A team of engineers from Technische Universität Kaiserslautern (TUK) is developing a software tool to support companies in the semiconductor industry in agile chip design. Through early testing and verification, the solution can help customers provide feedback and correct defects early in the design process. The engineers aim to market their tool under the name "LUBIS EDA". From April 12 to 16, they will present their platform at the digital Hanover Fair at the Rhineland-Palatinate research and innovation stand.
The engineers Tobias Ludwig, Michael Schwarz, and Dr. Max Birtel have joined forces with software developer Tim Burr to bring concepts from the software domain to chip development.
"In terms of hardware design, not much has changed in the industry over the last few decades," Ludwig explains. "The focus has been on making the existing process faster. The idea of completely redesigning this with the help of agile approaches and thus making a big leap forward when it comes to achieving time-to-market has not yet taken off."
The founding team is now offering the semiconductor industry the right toolbox to unleash this untapped potential. "Our software solution enables companies to transfer proven approaches from agile software development to the world of hardware," says Ludwig. "More customer proximity, faster releases, error minimization in initial design - all of this is also possible in hardware development."
The decisive advantage lies in early and continuous testing because it takes place not at the end, but after each adaptation step. This significantly reduces the total time required to verify the chip. "Based on experience, we can guarantee at least 10 percent time savings, just in testing," says Birtel. "Since the development costs for a chip range from two to as much as six million euros per project, depending on the complexity, it's obvious what savings potential opens."
The new methodology can be introduced easily, as the development tool can be operated in parallel with existing development environments. “Chip manufacturers can use our software to convert abstract specifications into a virtual prototype that provides all the functionality, before producing the hardware," Birtel says.
The agile system can be used to achieve all development goals that are relevant in the semiconductor industry - from the smallest possible chips to the most energy-efficient to the most powerful. "Almost 15 years of development work have gone into our tool. Now we are ready for pilot projects to evaluate our software solution in specific use cases," Birtel says.
The German Federal Ministry for Economic Affairs and Energy and the European Social Fund's development to market maturity was being funded until March 2021 as part of an EXIST research transfer called "Syncopate" (03EFORP026). Also, the start-up office of the TU Kaiserslautern and the Kaiserslautern University of Applied Sciences has advised the engineers.
How it all began: Ludwig further developed existing methods that enable agile hardware development as part of his doctoral thesis at the chair of Electronic Design Automation under Professor Dr. Wolfgang Kunz at TUK. Together with his doctoral colleague Schwarz, he recognized their potential, set his sights on founding the company, and brought on board Birtel, an industrial engineer who complements the technical engineer's perspective with business skills. Most recently, software developer Burr completed the team.
One of the most important classes of problems that all scientists and mathematicians aspire to solve, due to their relevance in both science and real life, is optimization problems. From esoteric computer science puzzles to the more realistic problems of vehicle routing, investment portfolio design, and digital marketing--at the heart of it all lies an optimization problem that needs to be solved.
An appealing technique often used in solving such problems is the technique of "quantum annealing", a framework that tackles optimization problems by using "quantum tunneling"--a quantum physical phenomenon--to pick an optimum solution out of several candidate solutions. Ironically, it is in quantum mechanical problems where the technique has found rather scarce application! "Chemists and materials scientists, who deal with quantum problems, are mostly unfamiliar with quantum annealing and so do not think to use it. Finding applications of this technique is therefore important for increasing its recognition as a useful method in this domain," says Prof. Ryo Maezono from Japan Advanced Institute of Science and Technology (JAIST), who specializes in applying information science to the field of materials science.
To that end, Prof. Maezono explored, in a recent study published in Scientific Reports, the phenomenon of ionic diffusion in solids, a topic of great interest in both pure and applied materials science, along with his colleagues, Keishu Utimula, a Ph.D. graduate in materials science from JAIST (in 2020) and lead author of the study, Prof. Kenta Hongo, and Prof. Kousuke Nakano, by applying a framework that combined quantum annealing with ab initio calculations, a method that calculates physical properties of materials without relying on experimental data. "While current ab initio techniques can provide information about diffusion path networks of the ions, it is difficult to map that information into useful knowledge of the diffusion coefficient, a practically relevant quantity," explains Prof. Maezono. ![In this figure, the hopping amplitude and existence of possible pathways for atomic migrations [panel (a)] can be identified at the microscopic level. But it is not easy to count all the chosen pathways using the magnitude of hopping amplitude [(b)]. To understand the properties of a material, it is necessary to be able to count the latter. CREDIT Ryo Maezono from JAIST In this figure, the hopping amplitude and existence of possible pathways for atomic migrations [panel (a)] can be identified at the microscopic level. But it is not easy to count all the chosen pathways using the magnitude of hopping amplitude [(b)]. To understand the properties of a material, it is necessary to be able to count the latter. CREDIT Ryo Maezono from JAIST](/images/261019_web_1a3f2.jpg)
Specifically, the team looked to calculate the "correlation factor", a key quantity in the diffusion process, and realized that this could be done by framing the process as a routing optimization problem, which is precisely what the quantum annealing framework is designed to solve! Accordingly, scientists calculated the correlation factor for a simple two-dimensional tetragonal lattice, for which they already knew the exact result, using quantum annealing and a variety of other computational techniques and compared their outputs.
While the evaluated correlation factors were consistent with the analytical result for all the methods employed, all the approaches suffered from limitations due to unrealistic computational costs for large system sizes. However, scientists noted that the computational expense for quantum annealing grew much more slowly in a linear fashion compared to the other techniques, which showed rapid exponential growth.
Prof. Maezono is excited by the finding and is confident that, with sufficient technological advancement, quantum annealing would present itself as the best possible choice for solving problems in materials science. "The problem of ion diffusion in solids is of central importance in building smaller batteries with higher capacity or improving the strength of steel. Our work shows that quantum annealing is effective in solving this problem and can expand the scope of materials science as a whole," he concludes.

How to resolve AdBlock issue? 