At ScaDS.AI Dresden/Leipzig, we focus on composable scalable agile, and secure computing architectures, investigated with a focus on the overall system architecture including software and hardware. Furthermore, we are exploring the benefits of our agile and scalable hardware/software computing platforms for various applications and therefore closely collaborate with other topic areas.
Our topic area is divided into three main directions:
In architecture, we focus on investigating highly flexible architectures that can adapt the performance and energy consumption to the application requirements to achieve better results under dynamic or unpredictable conditions.
In the direction of compilers and scalability, we focus on developing scalable, agile and secure computing architectures that enable the portability of AI and big data applications to existing CNM/CIM systems. We also develop novel accelerator designs, with an emphasis on performance, accuracy, and energy efficiency.
In Security, we focus on automatic detection of potential security issues and support for users without cybersecurity knowledge by using generative AI models to compensate for the user’s lack of expert knowledge.
Our strategic goals are to investigate composable and scalable computing architectures, including design and programming tools as well as online workflows. We hereby focus on cross-layer hardware-software-co design optimization, especially in heterogeneous systems consisting of CPUs, GPUs and FPGAs.