Enabling a new generation of system-level integration in data centers, NVIDIA today announced NVIDIA NVLink -C2C, an ultra-fast chip-to-chip and die-to-die interconnect that will allow custom dies to coherently interconnect to the company's GPUs, CPUs, DPUs, NICs and SOCs. With advanced packaging, NVIDIA NVLink-C2C interconnect would deliver up to 25x more energy efficiency and be 90x more area-efficient than PCIe Gen 5 on NVIDIA chips and enable coherent interconnect bandwidth of 900 gigabytes per second or higher.

"Chiplets and heterogeneous computing are necessary to counter the slowing of Moore's law," said Ian Buck, vice president of Hyperscale Computing at NVIDIA. "We've used our world-class expertise in high-speed interconnects to build uniform, open technology that will help our GPUs, DPUs, NICs, CPUs and SoCs create a new class of integrated products built via chiplets."NVIDIA NVLink-C2C is the same technology that is used to connect the processor silicon in the NVIDIA Grace Superchip family, also announced today, as well as the Grace Hopper Superchip announced last year. NVLink-C2C is now open for semi-custom silicon-level integration with NVIDIA technology.

NVIDIA NVLink-C2C supports the Arm AMBA Coherent Hub Interface (AMBA CHI) protocol. NVIDIA and Arm are working closely together to enhance AMBA CHI to support fully coherent and secure accelerators with other interconnected processors.

"As the future of CPU design is increasingly accelerated and multichip, it is critical to support chiplet-based SoCs across the ecosystem," said Chris Bergey, senior vice president and general manager of the Infrastructure Line of Business at Arm. "Arm is supporting a broad set of connectivity standards and designing our AMBA CHI protocol to support these future technologies, including collaborating with NVIDIA on NVLink-C2C to address use cases like coherent connectivity between CPUs, GPUs and DPUs."

NVIDIA NVLink-C2C is built on top of NVIDIA's world-class SERDES and LINK design technology, and it is extensible from PCB-level integrations and multichip modules to silicon interposer and wafer-level connections, delivering extremely high bandwidth while optimizing for energy and die area efficiency.

In addition to NVLink-C2C, NVIDIA will also support the developing Universal Chiplet Interconnect Express (UCIe) standard announced earlier this month. Custom silicon integration with NVIDIA chips can either use the UCIe standard or NVLink-C2C, which is optimized for lower latency, higher bandwidth and greater power efficiency.

Some of NVLink-C2C's key features include:

• High Bandwidth - supporting high-bandwidth coherent data transfers between processors and accelerators
• Low Latency - supporting atomics between processors and accelerators to perform fast synchronization and high-frequency updates to shared data
• Low Power and High Density - using advanced packaging, it is 25x more energy efficient and 90x more area-efficient than PCIe Gen 5 on NVIDIA chips
• Industry-Standard Support - works with Arm's AMBA CHI or CXL industry-standard protocols for interoperability between devices