Where is "nvscibuf.h" ?

Hello!!
I just installed cuda_10.2.89_440.33.01_linux.run on CentOS 8.0-1905.
After that in NVIDIA_CUDA-10.2_Samples, I got the following ERROR message.
Where is the include file, “nvscibuf.h”?

make[1]: Entering directory ‘/root/NVIDIA_CUDA-10.2_Samples/0_Simple/cudaNvSci’

GCC Version is greater or equal to 4.7.0 <<<
/usr/local/cuda/bin/nvcc -ccbin g++ -I…/…/common/inc -m64
–std=c++11 -gencode arch=compute_30,code=sm_30 -gencode
arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37
-gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52
-gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61
-gencode arch=compute_70,code=sm_70 -gencode arch=compute_75,code=sm_75
-gencode arch=compute_75,code=compute_75 -o cudaNvSci.o -c cudaNvSci.cpp
In file included from cudaNvSci.cpp:12:
cudaNvSci.h:14:10: fatal error: nvscibuf.h: No such file or directory
#include <nvscibuf.h>
^~~~~~~~~~~~
compilation terminated.
make[1]: *** [Makefile:394: cudaNvSci.o] Error 1
make[1]: Leaving directory ‘/root/NVIDIA_CUDA-10.2_Samples/0_Simple/cudaNvSci’
make: *** [Makefile:51: 0_Simple/cudaNvSci/Makefile.ph_build] Error 2


The issue is being researched. I don’t have further information at this time.

As a workaround, issue make -k

That should allow you to build other sample codes.

Thank you, Robert_Crovella!!

$man make
-k, --keep-going
Continue as much as possible after an error.

I was removing the directory in order to avoid the error.
I will “keep on going”…

PS:
I installed from cuda_10.2.89_440.33.01_linux.run.

I also just installed cuda_10.2.89_440.33.01_linux using deb on my Dell XPS 15 9560 with ubuntu 18.04 (dualbooted with windows 10 if that matters) and faced the exact same issue.

Tried “make -k” when in the NVIDIA_CUDA-10.2_Samples folder, but still got this error:

“make[1]: Entering directory ‘/home/hanna/cuda_samples/NVIDIA_CUDA-10.2_Samples/7_CUDALibraries/simpleCUBLAS’
/usr/local/cuda-10.2/bin/nvcc -ccbin g++ -I…/…/common/inc -m64 -gencode arch=compute_30,code=compute_30 -o simpleCUBLAS.o -c simpleCUBLAS.cpp
/usr/local/cuda-10.2/bin/nvcc -ccbin g++ -m64 -gencode arch=compute_30,code=compute_30 -o simpleCUBLAS simpleCUBLAS.o -lcublas
mkdir -p …/…/bin/x86_64/linux/release
cp simpleCUBLAS …/…/bin/x86_64/linux/release
make[1]: Leaving directory ‘/home/hanna/cuda_samples/NVIDIA_CUDA-10.2_Samples/7_CUDALibraries/simpleCUBLAS’
make: Target ‘all’ not remade because of errors.”

Any solutions? Looking forward to updates on the issue.

1 Like

Hello hanna, I think… “-k” means skipping some errors.
That is way you got the message,
“Target ‘all’ not remade because of errors.”
Please check each directories and there are executable files or compiled binary files.

@keizofk1f3l: I know, mostly just wanted to share that I also got the same error. And yes, all sample code I have tried so far has been executable, except for those using OpenGL. Not sure if it’s caused by this error or something else thought. Could you execute OpenGL samples?

Have you tried copying from
https://docs.nvidia.com/drive/active/5.1.6.0L/nvvib_docs/DRIVE_OS_Linux_SDK_Development_Guide/baggage/nvscibuf_8h_source.html

Cheers Paul J

Thank you, paul.johnston!!

As your advice I tried…
It seems that these 4 files are missing.

nvscibuf.h
nvsciipc.h
nvscisync.h
nvscierror.h

I had a quick look and got nvscibuf.h from the web site

Then running make

[paulj@pjux3 cudaNvSci]$ make

GCC Version is greater or equal to 4.7.0 <<<
/usr/local/cuda/bin/nvcc -ccbin g++ -I…/…/common/inc -m64 --std=c++11 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_75,code=sm_75 -gencode arch=compute_75,code=compute_75 -o cudaNvSci.o -c cudaNvSci.cpp
In file included from cudaNvSci.h:14,
from cudaNvSci.cpp:12:
…/…/common/inc/nvscibuf.h:18:11: fatal error: nvscierror.h: No such file or directory
#include “nvscierror.h” nvscierror.h:
^~~~~~~~~~~~~~
compilation terminated.
make: *** [Makefile:394: cudaNvSci.o] Error 1

So it found the nvscibuf.h and then realised nvscierror.h couldn’t be found

Haven’t had time to put nvscierror.h in …/…/common/inc but will let you know what happens when I do

Cheers Paul J

OMG!!

It seems some library files are missing too.
I think all NvSci’s development files are missing.

make[1]: Entering directory ‘/usr/local/cuda-10.2/samples/0_Simple/cudaNvSci’

GCC Version is greater or equal to 4.7.0 <<<
/usr/local/cuda/bin/nvcc -ccbin g++ -m64 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_37,code=sm_37 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_75,code=sm_75 -gencode arch=compute_75,code=compute_75 -o cudaNvSci cudaNvSci.o imageKernels.o main.o -L/usr/local/cuda/lib64/stubs -lcuda -lnvscibuf -lnvscisync
/usr/bin/ld: cannot find -lnvscibuf
/usr/bin/ld: cannot find -lnvscisync
collect2: error: ld returned 1 exit status
make[1]: *** [Makefile:403: cudaNvSci] Error 1
make[1]: Leaving directory ‘/usr/local/cuda-10.2/samples/0_Simple/cudaNvSci’
make: *** [Makefile:51: 0_Simple/cudaNvSci/Makefile.ph_build] Error 2

Seems to be the case
[paulj@pjux3 stubs]$ pwd
/usr/local/cuda/lib64/stubs
[paulj@pjux3 stubs]$ ls -l libcuda.so
-rwxr-xr-x. 1 root root 47216 Nov 25 12:40 libcuda.so
[paulj@pjux3 stubs]$ ls -l libvsci*
ls: cannot access ‘libvsci*’: No such file or directory
[paulj@pjux3 stubs]$

Paul J

Hi everyone

Just got the same error while compiling samples.

Used runfile installation with runfile “cuda_10.2.89_440.33.01_linux.run”.

Ubuntu 18.4 lts
gtx 1080

Hope there will be patch soon.

Greetings

Sieuwe

Confirmed on new cuda-10-2 installation on Ubuntu. The library is probably missing

From where i can get cublas64_100.dll library file.

From where i can get cublas64_100.dll files.

This thread seems to be about the Linux install, when you mention a missing dll can I assume you are refering to Windows?
Cheers Paul J

Has anyone gotten the samples compiled?

yes i am running on windows 10 and
graphics geforce 1050ti

This was sent to me, missed it was not also on this thread.
Basically when we got 10.2 it was not included.

As indicated in the CUDA 10.2 release notes:

https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html#cuda-general-new-features

the NvSci interop support is a new BETA feature to CUDA 10.2 and hence it’s support is limited at this time. The cudaNvSci sample that is contained in the CUDA 10.2 toolkit requires a new NvStreams development package that has, to date, not yet been released. NvStreams is a product of NVIDIA Automotive SW and is already supported in DRIVE OS aarch64 Linux on DRIVE OS 5.1.12.0 . The x86_64 support package for Nvstreams is new and is due to be released soon as a part of a future DRIVE OS release. If you are interested to obtain the upcoming release of Nvstreams support for DRIVE OS on x86_64 please register for an account at https://developer.nvidia.com/DRIVE-AGX-program

The initial release of the BETA Nvstreams x86_64 library package will be handled via https://partners.nvidia.com/ which is a credentialed access portal for automotive partners for obtaining all related DRIVE OS software.

Users who don’t have a particular concern with the above should disregard this sample code. The failure of it to build correctly at this time is expected. It does not indicate any sort of problem with the install process, GPU, or compatibility.

To work around this, please issue make -k or similar when building the sample codes on linux. This does not eliminate the failure to build this sample code, but allows the other samples which are buildable, to be properly built. This failure to build will still be indicated in a general make/build process and should be disregarded.

I won’t be able to answer further questions or future-looking questions about this at this time. This includes but is not limited to:

  • “what about windows?”
  • “what does this sample code do?”
  • “when will XXX be available?”
  • “are there any workarounds?”
  • “where is YYY located?”
  • “how can I get further information?”

General questions about DRIVE HW or SW products should be posted in the Drive section of these forums.

https://devtalk.nvidia.com/default/board/376/general/

2 Likes

/usr/local/cuda/samples/Makefile

edit line 41:

FILTER_OUT := 0_Simple/cudaNvSci/Makefile

1 Like