Hello list, today I tried to load a large dataset (just below 4G) on a Tesla, that I know I was able to load before reinstalling my machine, but now fails with an out of memory error.
Turns out that there are about 325 MB missing already just after device instanciation: >>> import pycuda.driver as cuda >>> import pycuda.autoinit >>> pycuda.autoinit.device.name() 'Tesla C1060' >>> cuda.mem_get_info() (3953234688, 4294705152) >>> pycuda.VERSION (0, 94) I'm running Ubuntu Hardy, Nvidia Driver version 185.18.36. How do I go about debugging this? Thank you! Sincerely yours, Florian Finkernagel _______________________________________________ PyCUDA mailing list [email protected] http://tiker.net/mailman/listinfo/pycuda_tiker.net
