REGISTER | LOGIN
Breaking News
News & Analysis

IBM Solves Quantum Computing

Google architecture obstacle to commercialization
4/29/2015 05:00 AM EDT
26 comments
NO RATINGS
2 saves
More Related Links
View Comments: Newest First | Oldest First | Threaded View
Page 1 / 3   >   >>
alex_m1
User Rank
Author
Re: Google Looses Out to IBM
alex_m1   8/25/2015 8:21:11 PM
NO RATINGS
Thanks.

R_Colin_Johnson
User Rank
Author
Re: Google Looses Out to IBM
R_Colin_Johnson   8/25/2015 7:55:18 PM
NO RATINGS
Looks like it was published in some pretty high-profile places, so I'm sure Intel is aware of the possibilities. I'll ask around to see if anyone's will to spill-the-beans on such a project at Intel.

alex_m1
User Rank
Author
Re: Google Looses Out to IBM
alex_m1   8/25/2015 7:24:20 PM
NO RATINGS
Happy to help.

 

it was published in all these places:

 

https://scholar.google.co.il/scholar?cluster=14424480274103347040&hl=en&as_sdt=0,5

R_Colin_Johnson
User Rank
Author
Re: Google Looses Out to IBM
R_Colin_Johnson   8/25/2015 2:10:30 PM
NO RATINGS
Yes, I would imagine so. Intel will of course want to combine CPUs with FPGAs to make them smart, but some sort of memristor material seems logical. Unfortunately, both Micron and Intel are keeping their cards close to their vest regarding just what material they are using, except to say it causes "bulk" changes at the crossbar intersections--rather than filaments.

Thanks for the reference to the paper on memristorFPGAs (mrFPGAs). Facinating reading. I can't find a publication or date. The latest reference is back in 2010 (from a cursory scan). Do you know where/when it was published?

alex_m1
User Rank
Author
Re: Google Looses Out to IBM
alex_m1   8/25/2015 10:17:43 AM
NO RATINGS
Colin: an unrelated question -

Given that intel's new 3DXpoint memory is pretty similar to memristor( in layout ) an given that memristors hold great promise for FPGA's [1], and of course, intel owns Altera - are we're going to see intel creating breakthrough FPGA's ?

 

[1]http://cadlab.cs.ucla.edu/software_release/mrVPR/mrFPGA.pdf

R_Colin_Johnson
User Rank
Author
Re: power dissipation
R_Colin_Johnson   5/4/2015 8:50:54 PM
NO RATINGS
traneus: I see your point. Refrigerators are fine for prototyping, but not for production. They did mention liquid helium, but I don't remember the exact context. Also I think that before commercialization, the will have to get quantum computers to operate at liquid nitrogen or higher temperatures, which is a long way down the road, barring some breakthrough. Right now they are trying to prove it can be done, becasue once you do that its often amazing what engineers will think of to successfully commercialize. On the other hand they may end up like bubble memories--an almost there technology obsoleted by others like flash. Thanks for the careful research and insightful comments.

traneus
User Rank
Author
Re: power dissipation
traneus   5/4/2015 8:38:19 PM
NO RATINGS
R_Colin_Johnson: Yes, plug-in-the-wall refrigerators exist. However, their cooling power at 0.015 K is tiny. Next time you interview the researchers, ask them how much power their refrigerator can pump out of a sample at 0.015 K. I attempt an estimate as follows:

At iceoxford.com I found dilution refrigerators capable of 0.015 K temperatures. Their largest unit is rated to extract 450 microwatts at 0.1 K, and to be able to cool down to 0.007 K (at which temperature the refrigerator can't extract any power from the sample because all its cooling power goes to removing heat leaking in from outside).

At en.wikipedia.org/wiki/Dilution_refrigerator, I read that the difficulty of heat transfer worsens inversely with the fourth power of temperature. Other difficulties get worse faster, but I use T^4 here:

0.1 K / 0.015 K = 7 approximately. 7 raised to the fourth power is 2401.

The refrigerator than extract 450 microwatts at 0.1 K, will be able to extract 450/2401 = 0.2 microwatts at 0.015 K.

0.2 microwatts is plenty for a demo of a few qubits. I wonder what they will do for millions or billions of qubits for a useful system. Where will they get the helium-3?

Dilution refrigerators operate by circulating both helium-4 and helium-3, so there is an unobtanium problem.

Remember that quantum computers are quantum analog computers. They have strengths and weaknesses similar to those of classical analog computers: high speed, limited accuracy, useful only for certain problems.

R_Colin_Johnson
User Rank
Author
Re: power dissipation
R_Colin_Johnson   5/4/2015 5:53:36 PM
NO RATINGS
traneus: I brought that up with the researcfhers during the interview, but they sluffed it off, saying that there were commercially avialable refrigerators for near absolute zero temperatures "that you just plug in the wall". I didn't research it, but sometimes you just have to take a scientists word for it. Thanks for the concern and comment, I was concerned abvout it too.

traneus
User Rank
Author
power dissipation
traneus   5/4/2015 5:46:28 PM
NO RATINGS
The article states that IBM's quantum computer operates at 0.015 kelvins. Room temperature is about 300 kelvins, which is 20,000 times the operating temperature. An ideal refrigerator would thus dissipate 20,000 times the power the quantum computer dissipates. Real cryogenic refrigerators dissipate even more power.

microjunkie
User Rank
Author
Re: Confused
microjunkie   5/4/2015 12:47:06 AM
NO RATINGS
Thank you Colin, I have gained in knowledge.

Page 1 / 3   >   >>

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed