Colorless WDM P2MP link for GPU-HBM interconnects

As data-intensive applications such as AI and machine learning continue to grow, there is an increasing need for faster, more efficient data transfer between GPU and HBM. Optical interconnects, leveraging technologies like wavelength-division multiplexing (WDM), offer higher bandwidth density and lower power consumption compared to electrical solutions. With growing memory requirements, future GPU systems will need a larger number of HBMs using cost- and energy-efficient optical interconnects based on advanced network topologies. This research aims to explore WDM passive optical networks for GPU-HBM interconnects by developing a system model and PoC of a colorless WDM network on a photonic integrated circuit.
 

Required background: Engineering Technology

Type of work: 40% modeling/simulation, 50 experimental, 10% literature

Supervisor: Peter Ossieur

Daily advisor: Joonyoung Kim

The reference code for this position is 2025-018. Mention this reference code on your application form.