Forums - 1.19.2 snpe-onnx-to-dlc Error

16 posts / 0 new
Last post
1.19.2 snpe-onnx-to-dlc Error
nokia_du
Join Date: 4 Jul 18
Posts: 4
Posted: Thu, 2018-10-04 01:29

I follow all the steps to setup the environment, and make sure all python package work fine.

All the converter looks good, [snpe-caffe-to-dlc, snpe-caffe2-to-dlc, snpe-tensorflow-to-dlc]

But when I run this command, [snpe-onnx-to-dlc --help], it only show Segmentation fault (core dumped),

The SNPE version is 1.19.2, onnx version is 1.3.0.

Could you please help to check this?

Could you provide Docker Image for us to saving the setup environment time?

Thanks.

Best regards,

  • Up0
  • Down0
nokia_du
Join Date: 4 Jul 18
Posts: 4
Posted: Tue, 2018-10-09 02:32

Use 1.18.0 and issue gone, please provide docker image to saving your clinet's time !!!

  • Up0
  • Down0
y-yamashita
Join Date: 1 Nov 18
Posts: 4
Posted: Wed, 2018-11-07 16:38
I have the same problem. I would like to try v1.18.0, but I can not find the download link. I want to tell you. Thank you.

 

  • Up0
  • Down0
haroldyang977
Join Date: 15 May 18
Posts: 4
Posted: Fri, 2019-01-25 02:15

I encounter the same issue using snpe-1.22.0

  • Up0
  • Down0
jihoonk
Profile picture
Join Date: 28 Jan 13
Location: Seoul
Posts: 55
Posted: Sat, 2019-01-26 17:55

I had the same issue. My workaround solution is to downgrade onnx from 1.3.0 to 1.2.2. This might solve the issue. 

Thanks,

Jihoon

  • Up0
  • Down0
haroldyang977
Join Date: 15 May 18
Posts: 4
Posted: Sun, 2019-01-27 21:42

I downgrad my onnx to 1.2.2, the same problem still occur (segmentation fault, core dumped)

my snpe is latest 1.22.0.212

Where can I download the earlier version of SNPE?

  • Up0
  • Down0
gesqdn-forum
Join Date: 4 Nov 18
Posts: 184
Posted: Mon, 2019-02-11 23:20

Hi,

 

Irrespective of SNPE version, the issue is been seen.

 

On debugging the core dump file generated on segmentation fault, we figured out that the error is because of missing folder "Object". Below are the details of the coredump.

 

jinka@jinka:~/git/SNPE/snpe-sdk$ gdb python2.7-dbg

GNU gdb (Ubuntu 7.11.1-0ubuntu1~16.5) 7.11.1

This GDB was configured as "x86_64-linux-gnu".

Reading symbols from python2.7-dbg...done.

(gdb) run ./bin/x86_64-linux-clang/snpe-onnx-to-dlc  --model_path models/bvlc_alexnet/bvlc_alexnet/model.onnx --dlc_path bvlc_alexnet_onnx.dlc

Starting program: /usr/bin/python2.7-dbg ./bin/x86_64-linux-clang/snpe-onnx-to-dlc  --model_path models/bvlc_alexnet/bvlc_alexnet/model.onnx --dlc_path bvlc_alexnet_onnx.dlc

[Thread debugging using libthread_db enabled]

Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".


Program received signal SIGSEGV, Segmentation fault.

0x00007ffff5bdf482 in PyString_FromFormatV (format=0x7ffff5ce4f98 "expected string or Unicode object, %.200s found", vargs=0x7fffff7ff010)

    at ../Objects/stringobject.c:169

169    ../Objects/stringobject.c: No such file or directory.

  • Up0
  • Down0
themachine013
Join Date: 12 Feb 19
Posts: 1
Posted: Tue, 2019-02-12 23:03

So where can I get these folder and files?

  • Up0
  • Down0
dmitry.yunitski
Join Date: 1 Nov 18
Posts: 3
Posted: Thu, 2019-03-28 02:59

We face the same issue, snpe-onnx-to-dlc --help fails with segfault with the latest 1.24 and 1.23 versions of SNPE. Tried with ONNX 1.2.2, 1.3.1, 1.41 with no luck. Did anyone solve this?

  • Up0
  • Down0
mstyura
Join Date: 28 Mar 19
Posts: 1
Posted: Thu, 2019-03-28 06:32

SEGFAULT disappears when protobuf version is 2.5.0, but then ONNX(1.4.1) failed to be imported, presumably due to protobuf version mismatch. I've not yet figured out which versions combination works :(

  • Up0
  • Down0
aleksandr.buslaev
Join Date: 14 Sep 18
Posts: 1
Posted: Tue, 2019-04-16 00:34

same here with most recent 1.25

  • Up0
  • Down0
pgaston
Join Date: 29 Mar 19
Posts: 5
Posted: Thu, 2019-05-16 04:19

This is still occurring.    Surely there is a fix somewhere.

I'm on 1.25.1.310

  • Up0
  • Down0
alex.veuthey
Join Date: 20 May 19
Posts: 2
Posted: Mon, 2019-05-20 07:52

I've tried all versions of ONNX with protobuf 2.5.0, it doesn't work unfortunately.

I've also tried with protobuf 2.6.1, which is the latest before 3.0.0, but the segfault is still there for me (also 1.25.1.310)...

  • Up0
  • Down0
pgaston
Join Date: 29 Mar 19
Posts: 5
Posted: Mon, 2019-05-20 09:35

Perhaps a clue, but it doesn't seg fault when you run at the top level of the git folder.     Still fails, but now with a Python error.    (See my topic from May 16.)

  • Up0
  • Down0
alex.veuthey
Join Date: 20 May 19
Posts: 2
Posted: Tue, 2019-05-21 00:42

I'm assuming you are talking of the $SNPE_ROOT folder, as I got a .zip file and not a cloned git... But that is the way I launched it before, nothing works but a segfault. At some point, I was able to see Python errors though. Debugging the segfault with @gesqdn-forum 's answer didn't show anything else than segfault as well.

At this point, I'm trying to debug the python code, see which import throws the segfaul... I'll update here if I find anything.

EDIT: finally made it work, but had to go through an additional conversion step... Caffe2. The following link, at the very bottom, shows how to export an ONNX model to Caffe2. https://github.com/onnx/tutorials/blob/master/tutorials/OnnxCaffe2Import...

I had to make a few modifications to my model and install some dependencies on my SNPE environment (python 2.7), but it finally worked.

Note: the tutorial in the following link is not very clear, but it expects the name of the ONNX/Caffe2 input node as the first argument of the --input_dim parameter... https://developer.qualcomm.com/sites/default/files/docs/snpe/model_conv_... .

  • Up0
  • Down0
shanesmiskol
Join Date: 19 May 19
Posts: 10
Posted: Sat, 2019-06-01 08:27

Hi! Does anyone here still have 1.19 (or similar) downloaded? I am in need of the 1.19 snpe-tensorflow-to-dlc file used to convert TensorFlow models. If someone would be kind enough to share the link or PM me, that would be great appreciated. Email is [email protected] Thanks

  • Up0
  • Down0
or Register

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries (“Qualcomm”). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.