cannot find the CUDA examples after JetPack installation

I just finished JetPack 4LT installation. My Host PC is on Ubuntu 16.04 (I couldn’t make ubuntu 14.04 recognize the network card on my hardware).
Anyway, I made all the way through, I flashed my Jetson TX2, successfully, no errors displayed.
The newly flashed Jetson TX2 rebooted ok.
Now I am trying to run the compiled examples, but I don’t find them either in my host PC or in my target. Where they are? I don’t have any directory in called “NVIDIA_CUDA-%” or anything similar.
Please advise.
Regards,
Mario

I have these in /home/ubuntu/NVIDIA_CUDA-8.0_Samples.

Thank you. Where are they? In the Jetson target or in the host PC?
In my case, there are not there, in neither the target or the host.
Mario

Hi,

By default, CUDA sample should locate at ‘/home/ubuntu/’. If not, please re-flash it again.
(Please make sure you flash TX2 with CUDA sample selected)

Another alternative is to compile CUDA sample by yourself.

$ /usr/local/cuda/bin/cuda-install-samples-8.0.sh .
$ cd NVIDIA_CUDA-8.0_Samples/
$ make

Thanks.

The target does not even have the /usr/local/cuda directory!
I have selected “full” radio button in the Component Manager.
I repeated the procedure, and now I am getting another error:
just after press recovery button on the target, in the host I am getting the following error:

Making system.img…
/dev/loop0 is not block device. Terminating…

Please help! what should I do?
Regards,
Mario

Hi,

Please attach JetPack log for us debugging.
Located at ‘[JetPack]/_installer/logs’

Just want to check:

  1. You can successfully flash tx2 at the first time, but the package is incomplete.
  2. Fail to flash image at the second time
    Is it correct?

We are sorry for the inconvenience and please help us provide the log information.
Thanks.

Thank you for your kind response.
Please find attached the log files, in ZIP format.
I responded to your questions, inline, please see [MT].
Thanks,
Regards,
Mario

Hi,

Please attach JetPack log for us debugging.
Located at ‘[JetPack]/_installer/logs’

Just want to check:

  1. You can successfully flash tx2 at the first time, but the package is incomplete.

[MT] Yes. I made to flash the os into tx2 once. But it was never complete. I never made to transfer the Cuda tools or examples.

  1. Fail to flash image at the second time
    Is it correct?

[MT] That’s correct. After first attempt, I am getting this error “/dev/loop0 is not a block device”. I tried many times again, same results.

We are sorry for the inconvenience and please help us provide the log information.
Thanks.
logs.zip (18.7 KB)

Hi,

It looks like that your host disk doesn’t have enough space to make the image.

Could you try following command:

cd [JetPack]/64_TX2/Linux_for_Tegra_64_tx2/
sudo ./flash.sh --no-flash jetson-tx2 mmcblk0p1

This command will only make the image without flash OS to the device.
If you still met error with this command, please check if you have enough disk space.

Thanks, and please let us know the results.

I have plenty of space in both the target and the host.

Please find attached results of df command in both the Host PC and the target.
I run the command you suggested. Exactly same results. Script ends with line:
/dev/loop0 is no block device. Terminating…

Please advise.
Thank you,
Mario

DF_HOST.txt (519 Bytes)
DF_TARGET.txt (484 Bytes)

Hi,

Thanks for the feedback.
Is your host Ubuntu14.04?

Yes. The host is running Ubuntu 14.04, kernel 4.6

I solved the problem.

  1. I found that the ubuntu linux in the Host PC did not have /dev/loop* devices installed.
    I created new ones by doing the following procedure:
    cd /etc vi modules
loop max_loop=64
lp

Reboot the host machine.

  1. I was trying to get lan connection between the host PC and the TX2 using wireless.
    I connected both node to a wired line with internet connection. The install went through no problem.

Now the NVIDIA_CUDA_EXAMPLES are there.
Thank you for your support.
Mario