culaDeviceDgeTranspose on non-square matrices

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

culaDeviceDgeTranspose on non-square matrices

Postby mjadidi » Fri Nov 09, 2012 7:14 am

Hello ,

I am using a cula (sickits.cuda) python binding to do matrix transformation but I am have problem with non-square matrices.There is nothing in the documentation that matrix should be square.


for example for matrix :
a=

[ 0.19949964, 0.19801633, 0.88893133],
[ 0.98248964, 0.8524396 , 0.42118522],
[ 0.02960244, 0.24876817, 0.8868297 ]]


with the code=
Code: Select all
cula.culaDeviceDgeTranspose(3,3,a_gpu.ptr,a_gpu.__len__(),d_gpu.ptr,d_gpu.__len__())


I am getting correct result ,but for

w=

[[ 3. , 0.19949964, 0.19801633, 0.88893133],
[ 2. , 0.98248964, 0.8524396 , 0.42118522],
[ 5. , 0.02960244, 0.24876817, 0.8868297 ]]


I am getting :

[[ 3. , 0.19949964, 0.19801633],
[ 0.88893133, 2. , 0.98248964],
[ 0.8524396 , 0.42118522, 5. ],
[ 0.02960244, 0.24876817, 0.8868297 ]]



any idea why it's happening ?
mjadidi
 
Posts: 15
Joined: Wed Jun 20, 2012 2:58 am

Re: culaDeviceDgeTranspose on non-square matrices

Postby kyle » Sat Nov 10, 2012 10:43 am

Make sure that:
* M and N are set correctly based on the input matrix
* Make sure you are not doing an in-place operation
* Make sure that your leading dimensions are correct (they are column major)
kyle
Administrator
 
Posts: 301
Joined: Fri Jun 12, 2009 7:47 pm


Return to CULA Dense Support

Who is online

Users browsing this forum: No registered users and 2 guests

cron