search

mmizzle9 / cudpp

Automatically exported from code.google.com/p/cudpp
Other
0 stars 0 forks source link

cudppscan gives wrong result under large input data size #96

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago 
What steps will reproduce the problem?
1. compile cudpp1.1 and the sample codes in apps dir;
2. run "cudpp_testrig -scan -n=200000", you may have to enlarge the input data 
size to reproduce the issue, 200000 is enough in my case.

What is the expected output? What do you see instead?
Expected: Test Passed.
Actually: Test Failed.

What version of the product are you using? On what operating system?
CUDA 2.3, cudpp1.1, CentOS 5.3

Please provide any additional information below.

Original issue reported on code.google.com by hchwang...@gmail.com on 28 Jul 2011 at 5:24

GoogleCodeExporter commented 8 years ago 
Hi, both CUDPP 1.1 and CUDA 2.3 are out-of-date, and we don't have resources to 
support them.  CUDPP 1.1.1 is the current release, but we are in the middle of 
getting CUDPP 2.0 released, so please try the trunk from SVN 
(http://code.google.com/p/cudpp/source/checkout) and report back if you still 
have a failure.

I also suggest upgrading to CUDA 4.0, or *at least* CUDA 3.2.

Please tell us which GPU you have, also (please just paste the output of 
running cudpp_testrig, since it prints the GPU information).

Original comment by harr...@gmail.com on 28 Jul 2011 at 5:29

GoogleCodeExporter commented 8 years ago 
BTW, n=200000 is a SMALL input data size, not a large one!

Original comment by harr...@gmail.com on 28 Jul 2011 at 5:30

GoogleCodeExporter commented 8 years ago 
No response from submitter, no repro.  Closing.

Original comment by harr...@gmail.com on 9 Aug 2011 at 1:22