Pages

February 13, 2007

The Dwave Quantum computer demo event, part 1

Today Dwave demoed their 16 qubit quantum computer in public I went to the event at the Mountain View Computer museum of history. There were a few hundred people in attendance.

There were some slides that discussed markets that Dwave systems had identified. Operations Research $0.5 billion
Bioscience $1.2 billion
Search
Chemistry
CAD/CAE (place and route) $1.3 billion
Finance & Economics
Security
Academia

More was revealed about the timeline that Dwave has set for itself.
32 qubits by Q4 2007 with an improved I/O system
512 qubits for Q1 2008
1024 qubits for Q3 2008

The problem solution times for the system are currently dominated by setup times.
They have to configure and load the parameters into the system.
But they could run many problem runs per second.
So if the actual problem run times were microseconds then even with setup taking 100 times longer they were still able run multiple problems multiple times.
To avoid the issues of error correction, they take a probabilistic approach and run the same problem say 100 times. The most common answer is chosen.
They are able to run problems that are larger than the 16 qubits by breaking up some problems and running local solutions and then putting them together. Like a 9 X 9 sodoku problem.

There is still some question of how much quantum computerness the system is exploiting. They have run experiments to confirm this but the real answer will come at the end of this year or early next year when the larger systems either do or do not demonstrate the speed advantage.

If they double the qubits every year after 2008 then they will get to 1 million qubits by 2018. They will be iteratively improving the quality of the qubits and system over time.

0 comments: