embARC Machine Learning Inference Library

The embARC Machine Learning Inference (MLI) library provides software functions optimized for DSP-enhanced ARC EMxD and ARC HS4xD processors. It enables ARC customers to efficiently develop or port data processing algorithms based on machine learning (ML) principles. Supported ARC processors include:

CIFAR-10 CNN Model on ARC EM Processor

CIFAR-10 CNN Model on ARC EM Processor

MLI addresses a broad range of NN applications

Primarily addressing IoT related applications, the embARC MLI library extends Synopsys' artificial intelligence offering to multiple IoT use cases:

Application Example NN use cases
Voice-based human machine interfaces
  • Noise reduction
  • Speech Recognition
  • Acoustic context awareness
Personal fitness and health monitoring
  • Human activity recognition
  • Early disease prediction
Industrial IoT
  • Multisensory data fusion
  • Behavioral prediction
  • Acoustic fault detection

MLI kernels support multiple machine learning models

The embARC MLI software library provides a set of essential kernels for effective inference of small or mid-sized ML models. It enables the efficient implementation of convolutional neural networks (CNNs) [ex. classic and depth-wise convolutions], recurrent neural networks (RNNs) [ex. long short-term memory (LSTM) cells and basic RNN cells], fully connected layers, poolings, activation functions [ex. rectified linear units (ReLU)], and data routing operations [ex. padding, transposing, and concatenation], while reducing the power and memory footprint.

Leveraging the right processors for machine learning

ML-based applications intensively use classic DSP, RISC, and matrix operations, each with unique processing needs. ARC EM DSP and ARC HS DSP processors offer the best combination of power and area on the ML spectrum.


The embARC MLI software library is available through embARC.org, a dedicated website that provides software developers centralized access to free and open source software, drivers, operating systems, and middleware supporting ARC processors. The embARC MLI distribution is managed by Synopsys for the community and all contributions are welcome.

Say Welcome to the Machine - Low-Power Machine Learning for Smart IoT Applications


  • Optimized for low-power IoT applications that utilize convolutional neural networks (CNNs) and recurrent neural networks (RNNs)
  • Supports the energy-efficient, DSP-enhanced ARC EMxD and HS4xD Processors
  • Boosts performance up to 16x for 2D convolution layers compared to unoptimized implementations
  • Acceleration of RNNs up to 5x for a wide range of topologies including those built with long short-term memory (LSTM) cells
  • Distributed as free and open-source software through the embARC.org website