HSE and Samara University boost chip reliability with new algorithm

Researchers from HSE MIEM and Samara University have created LRF-3D, a new algorithm that enhances processor reliability by automatically bypassing faulty nodes within three-dimensional networks-on-chip. The system employs a hierarchy of eight local algorithms to address “dead ends” caused by manufacturing defects or crystal degradation in the on-chip network, a common issue for data centers and supercomputers.

Aleksandr Romanov, Leading Research Fellow at HSE MIEM, explains that the algorithm works much like a car navigation system, finding an alternative route if a familiar one is blocked. Testing across 36 scenarios demonstrates LRF-3D achieves a path length deviation of only 1.64% from a reference algorithm, even with up to 50% of nodes failing.

LRF-3D Algorithm Bypasses Faulty Nodes in 3D Networks-on-Chip

Developed by researchers at HSE MIEM and Samara University, LRF-3D addresses a critical challenge in modern chip design: maintaining functionality despite manufacturing defects or the gradual degradation of silicon crystals. These imperfections create “dead ends” within the intricate on-chip communication network, potentially halting data flow and compromising performance in data centers, supercomputers, and artificial intelligence systems.

The LRF-3D system distinguishes itself through a hierarchical structure comprised of eight local algorithms, each designed to navigate around these faulty nodes without requiring a complete map of the network. The researchers validated LRF-3D’s performance across 36 scenarios, including mazes, corridors, and various failure patterns.

Testing revealed substantial improvements over existing routing methods; with a 13-30% node failure rate, LRF-3D successfully delivered data packets in 86% of attempts. The algorithm’s speed surpasses that of established alternatives, making routing decisions 16.7 times faster than the A* algorithm and 22.5 times faster than LOFT, HSE MIEM says. The team reports that LRF-3D outperforms LOFT by a factor of more than 137 in terms of route quality.

Romanov anticipates future work will focus on real-world chip testing to evaluate power consumption and data transfer rates, and suggests that if the results are confirmed, the technology could help create a new generation of multiprocessor systems-on-chip that operate more reliably, even under challenging conditions. The study received support from the Russian Science Foundation, and the findings are published in IEEE Access.

Stay current

See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.

Avatar of Ivy Delaney

Ivy Delaney

Ivy Delaney has been working with neural networks and machine learning since the mid-nineties, back when a couple of hidden layers and a long afternoon of training counted as ambitious. She has watched the field go from academic curiosity to the thing quietly running underneath everything, and she brings that long view to quantum computing. For Quantum Zeitgeist she covers the ground where the two fields meet. That means quantum machine learning and the variational algorithms it leans on, and it also means the less glamorous but more interesting story of classical machine learning already doing real work inside quantum machines, decoding error-correcting codes, calibrating noisy hardware and learning the error models that simulators depend on. She writes about the hardware those algorithms have to run on too, and about the post-quantum cryptography scramble that the same hardware has set off. Her stories typically start with the paper, whether that is peer-reviewed work, conference proceedings or an arXiv preprint, with the source linked so you can hold a claim up against the research it came from. She is unimpressed by benchmarks that will not say what they beat, and by demonstrations that only work in the press release.

Latest Posts by Ivy Delaney: