New research in next-gen optical interposers: mutliple chiplets connected with low latency

The future of the semiconductor industry is the new arms race, with European researchers showing that the use of optical interposers, based on silicon photonics, could be an effective method for chiplet interconnects and could have reduced communication delays.

2

CEA-Leti is a European technology research institute, with their new optical interposers called Starac, with the use of silicon photonics and its conventional techniques that makes this new tech unique, and very capable. Starac's active optical interposers have infused electronic and photonic circuitry into a single package, paving the way for complex data routing and processing.

• Read more: Check out these beautiful high-res die shots of AMD's new Zen 5 CCD

The new optical interposer technology uses a dedicated ONoC (Optical Network-on-Chip) which handles the high-speed data transmission between the chiplets, all without needing intermediate hops through the ring topology structure.

• Read more: Intel unveils first optical interconnect tech using photonics, 100x distance
• Read more: TSMC's next-gen COUPE tech: silicon photonics packaging will be ready in 2026

Starac hasn't been finalized just yet, so there's no real-world performance but the team at CEA-Leti says that the new technology will decrease latency, have higher bandwidth, and increase the power efficiency by a large margin, and could find its way into the semiconductor industry.

• Read more: Apple A18 Pro, A18 SoC die sizes compared: built on TSMC N3E node
• Read more: This is what 2.5D advanced packaging looks like: GPU logic die, HBM, interposer
• Read more: AMD's new Ryzen 9000 series CPU delidding 'goes wrong' with a cracked die
• Read more: Check out this beautiful die shot of AMD's new Ryzen AI 300 'Strix Point' APU

Jean Charbonnier, R&D project leader at CEA-Leti, said: "In a big compute system, there are several compute chiplets with cores and several HBMs [high-bandwidth memories]. This is true of the latest processors from Intel, AMD and NVIDIA. It's easy to go from a core to an HBM that's close by. But if you need to go from a core to a more distant HBM, then there's a whole sequence of operations you need to carry out to fetch the data".

"With our solution, the latency would be greatly improved, because the intrinsic latency of the light being guided within our optical network-on-chip is very small compared with a trip through all the hops that would be needed in more conventional architectures. We hope to establish industrial partnerships within the next year or so to help us work out some of the process and packaging issues and to bring us closer to the real-word problems this technology might solve".