It’s time to begin constructing tomorrow’s hybrid quantum computer systems.
The motivation is compelling, the trail is obvious and key parts for the job can be found at this time.
Quantum computing has the potential to bust by a few of at this time’s hardest challenges, advancing every thing from drug discovery to climate forecasting. In brief, quantum computing will play an enormous position in HPC’s future.
Today’s Quantum Simulations
Creating that future gained’t be straightforward, however the instruments to get began are right here.
Taking the primary steps ahead, at this time’s supercomputers are simulating quantum computing jobs at scale and efficiency ranges past the attain of at this time’s comparatively small, error-prone quantum methods.
Dozens of quantum organizations are already utilizing the NVIDIA cuQuantum software program growth package to speed up their quantum circuit simulations on GPUs.
Most not too long ago, AWS introduced the supply of cuQuantum in its Braket service. It additionally demonstrated on Braket how cuQuantum can present as much as a 900x speedup on quantum machine studying workloads.
And cuQuantum now permits accelerated computing on the foremost quantum software program frameworks, together with Google’s qsim, IBM’s Qiskit Aer, Xanadu’s PennyLane and Classiq’s Quantum Algorithm Design platform. That means customers of these frameworks can entry GPU acceleration with none extra coding.
Quantum-Powered Drug Discovery
Today, Menten AI joins corporations utilizing cuQuantum to help its quantum work.
The Bay Area drug-discovery startup will use cuQuantum’s tensor community library to simulate protein interactions and optimize new drug molecules. It goals to harness the potential of quantum computing to hurry up drug design, a subject that, like chemistry itself, is regarded as among the many first to learn from quantum acceleration.
Specifically, Menten AI is creating a collection of quantum computing algorithms together with quantum machine studying to interrupt by computationally demanding issues in therapeutic design.
“While quantum computing {hardware} able to working these algorithms remains to be being developed, classical computing instruments like NVIDIA cuQuantum are essential for advancing quantum algorithm growth,” mentioned Alexey Galda, a principal scientist at Menten AI.
Forging a Quantum Link
As quantum methods evolve, the following massive leap is a transfer to hybrid methods: quantum and classical computer systems that work collectively. Researchers share a imaginative and prescient of systems-level quantum processors, or QPUs, that act as a brand new and highly effective class of accelerators.
So, one of many greatest jobs forward is bridging classical and quantum methods into hybrid quantum computer systems. This work has two main parts.
First, we’d like a quick, low-latency connection between GPUs and QPUs. That will let hybrid methods use GPUs for classical jobs the place they excel, like circuit optimization, calibration and error correction.
GPUs can pace the execution time of those steps and slash communication latency between classical and quantum computer systems, the principle bottlenecks for at this time’s hybrid quantum jobs.
Second, the trade wants a unified programming mannequin with instruments which are environment friendly and simple to make use of. Our expertise in HPC and AI has taught us and our customers the worth of a stable software program stack.
Right Tools for the Job
To program QPUs at this time, researchers are pressured to make use of the quantum equal of low-level meeting code, one thing outdoors of the attain of scientists who aren’t consultants in quantum computing. In addition, builders lack a unified programming mannequin and compiler toolchain that might allow them to run their work on any QPU.
This wants to vary, and it’ll. In a March weblog, we mentioned a few of our preliminary work towards a greater programming mannequin.
To effectively discover methods quantum computer systems can speed up their work, scientists want to simply port elements of their HPC apps first to a simulated QPU, then to an actual one. That requires a compiler enabling them to work at excessive efficiency ranges and in acquainted methods.
With the mix of GPU-accelerated simulation instruments and a programming mannequin and compiler toolchain to tie all of it collectively, HPC researchers will probably be empowered to begin constructing tomorrow’s hybrid quantum information facilities.
How to Get Started
For some, quantum computing could sound like science fiction, a future many years away. The truth is, yearly researchers are constructing extra and bigger quantum methods.
NVIDIA is totally engaged on this work and we invite you to hitch us in constructing tomorrow’s hybrid quantum methods at this time.
To be taught extra, you may watch a GTC session and attend an ISC tutorial on the subject. For a deep dive into what you are able to do with GPUs at this time, examine our State Vector and Tensor Network libraries.
https://blogs.nvidia.com/blog/2022/05/30/quantum-computing-hpc-isc2022/
