Cannot run any CUDA kernels CUDA runtime doesn't recognize NVIDIA GPU

Accelerated Computing CUDA CUDA Programming and Performance
21

See how the GL version is low.

A cuda and cudart mismatch. Drivers and toolkit. I assume the cuda/nvidia kernel driver is too low for the cuda toolkit?

This seems to be a setup issue.
edit: the toolkit and driver should be paired. The nvidia driver should be installed because it contains cuda code.

22

If I run any of the driver examples I always get:

bin/linux/release/matrixMulDrv 

Using device 0: Quadro FX 570M

cuSafeCallNoSync() Driver API error = 0002 from file <matrixMulDrv.cpp>, line 72.

To clear things up:

[list=1]

[*] I am using nvidia driver 190.53

OpenGL vendor string: NVIDIA Corporation

OpenGL renderer string: Quadro FX 570M/PCI/SSE2

OpenGL version string: 3.2.0 NVIDIA 190.53

[*] The nvcc compiler version matches the cudart lib version

nvcc --version

nvcc: NVIDIA (R) Cuda compiler driver

Copyright (c) 2005-2009 NVIDIA Corporation

Built on Thu_Jul__2_10:56:25_PDT_2009

Cuda compilation tools, release 2.3, V0.2.1221

ldd bin/linux/release/bandwidthTest 

	linux-gate.so.1 =>  (0x006dd000)

	libcudart.so.2 => /usr/local/cuda/lib/libcudart.so.2 (0x007e0000)

	libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00240000)

	libm.so.6 => /lib/tls/i686/cmov/libm.so.6 (0x008f4000)

	libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x00ca0000)

	libc.so.6 => /lib/tls/i686/cmov/libc.so.6 (0x00332000)

	libpthread.so.0 => /lib/tls/i686/cmov/libpthread.so.0 (0x00e63000)

	libdl.so.2 => /lib/tls/i686/cmov/libdl.so.2 (0x0054d000)

	librt.so.1 => /lib/tls/i686/cmov/librt.so.1 (0x00b03000)

	/lib/ld-linux.so.2 (0x0050b000)

[*] I can run deviceQuery and get the correct result.

bin/linux/release/deviceQuery

CUDA Device Query (Runtime API) version (CUDART static linking)

There is 1 device supporting CUDA

Device 0: "Quadro FX 570M"

  CUDA Driver Version:						   2.30

  CUDA Runtime Version:						  2.30

  CUDA Capability Major revision number:		 1

  CUDA Capability Minor revision number:		 1

  Total amount of global memory:				 133496832 bytes

  Number of multiprocessors:					 4

  Number of cores:							   32

  Total amount of constant memory:			   65536 bytes

  Total amount of shared memory per block:	   16384 bytes

  Total number of registers available per block: 8192

  Warp size:									 32

  Maximum number of threads per block:		   512

  Maximum sizes of each dimension of a block:	512 x 512 x 64

  Maximum sizes of each dimension of a grid:	 65535 x 65535 x 1

  Maximum memory pitch:						  262144 bytes

  Texture alignment:							 256 bytes

  Clock rate:									0.95 GHz

  Concurrent copy and execution:				 Yes

  Run time limit on kernels:					 Yes

  Integrated:									No

  Support host page-locked memory mapping:	   No

  Compute mode:								  Default (multiple host threads can use this device simultaneously)

But: I cannot run any real application using cuda; neither the ones compiled against the cudart nor the ones using cuda driver API.

I honestly have no idea as to what might be the reason for this. And I haven’t given up hope, yet! ;-)

23

The card shows having 128MB of memory. (I don’t know if that low amount is supported). I have 768MB
Driver API error = 0002

I looked up error 0002, and if it is the same as CUresult, it is CUDA_ERROR_OUT_OF_MEMORY.

This is a laptop, probably a little older. Maybe upgrade.

edit: my simple app uses 170MB including display

24

I have good news to report - the problem I had was that the device driver, toolkit and sdk were all different versions. I was pointed to

[url=“http://forums.nvidia.com/index.php?showtopic=149959”]http://forums.nvidia.com/index.php?showtopic=149959[/url]

This has pointers to a consistent set of 3.0-beta driver, toolkit and sdk. Once I deleted all the 2.3 (or whatever it actually was) and loaded the 3.0 versions, it all works.

25

You’re absolutely right about the little memory this card has, but for developing purposes this should suffice. And futhermore the basic SDK examples do run under Windows (like bandwithTest), so there should be no reason as to why this example shouldn’t run under linux.

Well I am still kind of puzzled and I am currently trying to remote develop on another cuda-capable machine.

Regards and a happy Christmas! :)

26

I’m using SSH connection to server for programming this. The same problem…

Any heads up when using ssh?

27

I’m using SSH connection to server for programming this. The same problem…

Any heads up when using ssh?