MENU

Scalable sub-pJ/bit chiplet interconnect with 10m reach

Scalable sub-pJ/bit chiplet interconnect with 10m reach

Technology News |
By Jean-Pierre Joosting



Avicena is announcing its new scalable LightBundle chiplet interconnect at OFC 2024 in San Diego, CA.

The chiplet interconnect extends ultra-high density die-to-die (D2D) connections up to 10m at multi-Tbps/mm shoreline bandwidth density and class leading sub-pJ/bit energy efficiency. Based on Avicena’s LightBundle platform which supports unprecedented shoreline density and energy efficiency, it unlocks increased performance from HPC and AI cluster architectures.

AI is driving an unprecedented surge in demand for compute and memory performance, driven by applications like ChatGPT based on large language models (LLMs). This requires ever higher density, lower power, longer interconnects between processors and different types of memory modules.

Today, High Bandwidth Memory (HBM) modules must be located within a few millimeters of a Graphics Processing Unit (GPU), limiting accessible memory bandwidth and capacity based on GPU shoreline. The LightBundle chiplet interconnect extends HBM and other ultra-high performance D2D connections up to 10m while dissipating < 1pJ/bit for the optical interconnect and supporting multi-Tbps/mm beachfront density. This enables GPUs and other high-performance ICs to greatly increase their total IO bandwidth, accessing vastly more HBM and relieving inter-processor bottlenecks.

The LightBundle chiplet is compatible with standard multichip packaging and supports a wide range of D2D interfaces including standard and advanced versions of UCIe and BOW. Avicena is working with selected partners on different implementations. Initial prototypes will be available in the second half of 2025.

“The first D2D implementation will be an 8-Tbps UCIe advanced interconnect with a total chiplet footprint of 4- x 7-mm, beachfront density of 2-Tbps/mm and power consumption of < 12 W,” says Bardia Pezeshki, Founder and CEO of Avicena.

“As generative AI continues to evolve, the role of high bandwidth-density, low-power and low latency interconnects between GPUs and HBM modules cannot be overstated”, says Vlad Kozlov, Founder and CEO of LightCounting. “Avicena’s microLED based interconnect technology has the potential to deliver a paradigm shift in die-to-die interconnects.”

https://avicena.tech
OFC 2024

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News

Share:

Linked Articles
10s