Hi @Kaipingli88, it will also be useful to send us the full cmake configuration and build log for us to diagnose the issue.
Thanks,
Kshitij
Please see attached build log. I am not sure which file is full cmake configuration so I attached Makefile, cmake_install.cmake, CMakeCache.txt, CTestTestfile.cmake from cmake-build-mps3-sse-300-ethos-u55-128-arm directory.
Many Thanks!
Kaiping
(Attachment log_build_default.log is missing)
(Attachment Makefile is missing)
(Attachment cmake_install.cmake is missing)
(Attachment CMakeCache.txt is missing)
(Attachment CTestTestfile.cmake is missing)
I just tried attached log files by replying to Kshitij Sisodia Arm via Discourse noreply@discuss.mlplatform.org with attachment but it is apparently failed.
Please let me know how to send your required files.
Many Thanks!
Kaiping
Hi @Kaipingli88,
You should be able to attach the log within the message if nothing else works:
cat log_build_default.log
INFO:root:Downloading resources.
INFO:root:'resources_downloaded' directory exists.
INFO:root:/home/user/ml-embedded-evaluation-kit/resources_downloaded/env/bin/python3 -m pip freeze
INFO:root:cffi==1.14.3
cmake==3.23.3
ethos-u-vela==3.5.0
flatbuffers==1.12
Jinja2==2.11.2
llvmlite==0.38.0
lxml==4.9.1
MarkupSafe==1.1.1
numba==0.55.1
numpy==1.21.3
Pillow==7.2.0
pkg_resources==0.0.0
pycparser==2.20
However, while we wait for these logs, one thing you can try straight away is:
Navigate to dependencies/tensorflow from the root of the repo and run these commands:
$ make -f tensorflow/lite/micro/tools/make/Makefile clean_downloads
tensorflow/lite/micro/tools/make/downloads/flatbuffers already exists, skipping the download.
tensorflow/lite/micro/tools/make/downloads/kissfft already exists, skipping the download.
tensorflow/lite/micro/tools/make/downloads/pigweed already exists, skipping the download.
find: ‘tensorflow/lite/micro/tools/gen_micro_mutable_op_resolver_test/person_detect’: No such file or directory
rm -rf tensorflow/lite/micro/tools/make/downloads
$ make -f tensorflow/lite/micro/tools/make/Makefile third_party_download
...
If the above command succeeds, then, re-run the default build script again.
Hoping the above helps clear out any stale downloads from TensorFlow Lite Micro. But if this doesn’t fix the issue, we will try and diagnose further with the logs.
Thanks,
Kshitij
I cannot past the build log here and send out because limitation of message size & 2 links for new user.
I just tried your suggestion. The first command executed without any issue. The 2nd one seems has issue and below is the last part of output executing your 2nd command in dependencies/tensorflow directory.
…
Switched to a new branch ‘tflm’ Applying /home/likaiping/ml-embedded-evaluation-kit/dependencies/tensorflow/tensorflow/lite/micro/tools/make/downloads/kissfft/.//…/…/kissfft.patch to /home/likaiping/ml-embedded-evaluation-kit/dependencies/tensorflow/tensorflow/lite/micro/tools/make/downloads/kissfft/./ Cloning into ‘tensorflow/lite/micro/tools/make/downloads/pigweed’… remote: Sending approximately 36.88 MiB … remote: Total 55790 (delta 27167), reused 55790 (delta 27167) Receiving objects: 100% (55790/55790), 36.88 MiB | 1.92 MiB/s, done. Resolving deltas: 100% (27167/27167), done. Updating files: 100% (3501/3501), done. Note: switching to ‘47268dff45019863e20438ca3746c6c62df6ef09’. You are in ‘detached HEAD’ state. You can look around, make experimental changes and commit them, and you can discard any commits you make in this state without impacting any branches by switching back to a branch. If you want to create a new branch to retain commits you create, you may do so (now or later) by using -c with the switch command. Example: git switch -c Or undo this operation with: git switch - Turn off this advice by setting config variable advice.detachedHead to false HEAD is now at 47268dff pw_hdlc_lite: Client I/O improvements Switched to a new branch ‘tflm’ Applying /home/likaiping/ml-embedded-evaluation-kit/dependencies/tensorflow/tensorflow/lite/micro/tools/make/downloads/pigweed/.//…/…/pigweed.patch to /home/likaiping/ml-embedded-evaluation-kit/dependencies/tensorflow/tensorflow/lite/micro/tools/make/downloads/pigweed/./ make: *** No rule to make target ‘third_party_download’. Stop.
Many Thanks!
Kaiping
Hi @Kaipingli88,
Sorry, there’s an “s” missing at the end (third_party_downloads) the command should be:
We will publish a change soon that will do this at configuration stage. Will let you know once that change is in.
Thanks,
Kshitij
I added “s” (third_party_downloads) and the 2nd command executed fine. However, re-run the default build script still has problem. Please give instruction on what is the next step or provide fix to test.
Thanks,
Kaiping
Continuing the conversation via messages. @Kaipingli88, you should have a message from me in your inbox 
.
The issue was resolved - the cause was a stray “(” character in the PATH variable - passed from Windows’ mount directory in the WSL2 environment.
Hello @kshitij-sisodia-arm, I have same issue as well. I am wondering the clear solution about it. Can you share with me too?
Hi @ek-10, I wouldn’t have expected you to see this issue if you are on one of the latest commits. It was fixed here: Diff - b891dd35e4f7403547ade73cbf6ad209193d728a^! - ml/ethos-u/ml-embedded-evaluation-kit - Gitiles. If you are still seeing this issue, can you share more details on it here please?
Thanks,
KS
Hello again, I’ve checked the diff in tensorflow.cmake file there is no problem in it. I’ve added error screenshot. They can be make it clear maybe.
You’re right, looks like you’re running a virtual Ubuntu instance, so that problem should not affect you.
Can you please share:
- Configuration command - are you configuring with additional logging params perhaps?
- Complete build log if possible
These will help us reproduce it locally and get back to you.
Hi, log_build_default.log file is very long and I can’t add it here. I didn’t make any additional configurations. I’ve just follow followed the quick start guide.
So, you’re using the build_default.py to build. That still assumes you have the prerequisites Which toolchain (and version) are you using?
Yes. I’ve installed them all.
here is the output:
"
arm-none-eabi-gcc --version
arm-none-eabi-gcc (15:10.3-2021.07-4) 10.3.1 20210621 (release)
Copyright (C) 2020 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. "
@kshitij-sisodia-arm its version is 10.3.1 maybe thats the problem? should it be 10.2.1 for the SSE-300?
I would advise trying with a toolchain downloaded from https://developer.arm.com/downloads/-/arm-gnu-toolchain-downloads; preferably 13.2.1: https://developer.arm.com/-/media/Files/downloads/gnu/13.2.rel1/binrel/arm-gnu-toolchain-13.2.rel1-x86_64-arm-none-eabi.tar.xz?rev=e434b9ea4afc4ed7998329566b764309&hash=688C370BF08399033CA9DE3C1CC8CF8E31D8C441
1 Like
