A Guth Labs publication ↗
Back to News

EMASS releases ECS-Studio for deploying models on edge-AI chips

Green glass microchip on a cream circuit board surrounded by floating geometric blocks
AI-generated editorial illustration

EMASS announced ECS-Studio on November 18, a development suite that combines model porting, application building and hardware debugging for its ECS-DoT edge-AI chips.

The Nanoveu subsidiary says the software is available to registered developers on Windows, macOS and Linux. Its Visual Studio Code-based workspace bundles the chip's software-development kit, RISC-V toolchain and sensor libraries.

The compiler imports TensorFlow Lite, PyTorch and ONNX models and converts them to eight-bit integer representations. It reports whether operations will run on the chip's deep-learning accelerators or its RISC-V core and checks whether the model fits in on-chip memory.

A model library supplies pre-ported audio, vision and time-series examples. Developers can test these or their own models on the ECS-DoT-22 evaluation board.

The launch release explicitly says ECS-Profiler, a tool for measuring latency and memory use on hardware, is coming later. The broader product page describes that capability but should not be read as confirmation it shipped with this release.

The suite's practical role is reducing the integration work between a trained model and an embedded device. Actual performance still needs to be measured on the developer's model and hardware.

Sources and citations

Revision history

Published .