IBM proposes new quantum computing speed metric: CLOPS

The quantum computing sector seems to be evolving at blinding speed, and that could become a big problem as the industry and potential users try to get a better understanding of the processing speeds emerging quantum computers really are capable of in a practical sense.

The sector already has ways to measure the scale of quantum computers--in qubits, and the quality of quantum computers--in quantum volume. It has lacked a way to accurately measure processing speed, but quantum computing pioneer IBM this week proposed circuit layer operations per second (CLOPS), which indicates how many circuits can run on hardware in a given time.

“As we test these systems, we need to focus on real-world scenarios, and what is actually practical,” said Bob Sutor, chief quantum exponent at IBM during a presentation this week at the Inside Quantum Technology Fall conference in New York City. “Speed tests should not allow people to extrapolate future speeds. The benchmarks we use should not allow gaming of the system.”

The 65-qubit quantum processor that IBM has today is capable of 1,400 CLOPS. For comparison, the 27-qubit system the company had in 2019 operated at 200 CLOPS. The CLOPS figure should go up dramatically as IBM releases its next system at 127 qubits.

CLOPS was described in greater detail in an IBM Research whitepaper that came out late last week. Sutor on Tuesday told Fierce Electronics via email, “No one else has had time to sign on yet, but we are hopeful that the momentum behind Quantum Volume will carry over to CLOPS. Since the implementation of OpenQASM 3 [a language that allows for dynamic blending of quantum operations and classical computations] will be essential to improve performance, we are very pleased with industry cooperation that is happening around quantum computing.”

He added, “I think once people have experience implementing the benchmark in open source code, we can then look at a more formal standardization process.”

In his presentation at IQT fall, Sutor also called for the industry to focus in the coming years on improving quantum computing scale, quality and speed, and not to spend too much of its time celebrating its own newfound popularity.

“Scale, quality and speed are the three metrics we need to be focusing on,” Sutor said. “Go ahead and celebrate the industry, but we need to show how all of these parts fit together, and each of them must be improved. There are speed bumps in all of these areas, and if anyone tells you we’re going to scale smoothly from here to there, just smile.”

RELATED: IBM adopts heavy-hex lattice for quantum computing