UK Consortium Releases Specification for a Hardware Abstraction Layer (HAL) for Quantum Computers

One of the things that helped accelerate the usage of classical computers was the standardization of the classical computer languages Fortran (1966) and Cobol (1968). This allowed a programmer to develop a program and then run it on different machines with minimal changes. We are starting to see similar things happen in the quantum world. The latest is the release from a UK consortium led by Riverlane and the UK's National Physical Laboratory (NPL) of a Hardware Abstraction Layer (HAL) that is designed to be portable across four leading qubit technologies including superconducting qubits, trapped-ion qubit, photonic systems and silicon-based qubits. This project has received in funding of £7.6 million ($10.4M USD) from the UK government's Industrial Challenge Strategy Fund and includes other members of the consortium SeeQC, Hitachi Europe, Universal Quantum, Duality Quantum Photonics, Oxford Ionics, and Oxford Quantum Circuits, as well as UK-based chip designer, ARM. This specification will allow quantum programmers to create portable programs that can be used across different quantum technologies. The first version 0.1.3 is now available on GitHub here and a news release provided by Riverland and NPL announcing this release can be found here.

There have been a number of similar efforts started up recently that also try to encourage development of tools and specifications to enable program portability. IBM has created a specification called OpenQASM3 and has announced a technical steering committee consisting of itself, Microsoft, and Amazon to guide evolution of the language. Microsoft has announced a Quantum Intermediate Representation (QIR) that is based upon the popular open-source LLVM intermediate language used in classical computing. Member of the the Quantum Economic Development Consortium (QED-C) have been holding workshops a topic called PIRQ (Practical Intermediate Representation for Quantum). A workshop will be held at IEEE Quantum Week in October on Quantum Intermediate Representations. And Zapata Computing has developed an internal language called ZQuantum that allows them to translate programs back and forth between ZQuantum and Qiskit, Cirq, Pyquil, Pennylane and others.

All of these efforts are still very early and time will tell how well they work out. One of the questions is whether a translated program will perform as well as one which has been originally written in the language specifically developed for the target hardware. The optimization challenges can be very significant and are compounded when one tries to support machines which can be so much different in their architectures. So although it may take a while for these efforts to come to fruition and provide large benefits, we are hopeful that it will have the same positive affect as the standardization of Fortran, Cobol, and other languages had in classical computing.

September 25, 2021