Graham Carlow Photography
IBM laid out its quantum software roadmap and ambitions to build an ecosystem that includes collaboration with open source developers.
Working with a mix of independent open source software developers, academic institutions and its own internal development teams, IBM hopes the new ecosystem can speed development of production-level software for quantum computers and accelerate the integration of quantum and classical systems. This effort could democratize quantum technology by making it available to nontechnical corporate users.
IBM vice president of quantum computing Jay Gambetta wrote in a blog post that open source software development is driven by the understanding that "an ecosystem of different human needs drives the best outcomes" and said IBM's approach to quantum computing is no different.
IBM will segment developers in the ecosystem on three levels:
- At the lowest level is quantum kernel developers focused on creating high-performance quantum circuits with timing and pulse-level controls.
- The next level up are quantum algorithm developers who rely on these high-performance circuits to develop quantum algorithms that provide an advantage over present-day classical computing offerings.
- The third segment includes quantum model developers that will apply these algorithms to use cases to create quantum models for multiple disciplines, including chemistry, physics, biology, machine learning and finance.
"We need software developers to prioritize themselves relative to the kind of work they want to do," said Bob Sutor, vice president of IBM Quantum Ecosystem Development group. "Do they want to work less directly to the hardware, do they want to work at the algorithmic level or do they want to work up closer to the actual use cases, or what we call model developers?"
As part of the roadmap rollout, IBM -- made "crucial" updates for quantum kernel developers who write at the lowest level. The company will release Qiskit runtime, an execution environment designed to increase the capacity to run more circuits at a faster rate. The offering also has the ability to store quantum programs, allowing users to run them as a service.
Qiskit runtime "rethinks" the classical-quantum workload so programs can be uploaded and executed on classical hardware that is located directly beside a quantum computer. This reduces latencies that stem from the typical communications between a user's classical system and a quantum processor. IBM said these enhancements will speed workloads by 100 times. This in turn enables quantum systems to run jobs in a few hours instead of a few months.
"The Qiskit runtime is important because it starts the process of selectively moving certain types of local classical computing output over to quantum environments," Sutor said. "This means you will experience less latency and you can use quantum resources more efficiently than you could before."
Some analysts believe the ecosystem could finally result in production-level offerings for large corporations.
"This feels like the first step toward thinking about how [quantum technologies] can make it into a production environment," said Bob Sorensen, senior vice president of research and chief analyst for quantum computing at Hyperion Research LLC. "It's a point where can we start to optimize the weak links in the chain and realize the things you can control to make a production environment more useful and higher performing."
Segmenting the open source ecosystem into three levels not only gives focus to software companies part of the ecosystem, it also gives focus to corporate IT pros in putting together their own internal quantum development projects, according to another analyst.
"By segmenting the population into model, algorithm and kernel developers, IBM is providing a structure for enterprises as to what activities these various roles should perform," said Chirag Dekate, vice president of research at Gartner. "More importantly, it gives them a way to go about matching their developers with the right skill sets which will accelerate their quantum journeys."
Chirag DekateVice president of research, Gartner
IBM laid out the first part of its quantum roadmap last September when it announced the delivery schedule for its quantum server hardware through 2023. The company promised to deliver four quantum systems culminating in the release of its Condor processor, which will contain 1,121 qubits. IBM said Condor will enable developers to jump over a number of technical hurdles that have prevented them from solving a wide range of complex scientific and business problems.
A key piece of IBM's software strategy is to continue to use and create open source tools, eventually converting some of them into cloud-native components, the company said. This reportedly will allow the company to continue scaling and extending its quantum software so IT pros can take advantage of its evolving quantum architecture while running quantum programs more reliably.
By 2023, IBM expects to have available multiple families of prebuilt runtimes that can be summoned from cloud-based APIs using a variety of common development frameworks, according Gambetta. Once the prebuilt runtimes are in place, foundations laid down by quantum kernel and algorithm developers will allow model and enterprise developers to explore quantum computing models without having to think about the quantum physics, Gambetta wrote.
According to IBM's roadmap, by 2025 the goal of frictionless quantum computing will be accomplished. This will mean hardware is no longer a concern to developers or users across any quantum computing stack. Developers can then depend on not just quantum systems but also on classical high-performance computing stacks using cloud-based APIs to create applications and services.