Compiling SDK on opensuse

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello,

I am completely new to CUDA as I am still in the process of installation. I have searched this forum for the issue I get, but found only some old posts which doesn’t seem to be applicable to my case.

Generally my problem is that I can’t compile SDK projects:

make

make[1]: Entering directory `/home/c111269/NVIDIA_CUDA_SDK/sdk/common'

src/cutil.cpp:47:27: error: builtin_types.h: No such file or directory

In file included from ./../common/inc/stopwatch.h:36,

				 from src/cutil.cpp:52:

./../common/inc/stopwatch_base.h:74: warning: type qualifiers ignored on function return type

./../common/inc/stopwatch_base.h:78: warning: type qualifiers ignored on function return type

In file included from ./../common/inc/stopwatch_base.h:90,

				 from ./../common/inc/stopwatch.h:36,

				 from src/cutil.cpp:52:

./../common/inc/stopwatch_base.inl:85: warning: type qualifiers ignored on function return type

./../common/inc/stopwatch_base.inl:96: warning: type qualifiers ignored on function return type

In file included from ./../common/inc/stopwatch.h:43,

				 from src/cutil.cpp:52:

./../common/inc/stopwatch_linux.h:64: warning: type qualifiers ignored on function return type

./../common/inc/stopwatch_linux.h:68: warning: type qualifiers ignored on function return type

./../common/inc/stopwatch_linux.h:144: warning: type qualifiers ignored on function return type

./../common/inc/stopwatch_linux.h:161: warning: type qualifiers ignored on function return type

In file included from src/cutil.cpp:52:

./../common/inc/stopwatch.h:53: warning: type qualifiers ignored on function return type

In file included from src/cutil.cpp:53:

./../common/inc/bank_checker.h:52: warning: type qualifiers ignored on function return type

./../common/inc/bank_checker.h:238: warning: type qualifiers ignored on function return type

make[1]: *** [obj/release/cutil.cpp.o] Error 1

make[1]: Leaving directory `/home/c111269/NVIDIA_CUDA_SDK/sdk/common'

make: *** [lib/libcutil.so] Error 2

nvcc -V

nvcc: NVIDIA (R) Cuda compiler driver

Copyright (c) 2005-2009 NVIDIA Corporation

Built on Thu_Apr__9_05:05:52_PDT_2009

Cuda compilation tools, release 2.2, V0.2.1221

gcc --version

gcc (SUSE Linux) 4.3.2 [gcc-4_3-branch revision 141291]

Copyright (C) 2008 Free Software Foundation, Inc.

This is free software; see the source for copying conditions.  There is NO

warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

I am running 64 bit openSUSE 11.1 with

lspci | grep -i nVidia

01:00.0 VGA compatible controller: nVidia Corporation GeForce 9500 GT (rev a1)

If you could give me some feedback on that I would be much obliged.

2

Did you include “/usr/local/cuda/include” in your search path?
Or maybe you installed the cuda toolkit in a directory other than /usr/local/cuda?

N.

3

Thank you for your comment. Indeed I have not installed CUDA in the default directory. Due to some computer settings I decided to install in my home directory and eventually the path is \home\c111269\cfd\cuda\cuda

I have added \cuda\bin to my PATH and \cuda\lib to LD_LIBRARY_PATH. The installer did not mention about INCLUDE directory. I have tried it as well but to no avail.

4

There’s a file called “common.mk” in the NVIDIA_SUDA_SDK/common directory.

Adjust this line:

CUDA_INSTALL_PATH ?= /usr/local/cuda

to point to the directory where you installed cuda and you should be good to go.

N.

5

Many thanks. Compilation worked like a charm! I have compiled bandwithTest and several other projects. However the former one has given me a sad ouput:

./bandwidthTest 

Running on......

	  device 0:Device Emulation (CPU)

Quick Mode

Host to Device Bandwidth for Pageable memory

cudaSafeCall() Runtime API error in file <bandwidthTest.cu>, line 667 : no CUDA-capable device is available.

I have found my card

/sbin/lspci | grep -i nVidia

01:00.0 VGA compatible controller: nVidia Corporation GeForce 9500 GT (rev a1)

on the supported products list. Have I installed the driver incorrectly? (I did through opensuse install software).

Moreover the device test gives

./deviceQuery 

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

There is no device supporting CUDA.

Device 0: "Device Emulation (CPU)"

  CUDA Capability Major revision number:		 9999

  CUDA Capability Minor revision number:		 9999

  Total amount of global memory:				 4294967295 bytes

  Number of multiprocessors:					 16

  Number of cores:							   128

  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:									 1

  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:									1.35 GHz

  Concurrent copy and execution:				 No

  Run time limit on kernels:					 No

  Integrated:									Yes

  Support host page-locked memory mapping:	   Yes

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

Test PASSED

Press ENTER to exit...
6

Which driver and SDK version are you using?

Also make sure that libcuda.so is in your library path.

N.

7

Thanks again for your help. As for the versions:

NVIDIA Driver for Linux (SUSE 11.1) 185.18.14

Cuda compilation tools, release 2.2, V0.2.1221

SDK is also 2.2

Did you mean libcudart, by licuda? This is the only thing I can find and yes I have it in my LD_LIBRARY_PATH:

find ./cuda/ -name libcuda*

./cuda/cuda/lib/libcudart.so

./cuda/cuda/lib/libcudart.so.2

./cuda/cuda/lib/libcudart.so.2.2
8

No, libcuda.so should be located in your /usr/lib directory
or /usr/lib32 or /usr/lib64

N.

9

You’re absolutely right:

find /usr/ -name libcud*

/usr/lib64/libcuda.so.185.18.14

/usr/lib64/libcuda.so.1

Alas, adding it to PATH or LD_LIBRARY_PATH did not resolve the problem.

Have you noticed that the Device 0 is actually CPU not the graphic card? It look odd. This why I am concerned about my driver installation went smoothly.

10

I’ve encountered this problem before. Try creating a symbolic link /usr/lib64/libcuda.so which points to /usr/lib64/libcuda.so.1

N.

11

Very many thanks - problem resolved! Now both tests display GeForce9500 as device 0 and other programs are working as well. Your help was greatly appreciated. :-)

12

Hi, Nico and all:

I am having the same problem as AlmostSurelyRob has, except I am seeing it on a MacBook Pro. I had a fresh installation of cudatookit 2.2 and cudasdk 2-2.21 on a MacBook Pro with Mac OS X 10.5.8.

I have /usr/local/cuda/bin added into path, and /usr/local/cuda/lib added into DYLD_LIBRARY_PATH .

I post below the output of deviceQuery and bandwidthTest :

=============================
[dh11:bin/darwin/release] % ./deviceQuery
CUDA Device Query (Runtime API) version (CUDART static linking)
There is no device supporting CUDA.

Device 0: “Device Emulation (CPU)”
CUDA Capability Major revision number: 9999
CUDA Capability Minor revision number: 9999
Total amount of global memory: 4294967295 bytes
Number of multiprocessors: 16
Number of cores: 128
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: 1
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: 1.35 GHz
Concurrent copy and execution: No
Run time limit on kernels: No
Integrated: Yes
Support host page-locked memory mapping: Yes
Compute mode: Default (multiple host threads can use this device simultaneously)

Test PASSED

Press ENTER to exit…

=============================

[dh11:bin/darwin/release] % ./bandwidthTest
Running on…
device 0:Device Emulation (CPU)
Quick Mode
Host to Device Bandwidth for Pageable memory
cudaSafeCall() Runtime API error in file <bandwidthTest.cu>, line 667 : no CUDA-capable device is available.

========================

Could anybody give an advice please?

Thanks

13

Change the LD_LIBRARY_PATH /usr/local/cuda/lib

to

/usr/local/cuda/lib64

and try again…for example by doing a:

export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

…where /usr/local/cuda is the cuda install directory…