"The frames get rebuilt from the ground up." Really? Does that include the data? I could not believe I was reading a "review". It read like marketing babble from the company that make the switch. I loved the bit about how turning the LEDs off resulted in better sound quality. Of course it did...
I think they forgot a huge marketing opportunity, though. The switch clearly needs a separate linear power supply with a huge R-core transformer. You know... For maximum audiophoolery.
Regarding phase noise: As someone who spent seven years for National Semiconductor and TI chasing the last dBc of phase noise, I can tell you that phase noise is very real. That doesn't mean that low installing the latest femtosecond clock source will do anything for audio quality, however. Most (all?) modern DACs include some form of clock jitter cleaner or clock/data recovery circuit on their inputs. I know for sure that's the case with the ESS DACs, for example. Now the name "clock jitter cleaner" is a bit misleading. A jitter cleaner is more of a "jitter replacer". It synchronizes a known clean clock from a VCO to the incoming dirty clock using a PLL. The PLL forms a lowpass filter for the incoming clock and a highpass filter for the VCO clock. The bandwidth of the PLL is low, which means that the DAC will see only the VCO phase noise within the bandwidth of interest (i.e. the audio band). Thus, the internal clock in the DAC has exactly the same jitter regardless of the cleanliness of the incoming clock. That's probably a good thing as I doubt the various femtosecond clocks out there actually meet their specs. Femtosecond noise floors are hard to reach and without $150k to blow on an HP/Agilent/Keysight/Nom d'Jour E5052B, we won't know if they meet their specs.
Tom