“Quantum Computing Advancements”
Photo by Scott Hamilton A 1/2 scale model of Fujitsu’s 256-qubit quantum processor on display at SuperComputing25 in St. Louis.
By Scott Hamilton
Last week was the annual SuperComputing conference and this year it was held in St. Louis, Mo., at the America’s Center Conference Center. I went with a group of home-school students from the Licking, Salem and Houston area. I figured over the next couple of weeks I would write about the things these students found to be the most interesting at the conference, as well as some of the things I found exciting about where the industry is going. The big topic was quantum computing, and several of the students realized that it involves way more than just computer engineering to build an effective quantum computer.
I would say the most impressive thing about quantum computing we saw at the convention was Fujitsu’s new 256-qubit quantum computer (well, a scale model of it, at least). You see, their quantum computer is really mostly a giant refrigeration unit, which one student noticed right away, as he is working part time in heating and air-conditioning. Like most quantum processors, the one currently being tested by Fujitsu requires extremely low operating temperatures nearing absolute zero. This particular one is designed to operate at 1 degree Kelvin, which is -457.87 degrees Fahrenheit. To achieve such low temperatures requires multiple stages of cooling, each accomplished by compressing refrigerant into smaller and smaller tubes; each smaller size results in higher pressures and lower temperatures.
You can see in the image a 1/2 scale model of the Fujitsu quantum processor. The model stood approximately 20 feet high. If you look closely at the image, the processor itself is housed in the little gold cube in the center of the bottom of the image. Everything else in the image is the cooling system. Each round gold disk is another layer of cooling. The gold cube is held at 1 degree Kelvin, the bottom gold disk is 17 degrees Kelvin, each disk doubles the temperature to the top disk which is held at 273 degrees Kelvin.
The technology involved in dropping temperatures to this level is really no different than how your household refrigerator works. It utilizes a coolant that has extreme temperature drops as the pressure of the coolant increases. If you pump the coolant into smaller and smaller pipes, the pressure increases and the temperature decreases. You can clearly see in the photo that the pipes between the layers get progressively smaller, resulting in progressively higher pressures and lower temperatures.
Last year the big deal was their announcement of their 64-qubit quantum computer, which was launched in October 2023. It might not sound like a big accomplishment to make a computer just four times more powerful, at least not until you understand the real math behind quantum computers. Quantum computers work off of the probability that a particular set of qubits will be a particular value based on the wave function they are assigned to optimize. If you try to do the same probability calculation in a classic computer, it doubles the amount of memory required to store the full probability state for each additional qubit. The most we can simulate with any kind of stability today is 41 qubits, and that is accomplished with a fairly large supercomputer, with multiple processors and nearly 32 terabytes of memory. For simulating, the quantum processor from two years ago would require 132 petabytes of memory. Let’s just say reaching 256 qubits is an impossible task for a classic computer.
Fujitsu would not share the budget of their project, but I expect it is in the hundreds of millions of dollars, and when I asked the head researcher how close to an actual useful quantum computer we were, she said we are at least two decades away. In order to run a truly useful quantum algorithm, we need to reach somewhere near 10,000 qubits and they have plans to double the capacity every two to five years. It really made me begin to wonder if there is a better way to model these systems, making it easier to design a more robust system with a faster time to market for a useful system.
Next week we will talk about an alternative to quantum computing, which is also based on bit probability, but uses more conventional hardware, operates at room temperature, and seems at first glance to accomplish nearly the same advantages as the current quantum computing technologies. It is a U.S.-based project out of the University of Kentucky. Stay tuned for more information on their project. Until next week, stay safe and learn something new.
Scott Hamilton is an Expert in Emerging Technologies at ATOS and can be reached with questions and comments via email to shamilton@techshepherd.org or through his website at https://www.techshepherd.org.
