Sync-n-Scale™
NEL Frequency Controls has developed an exciting new distributed clocking technology known as Sync-n-Scale™. For starters, imagine a clock tree with virtually unlimited scalability with agent-to-agent skews that don't increase as the number of branches increase and jitter that actually decreases as the number of branches increases. Additionally, the distributed design means that the frequency sources are redundant and there is no single point of failure. And of course, all frequency sources operate in strict frequency lock, so the entire distributed clock is a single clock domain. Just as the scalability is virtually unbounded, so is the potential physical size of the clock tree. With it's low-loss, low frequency interconnect, the distribution size is not bounded by high-frequency losses of traditional square wave digital distribution. The small, form factor parts are comparable in size to traditional clocking components – making this distributed approach perfectly suitable for everything from small blade cabinets to large, building-sized supercomputer applications. In addition the low-frequency, low amplitude sine wave synchronizing distribution greatly reduces electromagnetic emissions as compared with square wave distribution. Also, Sync-n-Scale™ is hot-swappable, RoHS compliant and available today.
Reducing interconnect latency and complexity is now achievable with NEL's flexible and reliable synchronous clocking solutions. Asynchronous compute nodes have been the norm for several computer generations due to the scalability limitations of traditional clock trees and single point of failure concerns with traditional cascaded fan-out buffers driven by a single oscillator. However, passing data across asynchronous clock domains involves added latency along with the inherent complexity introduced. However, if all domains are truly synchronous (i.e. everything one clock domain), then there is no need for the additional hardware complexity and latency is reduced as well. Simpler hardware leads to lower cost, less design, more straightforward verification and faster time to market; and in this case -- more performance.
Just as hardware can benefit from synchronous operation, operating systems and applications can as well. Operating systems rely on an accurate system-wide set of interval time counters to precisely schedule processes and effectively utilize all compute resources in the system. If the interval counters drift due to different sources clocking individual compute nodes, then performance is lost either through slipping cycles to re-synchronize, and/or running code to constantly compare and re-synchronize interval counters.
Both hardware and software can be simplified, latency reduced and performance increased through NEL's proprietary synchronous, Sync-n-Scale™ ultra-scalable, ultra low-jitter, redundant, hot swappable, clocking solutions.
Contact Us at 262-763-3591 for more information!