tflite-modeldirectories from the downloaded '.zip' file into the
source/directory of the example-standalone-inferencing-alif repository. Make sure to overwrite any existing files in the
ei_main.cpp. In this minimal code example, inference is run from a static buffer of input feature data. To verify that our embedded model achieves the exact same results as the model trained in Studio, we want to copy the same input features from Studio into the static buffer in
ei_main.cpppaste the raw features inside the
static const float featuresdefinition, for example:
ei_main.cppto run classification on live data.
-DTARGET_SUBSYSTEMparameter when building. The commands below all default to the high performance core, but you can easily switch cores by swapping any
app.axfwill now be available in the
build/bindirectory, and note whether you built the application for the high performance or high efficiency core
Secure Enclaveto manage secure firmware storage and bootup in production environments. Alif provides documentation on converting .axf files for use with their secure enclave, and then programming the resulting binary regions to the secure enclave in AUGD0002.
armclangthe linker file attempts to place as much as possible in DTCM, and overflows into SRAM if needed.
gcc, the linker is unable to auto place based on size. If you get an error during link, see ensemble.ld and un-comment the line that places the model in SRAM (instead of DTCM). This will only slow down DSP, as the U55 has to use the SRAM bus to access the model regardless of placement.