Artificial Intelligence (AI) has demonstrated strong capabilities across various domains over the past decade. Edge and specifically mission-critical applications, such as automotive and aerospace, require both high performance and efficiency without compromises in security and reliability. This stems from tightly constrained power consumption, failures that can have catastrophic consequences and devices that may be physically accessible to malicious actors. AI algorithm deployment to hardware also presents significant barriers, requiring specialized knowledge and expensive development tools. The DI-EDAI project aims to offer a holistic approach for connecting high-level AI algorithms with hardware implementations while tackling the aforementioned issues. Unlike other approaches that address individual aspects of the AI deployment flow, we investigate solutions across multiple layers of the design stack. Through our work we develop efficient hardware, map AI algorithms to hardware while simultaneously ensuring security and reliability. Furthermore, we leverage AI-techniques to assist with Electronic Design Automation (EDA) workflows for design optimization, verification and implementation. Our open source approach aims to reduce entry barriers, promote transparency and education, and spark innovation. This paper presents the current state of the DI-EDAI project at midterm, highlighting our latest contributions, identifying limitations in existing state-of-the-art approaches, and outlining ongoing work to address these gaps.

Chip technologies are crucial for the digital transformation of industry and society. Machine Learning (ML) and Artificial Intelligence (AI) are increasingly shaping both daily life and industrial applications, with AI hardware playing a vital role in enabling efficient and scalable ML deployment. However, significant challenges remain in bridging the gap between ML algorithm development and hardware implementation, particularly for edge ML applications where efficiency, power constraints, and adaptability are critical. In such resource-constrained environments, hardware-software co-design becomes essential to achieve the necessary trade-offs between performance, energy efficiency, and system responsiveness. One of the key bottlenecks in ML hardware development is the lack of seamless integration between ML toolchains and electronic design automation (EDA) tools for hardware synthesis and mapping. Current solutions often require extensive manual optimization and costly proprietary software, limiting accessibility and innovation. Open-source tools can play a transformative role in democratizing ML hardware design, fostering collaboration, and addressing the growing shortage of skilled professionals. This paper covers key aspects of hardware-software co-design for ML systems, such as ML algorithms, hardware design, compiler technologies and system security, with a focus on open-source solutions. We highlight the critical need for open-source toolchains that connect ML model development with hardware synthesis and optimization and present solutions for custom hardware, as well as FPGA accelerators.

Communication in optical downlinks of low earth orbit (LEO) satellites requires interleaving to enable reliable data transmission. These interleavers are orders of magnitude larger than conventional interleavers utilized for example in wireless communication. Hence, the capacity of on-chip memories (SRAMs) is insufficient to store all symbols and external memories (DRAMs) must be used. Due to the overall requirement for very high data rates beyond 100 Gbit/s, DRAM bandwidth then quickly becomes a critical bottleneck of the communication system. In this paper, we investigate triangular block interleavers for the aforementioned application and show that the standard mapping of symbols used for SRAMs results in low bandwidth utilization for DRAMs, in some cases below 50 %. As a solution, we present a novel mapping approach that combines different optimizations and achieves over 90 % bandwidth utilization in all tested configurations. Further, the mapping can be applied to any JEDEC-compliant DRAM device.

Modern and future AI-based automotive applications, such as autonomous driving, require the efficient real-time processing of huge amounts of data from different sensors, like camera, radar, and LiDAR. In the ZuSE-KI-AVF project, multiple university, and industry partners collaborate to develop a novel massive parallel processor architecture, based on a cus-tomized RISC-V host processor, and an efficient high-performance vertical vector coprocessor. In addition, a software development framework is also provided to efficiently program AI-based sensor processing applications. The proposed processor system was verified and evaluated on a state-of-the-art UltraScale+ FPGA board, reaching a processing performance of up to 126.9 FPS, while executing the YOLO-LITE CNN on 224x224 input images. Further optimizations of the FPGA design and the realization of the processor system on a 22nm FDSOI CMOS technology are planned.

The simulation of Dynamic Random Access Memories (DRAMs) in a system context requires highly accurate models due to the complex timing and power behavior of DRAMs. However, cycle accurate DRAM models often become the bottleneck regarding the overall simulation time. Therefore, fast but accurate DRAM simulation models are mandatory. This paper proposes two new performance optimized DRAM models that further accelerate the simulation speed with only a negligible degradation in accuracy. The first model is an enhanced Transaction Level Model (TLM), which uses a look-up table to accelerate parts of the simulation that feature a high memory access density for online scenarios. The second model is a neural network based simulator for offline trace analysis. We show a mathematical methodology to generate the inputs for the Look-Up Table (LUT) and an optimized artificial training set for the neural network. The enhanced TLM model is up to 5 times faster compared to a state-of-the-art TLM DRAM simulator. The neural network is able to speed up the simulation up to a factor of 10×, while inferring on a GPU. Both solutions provide only a slight decrease in accuracy of approximately 5%.