Data error at pos 127 after SYEV

General CULA Dense (LAPACK & BLAS) support and troubleshooting. Use this forum if you are having a general problem or have encountered a bug.

Data error at pos 127 after SYEV

Postby huangchbii » Wed Mar 21, 2012 1:26 am

Hi, I experienced this. After Ssyev, I got Data error at pos 127. I am wondering if this is also because of a singular matrix?

Thanks.
huangchbii
 
Posts: 15
Joined: Wed Jul 07, 2010 1:27 am

Re: Data error at pos 127 after SYEV

Postby john » Wed Mar 21, 2012 5:32 am

It's likely that you've hit the limits of single precision; I would recommend trying again in double precision to see if the error persists.
john
Administrator
 
Posts: 587
Joined: Thu Jul 23, 2009 2:31 pm

Re: Data error at pos 127 after SYEV

Postby huangchbii » Thu Mar 22, 2012 4:49 am

Hi,

What do you mean "hit the limits of single precision"? I am using NVIDIA 9400m, can I still use double precision? Dsyev?

In fact, the response of this function some times is strange. Some times I got error at pos 127, some times I got NAN without any error.

My matrix is randomly generated. Do you think if it can be prevented by generating the matrix in a proper way?
huangchbii
 
Posts: 15
Joined: Wed Jul 07, 2010 1:27 am

Re: Data error at pos 127 after SYEV

Postby john » Fri Mar 23, 2012 9:04 am

Many eigenvalue problems are numerically sensitive beyond the bounds of single precision. You should probably try dsyev to see if the problem is cured by extra precision. I believe the 9400 can run this, albeit very slowly. CULA will indicate to you if your device does not support double precision by an error code when you call a D routine.
john
Administrator
 
Posts: 587
Joined: Thu Jul 23, 2009 2:31 pm


Return to CULA Dense Support

Who is online

Users browsing this forum: Yahoo [Bot] and 1 guest

cron