asherliu / thrust

Automatically exported from code.google.com/p/thrust
Apache License 2.0
0 stars 0 forks source link

Misaligned Shared or Local Address #471

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago
I have some code written which fails with a thrust::system::system_error

After trying to debug it for a few days i upgraded to CUDA toolkit 4.1
and ran cuda-memcheck

I get the following error:

terminate called after throwing an instance of 'thrust::system::system_error'
  what():  unspecified launch failure
========= Error: process didn't terminate successfully
========= Misaligned Shared or Local Address
=========     at 0x00002bc0 in 
/usr/local/cuda/include/thrust/iterator/detail/zip_iterator_base.h:277:thrust
=========     by thread (64,0,0) in block (12,0,0)
=========
========= ERROR SUMMARY: 1 error

I googled around and didnt find anything mentioning this issue. Is it possible 
this is a Thrust bug?

Also I am running on Ubuntu 11.04 with a dedicated GTX480 for cuda

Thanks,

Matthias

Original issue reported on code.google.com by Matthias...@gmail.com on 15 Feb 2012 at 3:24

GoogleCodeExporter commented 8 years ago
Thanks for the report Matthias.  If you could you attach a minimalistic program 
(e.g. < 30 lines) that exhibits the problem we'll be able to investigate the 
matter further.

In the meantime, I wouldn't be concerned about the warning unless you're 
getting the wrong results.

Original comment by wnbell on 15 Feb 2012 at 3:57

GoogleCodeExporter commented 8 years ago
This triggers the error for me.

Error only occurs when compiling with 
-encode=arch=compute_20,code="sm_20,compute_20". 

Using CUDA 4.1 and Visual Studio 2010.

Original comment by surban84 on 27 Feb 2012 at 8:29

Attachments:

GoogleCodeExporter commented 8 years ago
Interesting for me its CUDA 4.1 on Linux with the newest driver both on Ubuntu 
11.04 and RHEL 5.something 

Original comment by Matthias...@gmail.com on 27 Feb 2012 at 8:34

GoogleCodeExporter commented 8 years ago
Hi Nathan,
it get the same bug on Ubuntu 10.04, CUDA 4.1 -arch=sm_20.

Original comment by peterzas...@gmail.com on 5 Mar 2012 at 1:36

GoogleCodeExporter commented 8 years ago
think this must be a compiler issue

Original comment by jaredhoberock on 7 May 2012 at 9:58

GoogleCodeExporter commented 8 years ago
[deleted comment]
GoogleCodeExporter commented 8 years ago
[deleted comment]
GoogleCodeExporter commented 8 years ago
is there any update on this issue? I am running into the same bug. I am using 
CUDA toolkit 4.1 and fedora 14 and -arch=sm_20

Original comment by fly...@gmail.com on 25 Jan 2013 at 9:41

GoogleCodeExporter commented 8 years ago
Any news concerning this problem? I am having the same...

Original comment by pedro.fe...@gmail.com on 10 Jul 2014 at 10:11