On Thu, 02 Feb 2012 11:06:56 +0100, Philip Kaufmann <[email protected]> 
wrote:
> Hi Andreas,
> 
> > Odd--I did try that version to, just to see if it would provoke a crash
> > (it shouldn't), and it didn't. What version are you on?
> 
> self.kernel.XYZ(self.commandQueue, (data.size, ), (self.WORKSIZE, ), 
> *(kernelargs), global_offset=data.baseNum[i])
> 
> The above crashes with 0.92 and I guess the other one is 2011.2, but 
> this one crashes on: self.interface.debug('using PyOpenCL version ' + 
> cl.VERSION_TEXT), which 0.92 does not ^^.

Known bug, fixed in git. Also, when you say 'crash', that sounds like
'caught a signal' to me. Please distinguish Python exceptions (usually
not a problem) from segfaults (argh!), to avoid scaring me like that.

(In Windows, it might be hard to tell one from the other unless you're
running your code from the command line.)

Andreas

Attachment: pgpL0yvDmf6Tx.pgp
Description: PGP signature

_______________________________________________
PyOpenCL mailing list
[email protected]
http://lists.tiker.net/listinfo/pyopencl

Reply via email to