Summary
AMD, IBM and Oak Ridge National Laboratory have established a shared platform to study how quantum processors and classical computing systems can work together. Its first deployment colocates IBM quantum computers, including the 120-qubit IBM Quantum Nighthawk, with ORNL’s Umbra supercomputer.
AMD, IBM and Oak Ridge National Laboratory (ORNL) have established a shared codesign platform to evaluate how quantum computers can work alongside classical computing systems. The first deployment colocates IBM quantum computers—including the 120-qubit IBM Quantum Nighthawk—with ORNL’s Umbra high-performance computing system, which is powered by AMD EPYC CPUs and AMD Instinct GPUs.
The platform is intended for research into the hardware architectures, software and workflows needed by future hybrid systems. Its immediate role is to give the three partners a shared environment for studying those design requirements, rather than to announce a completed quantum error-correction result.
A shared environment for system design
Quantum processors and classical computers handle different parts of a hybrid workload. The announcement describes classical systems as supporting precise control and rapid feedback, parallel scientific calculations, and the coordination of work across computing resources. Error correction is another central challenge: classical processing must help identify and respond to errors that arise during quantum computation.
The partners contribute different capabilities. IBM brings quantum computing technology; ORNL contributes scientific research and its computing environment; AMD brings CPUs, GPUs and software intended to connect quantum processing with scientific workloads. The initial deployment brings IBM’s quantum computers and the Umbra system together in a colocated environment.
AMD also describes a wider set of technologies that could address different parts of a hybrid system. It assigns orchestration and general-purpose computing roles to EPYC CPUs, simulation and parallel computation to Instinct GPUs, and low-latency control, readout and decoding to Versal adaptive SoCs. AMD’s ROCm software ecosystem is presented as a way to connect applications with this hardware. The announcement does not say that every technology in this broader portfolio is part of the initial deployment.
Research toward larger hybrid systems
The collaboration builds on earlier work between AMD and IBM involving GPU-accelerated scientific calculations and FPGA-based error decoding, according to AMD. The partners say they will use the platform to investigate requirements for integrating quantum processors with classical computing and scientific workloads.
Plans call for connecting the platform with the Oak Ridge Leadership Computing Facility. AMD says the expanded research environment is intended to support work combining quantum, AI and scientific computing. The collaboration is also described as supporting the US Department of Energy’s Quantum Genesis Mission and its effort toward a scientifically relevant quantum computer by 2028.
The practical significance is in testing the classical side of quantum computing as part of the overall system design. Control, computation and coordination need to work alongside a quantum processor; the platform gives the partners a setting in which to study those interactions and the requirements of future, more integrated systems.