying to build a custom DLL in WinXP 32-bit using "nmake" comes with Visual Studio 2005. But I got the following error. Does anybody know what is going on? Thanks!

C:LibsIntelMKL10.0.1.015 oolsuilder>nmake ia32 interface=stdcall

Microsoft Program Maintenance Utility Version 8.00.50727.762
Copyright (C) Microsoft Corporation. All rights reserved.

if exist . emp_cdll rd /q /s . emp_cdll
NMAKE : fatal error U1077: '"C:Program FilesMicrosoft Visual Studio 8VCinlib.EXE"' : return code '0xc0000135' Stop.

Different answers between MKL and Netlib Lapack

Hi Everyone,

I am getting different results between MKL and netlibs LAPACK implementation. I am solving a large constrained linear system (solving Ax=B subject to Cx=D, where D = vector of zeros, A is on the order of 60K X 30 and is block-diagonal).

The results that I get from netlib LAPACK are:

1.001544335038921
-0.000754480410948
-0.118852846248803
-0.002819465096284
0.998960344715003
0.750248126784616

The result from MKL is:

Subscribe to Intel® Math Kernel Library