Comments on: Oh my god—it’s full of cores!

By: Javier Ruiz

Javier Ruiz — Wed, 14 Jul 2010 15:07:30 +0000

Hi, if you are lucky to get one of the new Fermi architecture cards (like the GTX470), you will enjoy all the bells and whistles of Compute Capability 2.0, which includes use of printf() inside a device function.
Also concurrent kernels and other funny things.
Saludos,
Javier Ruiz

By: extern _shader_ hyperfloat computeAuxVectors ( )

extern _shader_ hyperfloat computeAuxVectors ( ) — Sat, 27 Feb 2010 18:01:09 +0000

good…. i like that..

By: No, include me!

No, include me! — Tue, 16 Feb 2010 12:41:51 +0000

cutil_inline.h

By: No, include me!

No, include me! — Tue, 16 Feb 2010 12:41:33 +0000

#include

By: #include

#include — Tue, 16 Feb 2010 12:40:35 +0000

Include me ^

By: Morten Madsen

Morten Madsen — Wed, 10 Feb 2010 10:27:00 +0000

#1: I guess my answer comes a bit late but remember to include additional libraries. (cutil32.lib)

By: t3d

t3d — Fri, 09 Oct 2009 21:36:39 +0000

Regarding footnote #6:
Did you hear of cuda-gdb? I think using it could give even better results than printf()

Unfortunately i cannot use it, because i invoice my cuda code from FORTRAN, with a C wrapper added.

I came here looking for a way to use “cutilSafeCall()” Still no idea what to do to make nvcc stop yelling “error: identifier “cutilSafeCall” is undefined” at me…