The PCIe Gen 6 retimers improve connections between AI accelerators, GPUs, CPUs, and other components, addressing signal degradation and ensuring reliable communication.
Marvell Technology, Inc. has announced the expansion of its connectivity portfolio by introducing the new Alaska P PCIe retimer product line. This product line is designed to scale data centre compute fabrics inside accelerated servers, general-purpose servers, CXL systems, and disaggregated infrastructure. The first two products in this line are 8- and 16-lane PCIe Gen 6 retimers, which facilitate connections between AI accelerators, GPUs, CPUs, and other components within server systems.
The company claims that the retimers compensate for signal degradation and regenerate the signal, ensuring reliable communication over necessary distances. These retimers can be deployed on AI accelerator baseboards, server motherboards, and riser cards and integrated into active electrical cables (PCIe AEC) and optical cables (PCIe AOC) for emerging multi-rack server system architectures.
The 5nm 16-lane PCIe 6 retimer, with typical power consumption of 10 watts, operates at the lowest power in the industry today. Available in 8- and 16-lane configurations, these devices compensate for 40dB of channel loss compared to the spec of 32dB at PCIe 6. They can be used for on-board or cable copper connections or combined with electrical-to-optical components to produce optical PCIe modules, addressing different cloud customer data centre architectures.
Some of the key features of the Alaska P PCIe retimer include:
- Compatibility with PCI Express® Gen 6/5/4/3/2/1 and Compute Express Link™ 3/2/1.1
- Industry-leading PAM4 SerDes performance
- Low-latency mode for cache-coherent links
- Industry’s lowest power consumption (10W PCIe 6 x16)
- Industry-standard x16 and x8 footprints
- Advanced telemetry and diagnostics: in-band FEC monitoring, out-of-band SerDes eye monitoring, embedded logic analyzer, and software suite for fleet management in large-scale deployments
“Signal distance is a real dilemma for service providers. We estimate over 75% of cloud and AI servers shipping two years from now will rely on retimers and these servers will contain multiple retimers. An eight GPU server might contain 16 or more of these devices. Retimers will even percolate into enterprise servers,” said Alan Weckel, co-founder of 650 Group.
For more information, click here.
