Quantum Day Will Show Quantum Computing’s Future. A intriguing topic of computer science is quantum computing, which promises quicker computation than currently possible. To enhance quantum computing simulations, NVIDIA presents the cuQuantum SDK, which provides researchers and developers with unmatched speed and efficiency.
The technique is expected to solve many previously unsolvable or unfeasible problems. Quantum computing has great promise for advancements in everything from financial forecasts to medicine research and materials creation.
However, the current advances in quantum hardware, error correction, and algorithms are equally as intriguing as the potential of quantum computing.
By announcing its first Quantum Day at GTC 2025 on Thursday, March 20, NVIDIA is commemorating and investigating this incredible advancement in quantum computing. Leading experts come together in this new focus area to provide a thorough and fair assessment of what companies may anticipate from quantum computing over the next few decades, paving the way for practical quantum applications.
NVIDIA cuQuantum SDK
Although they are still in their infancy, quantum computers are having an impact on a new generation of simulations that are now being run on conventional computers and are now being accelerated by the NVIDIA cuQuantum SDK.
What is Quantum computing?
Using the physics governing subatomic particles to substitute the more basic transistors found in modern computers, quantum computing is a complex method of doing parallel operations.
Unlike regular computers, which use bits that are either on or off, one or zero, quantum computers use qubits, which are computing units that can be on, off, or any value in between. Quantum computers are superior for some types of math because of the qubit’s capacity to exist in the in-between state known as superposition, which adds a potent capability to the computing equation.
What Does a Quantum Computer Do?
Quantum computers might perform calculations that would take classical computers a very long time, if they could complete them at all, by using qubits.
For instance, any integer between 0 and 255 is represented by eight bits in modern computers. A quantum computer may employ eight qubits to represent every integer between 0 and 255 simultaneously because of properties like superposition.
Similar to parallelism in computing, it offers enormous speedups by computing all options simultaneously rather than sequentially.
Therefore, a quantum computer can determine the answer in a single step, but a classical computer must factor a huge number by stepping through long division processes one at a time. Boom!
This suggests that entire industries that rely on factoring what are currently impossibly big numbers, like encryption, could be completely transformed by quantum computers.
A Big Role for Tiny Simulations
That might be the beginning. According to some experts, quantum computers will overcome the current barriers that prevent simulations in materials science, chemistry, and any other field involving worlds constructed using the nanoscale building blocks of quantum mechanics.
By assisting engineers in developing more accurate simulations of the quantum effects they are beginning to detect in today’s smallest transistors, quantum computers may even help semiconductors last longer.
In fact, according to scientists, quantum computers will eventually enhance classical computers rather than replace them. And according to some predictions, quantum computers will be employed as accelerators in the same way that GPUs speed up modern computers.
What Is the Process of Quantum Computing?
You shouldn’t expect to construct your own quantum computer using parts you found in the local electronics store’s bargain bins.
The few systems that are still in use today usually need refrigeration that produces temperatures slightly over absolute zero. To manage the delicate quantum states that drive these systems, they require that computing arctic.
To illustrate the difficulty of building a quantum computer, one prototype creates a qubit by suspending an atom between two lasers. In your own workshop, give it a try!
Entanglement is the result of quantum computing using nano-Herculean muscles. This occurs when two or more qubits are in a single quantum state, which can occasionally be detected by millimeter-wide electromagnetic waves.
If you turn up the energy a bit too much, you will lose superposition, entanglement, or both. Decoherence, the quantum computing counterpart of the blue screen of death, is the resultant noisy condition.
What Prospects Does Quantum Computing Have?
Early versions of quantum computers are currently in use by a small number of businesses, including Alibaba, Google, Honeywell, IBM, IonQ, and Xanadu.
They now offer tens of qubits. However, qubits can occasionally be unreliable due to their noise. Systems require tens of thousands or perhaps hundreds of thousands of qubits to reliably solve real-world issues.
It may take a few decades before humanity reach a high-fidelity era where quantum computers are actually helpful, according to experts. There is a lot of disagreement in the industry over when it will achieve so-called quantum computing supremacy and when quantum computers will be able to perform tasks that classical ones cannot.
Accelerating Quantum Circuit Simulations Today
The good news is that the fields of artificial intelligence and machine learning have brought attention to accelerators like GPUs, which are capable of carrying out a wide range of calculations that quantum computers would use qubits to execute.
Thus, quantum simulations using GPUs are already being hosted by classical computers today. For instance, NVIDIA used an in-house AI supercomputer, Selene, to run a state-of-the-art quantum simulation.
During the GTC keynote, NVIDIA revealed the cuQuantum SDK, which will accelerate GPU-based quantum circuit simulations. According to preliminary research, cuQuantum could provide speedups of orders of magnitude.
The SDK has an agnostic strategy, offering users a selection of tools to suit their methodology. The state vector approach, for instance, yields high-fidelity results, but as the number of qubits increases, so does its memory requirements.
On the biggest classical supercomputers available today, that results in a practical limit of about 50 qubits. However, cuQuantum has shown excellent results (below) in speeding up quantum circuit simulations that employ this technique.
In session E31941 at GTC, Jülich Supercomputing Centre researchers will give a detailed overview of their work using the state vector method (free with registration). Tensor network simulations, a more recent method, use more processing and less memory to do comparable tasks.
This technique was used by NVIDIA and Caltech to speed up a cutting-edge quantum circuit simulator that runs on NVIDIA A100 Tensor Core GPUs using cuQuantum. In 9.3 minutes on Selene, it produced a sample from a full-circuit simulation of the Google Sycamore circuit, a process that experts said would require days and millions of CPU cores eighteen months ago.
That has generated a sample of the Sycamore quantum circuit at depth m=20 in record time less than 10 minutes using the Cotengra/Quimb packages, NVIDIA’s recently announced cuQuantum SDK, and the Selene supercomputer.
This sets the benchmark for quantum circuit simulation performance and will help advance the field of quantum computing by improving the capacity to verify the behavior of quantum circuits.
NVIDIA anticipates that cuQuantum‘s ease of use and performance improvements will establish it as a fundamental component of all quantum computing frameworks and simulators at the forefront of this research.