Combining Nvidia RAPIDS Technology with D-Wave Annealer
Does anyone have any practical experience combining Nvidia hardware, specifically RAPIDS cuDF, with a quantum annealer? My application would benefit from both technologies running in parallel, but I have no idea how to combine them practically.
I am looking at Nvidia hardware to speed up the digital signal processing and D-Wave for the AI symbolic logic constraint logic. They run asynchronously in parallel.
Comments
Hi Daro,
I don't have any experience combining RAPIDS cuDF with the QPU, but it sounds like the d-wave hybrid tool may be helpful here.
Can you give any more information about your problem?
A quantum annealer is an expensive piece of hardware. I am thinking of keeping it constantly fed with data by using Nvidia technology to preprocess data. RAPIDS cuDF is a simple tool to manipulate data in ways that a quantum annealer cannot. Does this make sense to you? Thanks.
Hello Daro, to add to Alex's comment, above, I would suggest signing up for the Leap offer (http://cloud.dwavesys.com/leap) and writing a small program to use the QPU. After that is done, you could feed data to it using your RAPIDS cuDF setup, on a very small scale. Like Alex, I'm not aware of anyone doing this thus far.
Hi Daro,
One kind of input where our hybrid solvers can be useful is clustering dissimilarity (distance) data. For high dimensional data, computing all of the pairwise distances could be expensive. I could see GPU-accelerated cuDF calculations would be useful in place of Pandas.
Best,
Marcus
Please sign in to leave a comment.