Our analogue computer project REDAC has reached an important milestone towards an analogue mainframe computer: together with Anabrid, we have successfully put the REDAC prototype into operation at the Ulm Innovation Centre and carried out the first calculations on it. This makes it clear that Anabrid’s modular approach to a reprogrammable analogue computer works. The next step is to increase the number of analogue computer modules and thus exploit the full potential of the REDAC.
The REDAC is particularly suitable for solving differential equations, simulating complex models and predicting dynamic systems and, despite its high speed and throughput, operates with minimal or no cooling requirements.
The REDAC is designed to solve industry-relevant problems in areas such as realistic simulations of turbulent flow dynamics, structural and flight dynamics, control systems, route optimisation, big data analyses and real-time sensor processing. It can also be used as a specialised, reconfigurable hardware-in-the-loop system for complex measurement, control and regulation systems. This is relevant for aerospace applications at DLR, for example.
Anabrid explains in detail how the REDAC works and how it can be used in the public manual.
Modular for the Future of Compute
For the current milestone, Anabrid combined two iREDAC modules in one rack: two chassis with six mREDAC modules each (iREDAC stands for intermediate REDAC, mREDAC for minimal REDAC).
In total, the REDAC prototype has over 2,000 arithmetic elements and 40,000 switches for routing arithmetic signals. This enables REDAC to solve normal and partial differential equations with high speed, precision and correctness. This makes it an interesting tool for researchers and developers – also at DLR QCI. The mission planning project QMPC, our battery project BASIQ, the analogue quantum computing machine AQuRA and the QML project Quant²AI are now working on the first REDAC.
By the end of the year, however, Anabrid will build an even larger and more complex REDAC system for us with even more iREDAC racks. It can be used to solve either even more complex problems or even more problems simultaneously. Until then, the feedback from the QCI projects, which are already working with the 2-rack REDAC, is important: They too are new pioneers in the development and discovery of a compute paradigm that is as unusual as it is promising.
But with the handover and acceptance of the final REDAC, the real work on the future of analogue computing has only just begun: Anabrid’s next goal is to miniaturise the space-filling rack structure to chip size.
anabrid
The deep-tech start-up anabrid has been developing modern analogue and hybrid computers since 2020. The founders have decades of experience in analogue computing, circuit design, IC design and in industrial application areas. The company’s medium-term goal is to develop a highly integrated analogue computer on a microchip that can be used as a coprocessor in industry and research.