We’re tracking fairly well to our earlier projections in most aspects of the project. Our first cut at a 128 qubit processor (Rainier) is going to glass within the next 2 weeks. I’m going to be talking a little bit about it in the blog over the weeks to come. We are very excited about it as you might imagine. This will be the first version of our designs with on-chip SFQ control. The design is scalable (as hopefully I will be able to describe shortly) so pushing to much larger numbers of qubits should be straightforward subject to fab yield and areal density issues. As far as timing, the next hard results you’ll see from us will be entanglement-related results obtained on the 28-qubit Leda processors, then similar experimental results on the 128-qubit Rainier processors. If the first rev works you’ll see data from Rainier by February 2009. If it doesn’t we have another 128-qubit rev scheduled (Tantalus) with data published by May 2009.
Dwave also hopes to address performance comparisons to classical computers:
We’ve been working on a way to quantify the performance of adiabatic quantum algorithms on this particular 128-qubit chip based on using a quantum monte carlo technique to extract the minimum gap and matrix element between ground & first excited states, pretty much identical to the work described in http://arxiv.org/abs/0803.3971 . We are very close to being able to compare the projected performance of this chip & associated systems apples to apples with classical approaches to solving problems of interest.
This is good news. A bit behind the original roadmap over the next few months but once a solid scalable architecture is proved with entanglement and performance advantage then being able to fill out a mostly empty the die with a lot more working qubits should have quite a large number of qubits. More importantly proving the performance advantage will enable the new solutions and provide a profitable business. Proving entanglement in a peer reviewed paper should prove that this is a quantum computing system.
If the 128 qubit system proves out and the entanglement proof is good, and Dwave gets some paying customers for the computers or their online service that would likely satisfy the long now prediction that I made that there would be a 100 qubit computer.
There will be a quantum computer with over 100 qubits of processing capability sold either as a hardware system or whose use is made available as a commercial service by Dec 31, 2010
Previous articles had reviewed Dwave's evidence of quantum coherence.
This site has written 125+ articles on quantum computers.
7+ articles on adiabatic quantum computers
Pretty went over every Dwave publication and Seth Lloyds original adiabatic quantum computer papers and papers on what if they only get quantum annealing.
Dwave has several scientific papers on the Dwave site, which were discussed in articles written on the Dwave CTO blog and in the previously mentioned articles on this site.
Their strategy is to see if they can make "good enough" qubits and make the kind of quantum computer that they can make now (which is adiabatic) and which should allow for optimization type algorithms to be run. Then make them achieve superior performance for problems of business interest. Use money or funding to iterate make better qubits and more of them.
At the Demonstrations
There was work was done by the device which is their quantum chip. They also used a regular computer to use a regular algorithm to chunk up a larger Soduku puzzle to fit the smaller number of qubits that they had.
However, the device might not have been in a quantum state. If a quantum computing device (in particular the Dwave one) goes out of being quantum then it can still get the right answer via classical annealing.
So there was no scamming-switcharoo going one but it could not be stated clearly how much of the work was being done with a quantum computer in quantum conditions if any. The thing could have solved it in classical mode and you could not tell.
That is where the papers with evidence of some level of quantum coherence and
the expected paper on entanglement come in. Plus the biggest factor which is when
128 qubits or thousands of qubits solve useful problems a lot faster than any classical computer.
That has always been the Dwave position. This thing will only be valuable if it solves real problems a lot faster than classical computers or problems that are unsolvable by classic computers.