Check it there: http://developer.nvidia.com/object/cuda_3_1_downloads.html
Did not find new features since 3.1beta, just bug fixes. 2D surface write intrinsics that were buggy in 3.1 Beta are now working for instance.
There is also a reference to GPUDirect in the "release highlights", but I did not find any documentation for it. GPUDirect seams to allow third party devices to do direct memory acces to the GPU memory (cf the anouncement here).
CUDA 3.1 Final out
Subscribe to:
Post Comments (Atom)
June 30, 2010 at 1:40 PM
Bonjour,
Je viens de mettre aussi à jour ma version de CUDA mais malheureusement j'ai une erreur récurrente que ça sur la compilation de mes codes ou de la SDK de Nvidia.
[code]
/usr/include/surface_functions.h: In function ‘void surf1Dread(T*, surface, int, int, cudaSurfaceBoundaryMode)’:
/usr/include/surface_functions.h:100:95: erreur: there are no arguments to ‘__surf1Dreadc1’ that depend on a template parameter, so a declaration of ‘__surf1Dreadc1’ must be available
/usr/include/surface_functions.h:100:95: note: (if you use ‘-fpermissive’, G++ will accept your code, but allowing the use of an undeclared name is deprecated)
/usr/include/surface_functions.h:101:96: erreur: there are no arguments to ‘__surf1Dreads1’ that depend on a template parameter, so a declaration of ‘__surf1Dreads1’ must be available
/usr/include/surface_functions.h:102:94: erreur: there are no arguments to ‘__surf1Dreadu1’ that depend on a template parameter, so a declaration of ‘__surf1Dreadu1’ must be available
/usr/include/surface_functions.h:103:94: erreur: there are no arguments to ‘__surf1Dreadu2’ that depend on a template parameter, so a declaration of ‘__surf1Dreadu2’ must be available
/usr/include/surface_functions.h:104:95: erreur: there are no arguments to ‘__surf1Dreadu4’ that depend on a template parameter, so a declaration of ‘__surf1Dreadu4’ must be available
[/code]
Et d'autres erreurs dans le genre. Je ne sais pas si vous avez rencontré ce type d'erreur et comment je pourrais la résoudre.
Merci d'avance ;)
July 2, 2010 at 11:26 AM
Finalement, après plusieurs tests, il s'avère qu'il faut utiliser le flag -fpermissive, c'est moche mais ça fonctionne quand on est sur Archlinux avec un gcc 4.5.