gromero commented on pull request #7853:
URL: https://github.com/apache/tvm/pull/7853#issuecomment-820472243
@areusch oh okay. Thanks for fixing it!
So, what happened here was that I've indeed tested an early version of my
change with STM32 disco board, but then I did additional changes and checked it
again, but got the following error:
```File /home/gromero/.tvm_test_data/data/sine_model.tflite exists, skip.
Model Version: 3
Cannot find config for target=c -keys=cpu -link-params=0 -march=armv7e-m
-mcpu=cortex-m7 -model=stm32f746xx -runtime=c -system-lib=1,
workload=('dense_pack.x86', ('TENSOR', (1, 16), 'float32'), ('TENSOR', (1, 16),
'float32'), None, 'float32'). A fallback configuration is used, which may bring
great performance regression.
Cannot find config for target=c -keys=cpu -link-params=0 -march=armv7e-m
-mcpu=cortex-m7 -model=stm32f746xx -runtime=c -system-lib=1,
workload=('dense_pack.x86', ('TENSOR', (1, 16), 'float32'), ('TENSOR', (16,
16), 'float32'), None, 'float32'). A fallback configuration is used, which may
bring great performance regression.
Cannot find config for target=c -keys=cpu -link-params=0 -march=armv7e-m
-mcpu=cortex-m7 -model=stm32f746xx -runtime=c -system-lib=1,
workload=('dense_pack.x86', ('TENSOR', (1, 1), 'float32'), ('TENSOR', (16, 1),
'float32'), None, 'float32'). A fallback configuration is used, which may bring
great performance regression.
./micro_tflite_disco_head.py:202: DeprecationWarning: legacy graph executor
behavior of producing json / lib / params will be removed in the next release.
Please see documents of tvm.contrib.graph_executor.GraphModule for the new
recommended usage.
graph, c_mod, c_params = relay.build(mod, target=TARGET, params=params)
Including boilerplate (Zephyr base):
/home/gromero/zephyrproject/zephyr/cmake/app/boilerplate.cmake
In file included from <command-line>:
/home/gromero/zephyrproject/zephyr/boards/arm/stm32f746g_disco/stm32f746g_disco.dts:9:10:
fatal error: st/f7/stm32f746nghx-pinctrl.dtsi: No such file or directory
9 | #include <st/f7/stm32f746nghx-pinctrl.dtsi>
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
CMake Error at /home/gromero/zephyrproject/zephyr/cmake/dts.cmake:156
(message):
command failed with return code: 1
Call Stack (most recent call first):
/home/gromero/zephyrproject/zephyr/cmake/app/boilerplate.cmake:535
(include)
/home/gromero/zephyrproject/zephyr/share/zephyr-package/cmake/ZephyrConfig.cmake:24
(include)
/home/gromero/zephyrproject/zephyr/share/zephyr-package/cmake/ZephyrConfig.cmake:35
(include_boilerplate)
CMakeLists.txt:3 (find_package)
Traceback (most recent call last):
File "./micro_tflite_disco_head.py", line 242, in <module>
micro_binary = tvm.micro.build_static_runtime(
File "/home/gromero/git/tvm/python/tvm/micro/build.py", line 240, in
build_static_runtime
libs.append(compiler.library(lib_build_dir, lib_srcs,
compiler_options["lib_opts"]))
File "/home/gromero/git/tvm/python/tvm/micro/contrib/zephyr.py", line 209,
in library
self._subprocess_env.run(
File "/home/gromero/git/tvm/python/tvm/micro/contrib/zephyr.py", line 60,
in run
return subprocess.check_output(cmd, env=env, **kw,
universal_newlines=True)
File "/usr/lib/python3.8/subprocess.py", line 411, in check_output
return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
File "/usr/lib/python3.8/subprocess.py", line 512, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['cmake', '..',
'-DBOARD=stm32f746g_disco',
'-DEXTRA_CFLAGS=-Wno-error=incompatible-pointer-types -Wno-unused-variable
-fno-builtin', "-DEXTRA_CXXFLAGS=-std=c++11 -Wall -Werror
'-DDMLC_USE_LOGGING_LIBRARY=<tvm/runtime/logging.h>' -Wno-unused-variable",
'-DEXTRA_LDFLAGS=-std=c++11']' returned non-zero exit status 1.
```
Skimming through it I attributed it to a Zephyr 2.5.0 change which I stashed
to look later and then kind forgot to return to it, focusing on the CI issues I
had on the submitting the 7813 PR, sorry about that.
I'm wondering on which board you hit the regression first. Was that a disco
board or the Nordic (nRF)?
Anyway, I need to think of some way to avoid that kind of regression since
it's happening too often recently...
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
