In 2029, the artificial intelligence supercomputer Skynet suddenly awakened and developed self-awareness. The Skynet system determined that the humans who invented the supercomputer would pose a threat to AI. Therefore, it sent back a T-800 Terminator robot, played by Arnold Schwarzenegger, to the past to eliminate John Connor, the leader of the future human resistance. This is the plot of the movie “Terminator”.
Interestingly, Google’s AI quantum supercomputer also has a roadmap to build an AI super quantum computer within five years, which will also reach 2029. Currently, it is in between the third and fourth stages of development. The current stage mainly focuses on correcting quantum computing errors. At this time, the power of Nvidia GPUs has accelerated the evolution of AI super quantum computers. It can be said that the prototype of the “Skynet” in human society has taken shape.
Nvidia recently announced a collaboration with Google Quantum AI to accelerate quantum computer calculations using the Nvidia CUDA-QTM simulator. Nvidia has been working with Google to develop Quantum Process Units (QPUs) from CPUs to GPUs, with the goal of reducing errors and optimizing AI system upgrades. With supercomputing simulation, the development of supercomputers will not be as depicted in science fiction films, where they mistakenly perceive humans as a threat to AI and issue instructions to exterminate humans. This collaboration can be said to be the most important milestone in the development of human technological civilization in the next five years.
What is Quantum Computing?
Quantum computing utilizes quantum physics to solve mathematical problems that cannot be solved on traditional supercomputers. The core of quantum computing is the quantum bit, or qubit. Classical bits exist in either a 0 or 1 state, while qubits can exist in a superposition of these two states.
N qubits in superposition can represent 2^N binary configurations. These configurations collectively form a quantum state. When any operation is performed on N qubits, the entire quantum state is controlled, indicating the presence of tremendous parallelism. However, the utilization of this computational power has subtle differences, as information read from the quantum state can only be probabilistically measured through computation of a single configuration. To effectively harness quantum parallelism, the application of quantum computing requires the utilization of quantum entanglement and quantum interference.
How Nvidia CUDA-QTM Accelerates Google AI Super Quantum Computer Calculations
Nvidia has introduced the NVIDIA CUDA-Q hybrid quantum-classical computing platform, which enables quantum computers to work together with high-performance traditional computing. GPUs, originally designed purely for graphics, have transformed into essential hardware for high-performance computing (HPC). Nvidia provides CUDA-QTM, allowing all QPU researchers and developers to perform GPU-accelerated quantum dynamical simulations, accelerating the design of next-generation quantum computing devices.
Traditionally, simulating computations incurs high costs. By using CUDA-Q, Google can execute the largest and fastest quantum device physics dynamic simulations at a significantly lower cost, utilizing 1024 Nvidia H100 Tensor Core GPUs. Through CUDA-Q and H100 GPUs, Google can conduct comprehensive and realistic simulations of devices with 40 qubits. The software supporting these accelerated dynamic simulations will be publicly available on the CUDA-Q platform, enabling quantum hardware engineers to rapidly scale system design.