LAPACK Archives

[Lapack] Fwd: LAPACK and INTEL MKL

(  solved  )

Begin forwarded message:

From: NelsonJT <nelsonjt@Domain.Removed<mailto:nelsonjt@Domain.Removed>>
Subject: RE: [Lapack] LAPACK and INTEL MKL
Date: May 6, 2014 at 6:17:56 AM MDT
To: "'Kobotov, Alexander V'" 
 "Langou, Julien" 
Cc: "'Kuznetsov, Sergey V'" 
"'Mozartova, Nadezhda'" 
"'Arturov, Konstantin I'" 
Reply-To: "nelsonjt@Domain.Removed<mailto:nelsonjt@Domain.Removed>" 


I was in error!!!  (Late night programming!)

Intel Math Kernel Library 11.0.5 Reference Manual:

call dgbsvxx( fact, trans, n, kl, ku, nrhs, ab, ldab, afb, ldafb, ipiv, equed, 
r, c, b, ldb, x, ldx, rcond, rpvgrw, berr, n_err_bnds, 
err_bnds_norm,err_bnds_comp, nparams, params, work, IWORK, info )

call cgbsvxx( fact, trans, n, kl, ku, nrhs, ab, ldab, afb, ldafb, ipiv, equed, 
r, c, b, ldb, x, ldx, rcond, rpvgrw, berr, n_err_bnds, 
err_bnds_norm,err_bnds_comp, nparams, params, work, RWORK, info )

I see that DGBSVXX uses the INTEGER working vector IWORK, and CGBSVXX uses the 
REAL working vector RWORK in both MKL and LAPACK_3.5.0.  I was confused and 
stand corrected.  I did not expect the subroutines to be different.

EQUED is declared CHARACTER in both LAPACK_3.5.0 and MKL packages as well!!!  
Must have been dreaming when coding.


Call to CGBSVXX with INTEGER IWORK and LOGICAL EQUED compiled and linked 
without glitch in INTEL FORTRAN, despite error in declarations.  Program ran 
without error as well, as I suppose it should.

COMPAQ DIGITAL Fortran 6.6 linker tripped on LOGICAL EQUED when linking CGBSVXX 
from LAPAC_3.5.0.  Would not recognize CGBSVXX.  In any case, unable to link 
code due to missing BLAS routines.  No longer attempting to use LAPAC_3.5.0 
with CVF.

Thank you for correcting me!!!

No need to respond.

James Nelson
Wilson, Ihrig & Associates
6001 Shellmound, Suite 400
Emeryville, CA 94608

From: Kobotov, Alexander V [mailto:alexander.v.kobotov@Domain.Removed]
Sent: Tuesday, May 06, 2014 1:04 AM
To: Langou, Julien; nelsonjt@Domain.Removed<mailto:nelsonjt@Domain.Removed>
Cc: Kuznetsov, Sergey V; Mozartova, Nadezhda; Arturov, Konstantin I
Subject: RE: [Lapack] LAPACK and INTEL MKL

Hi Nelson,

Julien, thanks for forwarding this.
The list of parameters in NETLIB LAPACK 3.5 is correct.
However I didn?t found the described difference in MKL releases, I?ve checked a 
number of them starting 2009. I see exactly the same interface in MKL as in 
NETLIB LAPACK, as it is documented and implemented.

Nelson, could I ask you to expand a bit more details where you see this 
difference(headers, documentation, etc.). Also could you please take a look 
which actual MKL version it is (we had a number of releases in 2010), it could 
be found in doc/Release_Notes.pdf.

I?ll be on vacation May 7 ? May 17, I?m ccing guys who could support you in my 

Thank you,

From: Langou, Julien [mailto:Julien.Langou@Domain.Removed]
Sent: Tuesday, May 06, 2014 12:08 PM
To: nelsonjt@Domain.Removed<mailto:nelsonjt@Domain.Removed>
Cc: Kobotov, Alexander V
Subject: Re: [Lapack] LAPACK and INTEL MKL

Hi Nelson

In cc is Alexander from Intel. (Hi Alex!) I am forwarding him your email 
because he might be able to help us. I hope this is OK with you.


From: NelsonJT <nelsonjt@Domain.Removed<mailto:nelsonjt@Domain.Removed>>
Subject: [Lapack] LAPACK and INTEL MKL
Date: May 4, 2014 at 7:24:01 PM MDT
To: "lapack@Domain.Removed<mailto:lapack@Domain.Removed>" 
Reply-To: "nelsonjt@Domain.Removed<mailto:nelsonjt@Domain.Removed>" 

The argument lists for CGBSVXX in LAPACK_3.5.0 and INTEL Math Kernel Library 
are of slightly different data types.
The working vector RWORK used in LAPACK_3.5.0 is a real array.  The 
corresponding working vector in 2010 INTEL MKL is an integer array.
EQUED is a character variable in LAPACK_3.5.0.  The corresponding EQUED is a 
logical variable in 2010 INTEL MKL.

Closed Joint Stock Company Intel A/O
Registered legal address: Krylatsky Hills Business Park,
17 Krylatskaya Str., Bldg 4, Moscow 121614,
Russian Federation

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.

-------------- next part --------------
An HTML attachment was scrubbed...

<Prev in Thread] Current Thread [Next in Thread>
  • [Lapack] Fwd: LAPACK and INTEL MKL, Langou, Julien <=

For additional information you may use the LAPACK/ScaLAPACK Forum.
Or one of the mailing lists, or