OpenCV 4.5

Autonomous Machines Jetson & Embedded Systems Jetson Xavier NX
#1

I fired up my new Xavier NX (Jetpack 4.1.1) and discovered OpenCV4.1.1 was not CUDA enabled. I downloaded and built OpenCV 4.5 using the script here:

All went well, it installed, and CUDA is now enabled. However, it did not detect BLAS. I have scoured every forum and installed everything related to BLAS to no effect - the build does not detect it.

There is a suggestion that I build OpenBlas. However I see that Nvidia has an accelerated cuBLAS. So my question is: what is the best way to add BLAS to my OpenCV build? And are there any other accelerated libraries to use in place of standard libraries?

Thanks,
Colin

#2

You would try adding -D WITH_CUBLAS=1 to your cmake options.

#3

You would try adding -D WITH_CUBLAS=1 to your cmake options.

Thank you, I will try that. And could you tell me if cuBLAS is present by default in Jetpack or do I need to install it from somewhere?

Thanks again :-)

#4

ref: nvblas for Jetson TX2 or another way to run Armadillo code with cublas

#5

It should be there. You may just check with:

sudo apt search cublas
Sorting... Done
Full Text Search... Done
libcublas-dev/unknown,stable,now 10.2.2.89-1 arm64 [installed,automatic]
  CUBLAS native dev links, headers

libcublas10/unknown,stable,now 10.2.2.89-1 arm64 [installed,automatic]
  CUBLAS native runtime libraries
...
#6

It was there, as you describe, same version. Many thanks.

Colin

#7

@colinalow
another handy tool to find a file with certain name on the local filesystem is `mlocate’

sudo apt install -y mlocate
 sudo updatedb
locate cublas | grep .so
/usr/lib/aarch64-linux-gnu/libcublas.so
/usr/lib/aarch64-linux-gnu/libcublas.so.10
/usr/lib/aarch64-linux-gnu/libcublas.so.10.2.2.89
/usr/lib/aarch64-linux-gnu/libcublasLt.so
/usr/lib/aarch64-linux-gnu/libcublasLt.so.10
/usr/lib/aarch64-linux-gnu/libcublasLt.so.10.2.2.89
/usr/lib/aarch64-linux-gnu/stubs/libcublas.so
/usr/lib/aarch64-linux-gnu/stubs/libcublasLt.so
/usr/local/cuda-10.2/doc/man/man7/libcublas.so.7
1 Like
#8

Thanks for your help :-)