Commit Graph

5 Commits

Author SHA1 Message Date
Stephanos Ioannidis
6ee8182e0a samples: tflite-micro: hello_world: Increase main stack size
This commit increases the main stack size to 2048 since the current
stack size is insufficient for the `qemu_x86` board (the default stack
size is 1024 for that board) and results in a stack overflow in the
main thread.

Note that this is required because the newlib full variant, which is
used by this sample, included in the Zephyr SDK 0.13.2 now has an
increased stack usage requirement due to enabling the C99 format
specifier support.

For more details, refer to the issue #40469 [1].

[1] https://github.com/zephyrproject-rtos/zephyr/pull/40469#issuecomment-977854095

Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
2021-11-25 06:58:45 -05:00
Stephanos Ioannidis
9d7c9ffefd samples: tflite-micro: hello_world: Enable standard C++ library
The TensorFlow Lite module makes use of the features provided by the
standard C++ library (e.g. `#include <functional>`), so the standard
C++ library config must be enabled for it.

This used to work without `CONFIG_LIB_CPLUSPLUS=y` due to the bug
described in the issue #36644.

Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
2021-09-21 11:39:18 -04:00
Lauren Murphy
bdae9399be samples: tflite-micro: fix magic_wand
Fixes accelerometer_handler.cc to import correct header
with TfLiteStatus since c_api_internal.h file was deleted in
the new standalone tflite-micro repo.

Removes import of version.h in main_functions.cc to match
tflite-micro version.

Signed-off-by: Lauren Murphy <lauren.murphy@intel.com>
2021-08-24 07:26:08 -04:00
Lauren Murphy
82f64e1f20 samples: tflite-micro: update hello_world
Updates tflite-micro hello_world sample code.

Signed-off-by: Lauren Murphy <lauren.murphy@intel.com>
2021-08-24 07:26:08 -04:00
Lauren Murphy
6fd1b02496 docs, samples: tflite-micro: change tensorflow to tensorflow lite micro
Changes name of folder from tensorflow to tflite-micro and updates docs
to reference TensorFlow Lite Micro specifically instead of TensorFlow.

Signed-off-by: Lauren Murphy <lauren.murphy@intel.com>
2021-08-24 07:26:08 -04:00