SiTime Precision can enable 3x better synchronisation and 800G network connectivity with 4x smaller footprint
Precision timing company SiTime Precision has unveiled its differential-ended SiT5977 Super-TCXO, becoming the newest member of the SiTime Elite RF family. The development delivers the most resilient performance for AI compute nodes with high bandwidth and network synchronisation whilst only being single-chipped.
Smart network interface cards (Smart NICs), acceleration cards, switches, and computer nodes are some of the applications the Super-TCXO can be used in, part of the wider $200Bn data centre market.
According to the IDC press release, IDC Report Reveals AI-Driven Growth in Datacenter Energy Consumption, Predicts Surge in Datacenter Facility Spending Amid Rising Electricity Costs: “IDC expects the surging demand for AI workloads will lead to a significant increase in datacenter capacity, energy consumption and carbon emissions, with AI datacenter capacity projected to have a compound annual growth rate (CAGR) of 40.5% through 2027.”
Dave Altavilla, president and principal analyst at HotTech Vision & Analysis, said: “Improving AI workload efficiency to reduce energy consumption and carbon emissions is an industry-wide challenge. Precision timing is one of the approaches to help solve this problem. SiTime is the only semiconductor company fully dedicated to developing innovative timing solutions required for the complex scaling of today’s AI datacenters.”
Having a more efficient AI cluster can bring lower total costs of ownership along with energy consumption. High bandwidth is required to enable clusters to work, as this allows for interconnectivity and tightly synchronised orchestration to minimise AI accelerator idle time. Precise network telemetry for AI operations is achieved by having tighter synchronisation.
Multiple timing components can be replaced with the SiT5977 alone, as it optimises the efficiency of AI compute clusters with 3x tighter synchronisation for data transmission and communication. When compared to other architectures, it is 4x smaller and enables more processing in compact systems, allowing architects to leverage high-speed 800G bandwidth network connectivity and maximise utilisation of the AI cluster.
Piyush Sevalia, executive vice president of marketing at SiTime, said: “AI training and inference are fundamentally distributed computing applications, which require accurate timing to synchronize activities. By enabling a new, more efficient architecture, the Elite RF timing solution uniquely supports more efficient AI workload processing, which may lead to higher revenue and lower TCO for datacenters.”
The architecture is environmentally robust with ±1 ppb/°C frequency slope (dF/dT) for optimum performance under airflow and thermal shock, whilst also capable of driving 800G and higher links via 80 fs phase jitter and LVDS outputs.
It enables embedded control loops with precise digital tuning of output frequency (DCTCXO), ±400 ppm pull range and 0.05 ppt (5e-14) resolution via I2C/SPI and eliminates link flaps from quartz timing activity dips or micro jumps.
Protection-wise, it is resistant to shock, vibration and board bending, eliminates external LDOs via on-chip voltage regulators, and 156.25 MHz output frequency enabling high-speed SerDes and 800G links.
