Different answers between MKL and Netlib Lapack

ankur28 — Fri, 18 Apr 2008 20:08:46 GMT

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:

1.00029802322388
-0.00063318014145
-0.02181071229279
-0.00281071662903
0.99918329715729
0.67531794309616

I have posted the first six entries of the result to outline the difference. The matrices are the same in both the cases.

Following are the flags that I am using for icc:

CFLAGS = -ggdb -fp-model precise -align -Zp16 -O0
INCPATH = -I/opt/intel/ipp/5.3.2.068/em64t/include -I/opt/intel/mkl/10.0.2.018/include -I/usr/include/atlas -I/usr/local/include/opencv -I/usr/share/qt4/mkspecs/linux-g++ -I. -I/usr/inclu de/qt4/QtCore -I/usr/include/qt4/QtCore -I/usr/include/qt4/QtGui -I/usr/include/qt4/QtGui -I/usr/include/qt4 -I.

The netlib answer is the correct answer in this case.

Any pointers to what might be causing this mystery difference?

Thanks,
-Ankur

Re: Different answers between MKL and Netlib Lapack

Michael_C_Intel4 — Wed, 30 Apr 2008 13:00:11 GMT

DearAnkur,

which LAPACK routines do you use to solve it?

Michael.

topic Re: Different answers between MKL and Netlib Lapack in Intel® oneAPI Math Kernel Library

Different answers between MKL and Netlib Lapack

Re: Different answers between MKL and Netlib Lapack