IBM pledges to push boundaries in quantum computer roadmap
IBM revised its quantum computing roadmap with a 4,000-qubit system by 2025 and unveiled its Modular Quantum Computing initiative to support multiprocessor systems.
IBM updated its roadmap for a number of key quantum technologies, including a 4,000-qubit system by 2025, an improved intelligent software orchestration layer to distribute workloads and a version of its Qiskit Runtime software that allows developers to access a broader array of functions in quantum computer systems.
Working in concert, the upcoming technologies are designed to make quantum computing more of a practical reality for users and developers, according to IBM, while also laying down the technical foundation for quantum-centric supercomputers to come over the next several years. The newly added technologies, to be unveiled at this week's IBM Think conference, play a central role in IBM's Modular Quantum Computing initiative, a blueprint for building more powerful and scalable systems.
Modular quantum computing
The new computing model is anchored by three different approaches to scaling. The first combines modular quantum processors with classical infrastructure, both to be controlled by Qiskit Runtime software. This sets up developers to build quantum calculations into their workflows.
Second, the company will use chip-level couplers connecting multiple chips, residing either inside one server or spread across multiple servers, that make multiple chips work in unison to solve a single problem. The third approach involves building quantum communication links among quantum processors. IBM will outline its proposal to deploy such links to connect clusters of systems together during Think.
"The connections tying the chips together are fast enough to enable the [entangled] gates between them to the point where users can't tell if they are dealing with one system or three," said Katie Pizzolato, director of IBM quantum strategy and applications research. "The infrastructure is going to be abstracted from the process to increase speed."
While IBM will showcase the power and speed of its upcoming hardware, Pizzolato said that starting this week and going forward, IBM will be turning its focus more toward tighter integration between the software and hardware.
IBM's quantum computing development roadmap through 2026 includes plans to scale quantum applications.
Multiprocessor quantum systems
One method of bringing hardware and software closer together -- making programming for future quantum systems easier -- uses primitives built into the Qiskit Runtime software. Primitives can encapsulate common hardware queries into algorithms that can then be abstracted by programmers and used for a variety of tasks, including application development.
"We are going to continue to extend these primitives with new capabilities," Pizzolato said. "The top-line message here is, we will push the boundaries of what we can do with existing systems using these techniques to make quantum computing more practical."
Primitives will also play a role in IBM's ability to deliver quantum serverless capabilities into its core software stack by 2023, the company said. This too is designed to make it easier for developers to access both quantum and classical resources.
If you can run a program that solves a problem in 15 minutes instead of five hours, that's a big deal.
Doug FinkeFounder, Quantum Computing Report
IBM's use of multiprocessors in quantum computers is a natural evolution, similar to what happened in the classical systems world decades ago.
"In the days of the IBM 360 mainframes, people only knew monolithic, single-processor systems, but then people built data centers with multiple processors," said Doug Finke, founder of Quantum Computing Report. "IBM is taking its first steps to do the same in the quantum industry. It's going to become quite common in the quantum world."
Given the time it takes classical and even some existing quantum computers to come up with answers to extremely complex problems, the added speed a modular quantum system could provide would be welcomed by many IT shops and researchers.
"Users really do care about how long it takes a program to run," Finke said. "If you can run a program that solves a problem in 15 minutes instead of five hours, that's a big deal."
Given that the Qiskit Runtime software's most important duty is to make classical and quantum computers work together more efficiently, the software plays a critical role in providing additional processing speed.
The new additions to the roadmap will exploit the capabilities of IBM's Quantum System Two, and also add modularity and more flexibility into each layer of that system's stack, as well as the ability to link multiple quantum systems.
As Editor at Large with TechTarget's News Group, Ed Scannell is responsible for writing and reporting breaking news, news analysis and features focused on technology issues and trends affecting corporate IT professionals.
