Fast GPU Interconnect over Radio(spectrum.ieee.org)
spectrum.ieee.org
Fast GPU Interconnect over Radio
https://spectrum.ieee.org/rf-over-fiber
12 comments
Tiny waveguides in cable form. Cute.
As someone with no experience in the domain, this sounds really interesting.
But one of the companies mentioned has been attempting this for 9 years.
Why hasn't this taken off already? It doesn't seem we're in need of any breakthroughs. So where do the economics break down?
But one of the companies mentioned has been attempting this for 9 years.
Why hasn't this taken off already? It doesn't seem we're in need of any breakthroughs. So where do the economics break down?
Precision manufacturing at scale. The physics of merging a hundred-gigahertz-scale circuit board track into a waveguide are very unforgiving. The physics governing the tolerances of said waveguide are similar.
It’s sorta funny to see one of the companies aiming for terahertz frequencies - long wave IR is only ~30THz.
It'd be nice to see the 'terahertz gap' closed finally.
Not for a practical reason, but it's been on my mind since I randomly found a wikipedia article about it.
Not for a practical reason, but it's been on my mind since I randomly found a wikipedia article about it.
Interestingly reading the terahertz radiation wikipedia article they mention peeling adhesive tape generates 2THz and 18THz peaks
Very clever stuff. I wonder how their power consumption compares to copper with a retimer.
> At 60 hertz—the mains frequency in many countries—most of the current is in the outer 8 millimeters of copper
That's a very fat copper wire!
The technology sounds interesting, but why wouldn't it have been developed previously? What's changed such that it is now deployable versus a decade or two ago?
That's a very fat copper wire!
The technology sounds interesting, but why wouldn't it have been developed previously? What's changed such that it is now deployable versus a decade or two ago?
For a moment I thought this was actual co-ax, which would be supremely ironic (it was used in the early days of Ethernet, but twisted pair proved cheaper).
But it looks like neither candidate has a conductive core, although they do have the conductive shell.
Coax, or more accurately twin-ax is still the underlying technology for Direct Attachment Copper cables for Ethernet using pluggable modules.
The whole point of this technology is to avoid the use of a conductive core, i.e. the use of the TEM propagation mode, in order to avoid the conductive losses caused by electrical currents that pass through the cable.
Instead of that, a propagation mode of the electromagnetic waves based on the reflection of the waves from the walls of the wave guide is used, like in optical fibers, but at much lower frequencies, in order to avoid the conversions between electrical signals and light.
Instead of that, a propagation mode of the electromagnetic waves based on the reflection of the waves from the walls of the wave guide is used, like in optical fibers, but at much lower frequencies, in order to avoid the conversions between electrical signals and light.
Both companies say their technologies can easily outdo copper in reach—spanning 10 to 20 meters without significant loss"
This is absolutely fascinating! For the longest time, I thought that optical fibers were the future, but waveguides (of whatever material appropriate) at whatever frequenc(y|ies) appropriate could give optical fibers a run (get it, a "run"? :-) ) for the money!
If we think about it, both fiber and copper cables are both very specific cases of a more broader
waveguide (first) principle...
That is, in theory you could make something that looks like a wire or cable out of any material(s) -- and if the material(s) and apertures and frequencies are correct, then you've created a transmission of path for data from point A to point B...
So, kudos to Point2, AttoTude (and other future companies!) that go down this technological tract! You're increasing both human knowledge (and data rates!) -- which could never be a bad thing!