F2 · Embedded & Security
Using Arm Tools to Enable Machine Learning On-Device (2018)
Overview
Using an MNIST CNN as a case study, this project combines DS-5/Streamline, C API refactoring and a SystemC/TLM accelerator model on a Cortex-A55 FVP to build an on-device ML analysis flow before physical hardware is ready. By observing instruction counts, memory access and hardware/software partitioning, it ties model execution behavior back to system-architecture decisions. Its ecosystem value is composing Arm virtual platform, performance analysis and hardware modeling into an early-stage ML/accelerator exploration baseline.
Scope of verification
- Results come from a Cortex-A55 FVP, MNIST CNN and a SystemC/TLM accelerator model, used to observe instructions, memory behavior and architectural assumptions — not a physical-silicon benchmark.
- Model and accelerator behavior depend on simulation parameters; timing, power, memory-system effects, driver integration and production inference performance on real hardware still require separate validation.
Canonical source: https://developer.arm.com/community/arm-community-blogs/b/tools-software-ides-blog/posts/using-arm-tools-to-enable-machine-learning-on-device
This page is a guided summary with boundary notes, not an official reproduction.
This page is a guided summary with boundary notes, not an official reproduction.