Stories
Slash Boxes
Comments

SoylentNews is people

posted by spiraldancing on Tuesday February 11 2020, @03:35PM   Printer-friendly
from the Tyranny-of-Moore dept.

A team of researchers affiliated with several institutions in China created a photonic computer that was able to solve the subset sum problem. In their paper published in the journal Science Advances, the group describes their computer and how well it performed.

[...] In this new effort, the researchers propose the idea of a photonic computer by creating one that can solve the subset sum problem.

The subset sum problem [...] is easy for a conventional computer when the list is small—but when it grows large, it becomes unworkable.

To solve the problem using a photonic computer, the researchers mapped it into a 3-D waveguide network etched onto glass using a femtosecond laser. Photons were then allowed to dissipate into the network in search of a solution in parallel. This allowed the researchers to try different combinations at the same time rather than grinding through them all, as is done with a conventional computer. Not only did the approach work, it was able to do so faster than a supercomputer—and it demonstrated that photonic computers are capable of solving such problems and are scalable, as well.

More information: Xiao-Yun Xu et al. A scalable photonic computer solving the subset sum problem, Science Advances (2020). DOI: 10.1126/sciadv.aay5853


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 2) by Runaway1956 on Wednesday February 12 2020, @06:03PM

    by Runaway1956 (2926) Subscriber Badge on Wednesday February 12 2020, @06:03PM (#957299) Journal

    Yes. And, I wonder how many people understand the scale of GPU computing.

    For comparison sake,

    dual hexacore socket F Opterons (12 cores total) rack up about 15,000* points per day on Folding at Home.

    A single GeForce GT 730 gets about 12,000 PPD

    quad octocore Opterons (32 cores total) get about 45,000 PPD

    A single GeForce GTX 1650 gets about 300,000 PPD

    Those are components of my existing systems. I'm looking at a GeForce RTX 2060 or 2070, and thinking hard about replacing that GT 730. Either of them should roughly quadruple the performance of the 1650's. I would like to get a 2080, but the cost is prohibitive.

    Anyway, you can see that the earliest of CUDA enabled consumer GPU's can do more math faster than a whole boatload of CPU's.

    * All of those PPD numbers vary day-to-day, and even hour-to-hour. Those are the round numbers showing currently in my fahcontrol screen.

    Starting Score:    1  point
    Karma-Bonus Modifier   +1  

    Total Score:   2