In the era of data centric computing, there is a critical demand for systems that provide high performance while maintaining robust data reliability. To address this demand, Dates Lab investigates error resilient memory architectures that sustain high throughput and ensure data integrity. This research ultimately strengthens RAS features, including reliability, availability, and serviceability.
Modern IC design faces a critical "Complexity Gap," where the scale of transistors grows exponentially while designer productivity lags behind. DATES focuses on bridging this gap by advancing electronic design automation (EDA) through novel, high-performance methodologies. By enhancing the traditional design flow—spanning from system specification to physical implementation—with intelligent and automated solutions, we aim to significantly optimize PPA (Power, Performance, Area) and ensure the reliability and testability of complex integrated systems.
Memory architecture becomes more critical in deep learning operations and big data system. To cope with new requirements such as speed, power, bandwidth and so on, DATES researches on opportunities of memory architectures. This not only includes conventional architectures, but also next generation memories and their architectures.
As artificial intelligence models grow in complexity, conventional hardware faces significant bottlenecks in performance and energy efficiency. DATES addresses this challenge by designing next-generation Neural Processing Units (NPUs). We focus on developing high-performance, energy-efficient architectures and robust computing systems that can reliably support the demanding workloads of future intelligent applications.
Quantum computing offers remarkable performance gains for problems that are intractable with classical computing. DATES focuses on fault-tolerant quantum computing and optimization for quantum algorithms.
