OK, we’re going to have to be careful with this one because ipXchange’s pic for the top MCU of 2023 could easily slip into the rest of the product family, which fuses the humble 32-bit microcontroller with application processor cores to run so much more than a standard MCU ever could.
That said, if you were ever thinking of greatly improving your product with AI functionality, these are the MCUs to consider, hands down. Don’t worry. All the development tools will be familiar to you.
You guessed it: ipXchange’s winner of best MCU discovery in 2023 is Alif Semiconductor’s Ensemble family of MCUs and ‘Fusion processors’; to stay within this category, lets focus on why the pure MCUs of the series, the E1 and E3 devices, get the top spot.
In short, Alif have changed the game when it comes to running AI workloads on MCUs. We’re not talking about TinyML here; we’re talking about serious image- and audio-recognition workloads, and more, at high frame rates, but more importantly, high energy efficiency. The secret is an MCU architecture that has been redeveloped from the ground up and presents the first market-ready implementation of Arm’s Cortex-M55 MCU core with the Ethos-U55 microNPU, a hardware-based accelerator for AI.
But let’s let the numbers do the talking. When running an image recognition workload, Alif’s novel MCU architecture performs 800x faster than the previous Cortex-M generation, 78x faster than the M-55 alone (8 ms vs 624 ms), and uses 76x less energy than the M-55 alone (3 mJ vs 228 mJ) – these numbers are per inference. This means two things for you as a designer:
- Alif’s MCU’s set a new bar for AI functionality in MCU-based systems
- High-level AI can now be done in a long-life, battery-operated, MCU-based system
The two MCUs of the Ensemble family offer a single- (E1) or dual-core (E3) implementation of Alif’s highly efficient, AI-capable MCUs – the latter can be used for a two-tier inferencing system for making some serious power savings. For example: the first core is connected to a PIR sensor in a home security system. When something passes in front of the detector it wakes up and performs a simple image recognition task from a camera feed – is this a dog, a leaf, or a person of interest? Then the second core can wake up and do the important task – is this someone I recognise?
We at ipXchange could go on and on about why these are the top MCU of 2023, so read our full technical writeup on the board page, watch the interview, and if your imagination has been sparked, apply for an evaluation board for your next big project.
Keep designing!
You must be a member to leave a comment. Sign up or Sign in