Replacing electronics with photonics will one day be an important way to run IT while consuming far less power than is the case today. But while that idea looks great on paper, the research is still young. The Internet’s voracious appetite for electricity needs some near-term solutions, so asThe Register followed-up the our …
But this reminds me of how great holographic memory was going to be.
Factor of 10 000 *below* the minimum
It's some times good to see how far we *still* have room to go.
I can't help feel that if the interview had been with Prof Malcolm Tucker it would have been more forthright.
You can guess what's in my pocket.
My physics teacher used to tell me that optics would soon replace the electronics I knew better than him 25 years ago. I still see little sign of it happening any time soon.
A river runs through your data center
Why worry about running air between the racks? Why not just boil nitrogen off the servers?
Re: A river runs through your data center
IIRC it costs energy to liquefy nitrogen, making it either expensive or problematic. Plus it's a natural asphyxiant (at concentrations above 80% of atmosphere), meaning a nitrogen leak can make the server room unsafe.
How are they going to make the photons? Probably with leccy, made with coal.
I'm sure we'll get our flying cars well before any of this is real.
I did my MS degree project on this, back in the mid-1980s. My conclusion? Electronics would win out for the computational elements, since electrons interact with matter a lot more easily than photons, at least at reasonable energy levels (All bets are off if the optical signal is strong enough to cause ionization, but, at that power level, all is lost anyway). Optical was the way to go for interconnects, whether long distance, to peripheral devices, or even between chips. There was some interest in optical for MIMD type computational operations (or even SIMD type computational operations), or for massively parallel operations, but these seemed doomed from the start.
There was a lot of interest, at least back in the 1980s, in crossbar switches for optical switching, rather than having to convert the optical signals to electricity, in order to do the switching, and then back to optical. But, while there was quite a bit of hope for using LCDs for such crossbar switching, nothing seems to have materialized. That might be an area of interest, especially since that could remove a lot of the power consumption in the routing function (Yes, some invention would be required.). Anyway, it may be an area of interest.
P.S. Mine's the one with the photons in the pocket.
Re: Optical computing
Yeah. Someone did it a decade ago. And they were real proud because the two peices of hardware (one for each end) cost more than installing the fiber between the two end-point cities.
Re: Optical computing
Sounds like the position optical computing was in the 1960's with photographic masks carrying out the FFT "instantly" compared to any computer implementation.
Possibly the furthest this work got was the Dunlop Aerospace developed "correlatron" which provided the terminal guidance for the Pershing 2 missile.
I'd suggest the 2 *big* challenges are 1) *gain* amplification of a light signal *by* a light signal and 2) eliminating interference techniques. I've seen *lots* of stuff which basically use this but it seems pretty useless if you want a device with *broad* bandwidth.
Perhaps the *next* generation will manage this.
And the answer is....
Iceland! Yes, that cold, bankrupt island with shedloads of clean geothermal power. Perfect for server farms even at today's (in-)efficiencies. Only lacking large enough pipes to the rest of the world, but no new technologies required.
Re: And the answer is....
"Only lacking large enough pipes to the rest of the world, but no new technologies required."
Well if enough companies agreed with that view things *could* change.....
chaos creates opportunity.