Memory footprint optimization

When developing an application, ROM and RAM footprint are important factors, especially when the firmware runs on the most resource-constrained devices like nRF52810 or nRF52811.

General recommendations

To reduce the memory footprint, ensure that your application uses the minimum required resources and tune the nRF Connect SDK configuration parameters. Complete the following actions to optimize your application:

  • Use the Memory report feature in the nRF Connect for VS Code extension to check the size and percentage of memory that each symbol uses on your device for RAM, ROM, and partitions (when applicable).

  • Follow the guides for optimizing Zephyr. Also see the implementation of the Minimal footprint sample.

  • Analyze stack usage in each thread of your application by using the Thread analyzer. Reduce the stack sizes where possible.

  • Limit or disable debugging features, such as logging or asserts.

  • Go through each component and subsystem and turn off all peripherals and features that your application does not use.

The following subsections give more information on how to optimize specific subsystems.

Bluetooth

Besides applying General recommendations, you can also complete the following actions to optimize the Bluetooth® part of your application:

For reference, you can find minimal footprint configurations of the Bluetooth: Peripheral LBS sample in nrf/samples/bluetooth/peripheral_lbs/prj_minimal.conf and the Bluetooth: Peripheral UART sample in nrf/samples/bluetooth/peripheral_uart/prj_minimal.conf.

Bluetooth Mesh

Besides applying General recommendations and Bluetooth optimization actions, there are some configuration options you can use to optimize the Bluetooth Mesh part of your application. Changing any of these options will change the functional capabilities of the Bluetooth Mesh device, and thereby result in changes to RAM and flash memory footprint.

Changing the values of the following options will affect the RAM footprint and the amount of space needed for persistent storage of the associated configuration data:

Changing the values of the following options will only affect the RAM footprint:

Model configuration options that affect stack size at runtime:

Gazell

To optimize the Gazell memory footprint, follow the General recommendations. Specifically, study the ISR stack size. If your application is in a pairing device, pay attention to the system workqueue stack size.

Do not enable features that your application does not use. The configuration options default to disabling optional features, such as:

  • Pairing

  • Pairing encryption

  • Pairing settings persistent storage

To reduce the logging level, set the CONFIG_GAZELL_LOG_LEVEL_CHOICE Kconfig option.

Matter

Besides applying General recommendations, you can also complete the following actions to optimize the Matter part of your application:

  • Make sure Zephyr’s Shell is disabled for your application. Related configuration options are listed in a dedicated section in each Matter sample’s prj.conf file.

  • Use prj_release.conf for building the application. The release configuration has a smaller memory footprint than the default, debug-enabled prj.conf.

  • If the logs in your application do not use the default log level, you can change the default log level of Zephyr modules from info to warning by setting CONFIG_LOG_DEFAULT_LEVEL to 2.

  • Change the log level of the Matter logs from debug to info by setting CONFIG_MATTER_LOG_LEVEL_INF to y.

  • Reduce the verbosity of assert messages by setting CONFIG_ASSERT_VERBOSE to n.

  • Check Thread memory footprint optimization actions, as the Matter application layer uses OpenThread.

Additionally, you can turn off logging for single Matter modules on the Matter SDK side, as described in Cutting off log regions for Matter SDK modules.

NFC

The Near Field Communication (NFC) protocol implementation in the nRF Connect SDK provides some options for optimizing memory footprint for both the tag and the poller roles.

Tag

To optimize your application that supports the NFC tag functionality, follow the General recommendations. The NFC Type 2 Tag and Type 4 Tag libraries do not provide configuration options that have an effect on memory usage in an application. However, there are a few Kconfig configuration options you can use to optimize memory usage related to NFC.

For an application that uses the Type 4 Tag library, you can set the following options:

Poller

To optimize an application that supports the NFC poller functionality using the NFC Reader ST25R3911B library, you can set the following options:

If the application supports the Tag NDEF Exchange Protocol (TNEP), you can set the following options:

If the application uses the NFC TNEP protocol for the Bluetooth LE out-of-band pairing procedure (NFC TNEP Connection Handover), regardless of the role (tag or poller), you can set the following options:

Thread

The current Thread memory requirements are listed on the OpenThread memory requirements page.

Besides applying General recommendations, you can also complete the following actions to optimize the Thread part of your application:

Wi-Fi

The current Wi-Fi® memory requirements are listed on the Memory requirements for Wi-Fi applications in Station mode and Memory requirements for Wi-Fi applications in Scan mode pages.

In addition to applying General recommendations, you can also see the Host device considerations page to optimize the Wi-Fi stack of your application. Specifically, you can refer to the section on networking stack or memory fine-tuning controls which lists Kconfig options that can be used to reduce memory requirements of your application.

Zigbee

The current Zigbee memory requirements are listed on the Zigbee memory requirements page.

Apply General recommendations to optimize the Zigbee part of your application.