The coolest thing about this article is that it's some work done by Nubis, I've been really excited to see where they go ever since Peter Pupalaikis left Teledyne LeCroy to work there/found the company. Word is he had a great presentation last week at DesignCon, unfortunately I couldn't make it but heard great things from a colleague.
I scrolled down looking for a cable schematic an all I found were ads. Did they leave out the cable schematic because it's largely indistinguishable from the previous iteration?
I hope this isn't a new level of disappointment in 'journalism' that I should brace myself for in 2024.
Pretty dang close, propagation delay is often given as a percentage of the speed of light (c). Here's a fun table with a comparison a few pages into this PDF:
It's all about the dielectric constant of the propagation medium. If you're building a system with copper for PCIe gen6, which is 64Gbps PAM-4, you're probably using something like Isola MT77 or Tachyon with Dk around 3.2 and very low dielectric loss. That being said though, I believe the signal attenuation in a FO system would be far better than copper interconnects.
Even when Thunderbolt was still codenamed Light Peak and had plans to use fiber, it had the goal of encapsulating a variety of protocols, not just PCIe.
