D-Wave's Upgraded Stride Solver Beats Its Old 120-Second Results In 15 Seconds

The company also launched a three-course developer training path, with benchmark gains on routing and assembly-line problems.

Share
D-Wave's Upgraded Stride Solver Beats Its Old 120-Second Results In 15 Seconds

D-Wave Quantum, which builds quantum computers and the cloud software to run them, said Wednesday that an upgraded Stride hybrid solver found better solutions in 15 seconds than the previous version found in 120 seconds in key benchmark tests. The update is live in its Leap cloud service, and the company paired it with a new three-course developer training path, according to an 8-K.

Benchmark gains on routing problems

On traveling salesperson problems, which ask for the shortest route through a set of locations, the company said the enhanced solver averaged results within 0.02% of the best possible in 15 seconds on problems with up to 10,000 locations. That was, on average, more than 200x closer to the best possible result than the previous solver managed at the same runtime.

On larger problems with more than 10,000 locations, solutions averaged within 3.8% of the best possible result in 120 seconds, more than 500x closer than the prior version at the same runtime.

On simple assembly line balancing problems, the new version found better solutions in 15 seconds than the old one did in 120. The company defines best possible results as reference solutions proven optimal by another solver or algorithm, and says the improvement factors measure closeness to the optimum, not processing speed.

How Stride fits D-Wave's business

D-Wave sells access to its quantum machines through Leap, which is available in 42 countries, and also sells systems for customers' own premises. Stride combines quantum and classical computing resources to tackle large optimization problems that can be hard for classical computing alone. The company said it enhanced the solver's core algorithm to extend performance across the problem types it already supports.

Alex Condello, D-Wave's senior vice president of algorithms, performance and tools, said "Developers need proven technology that finds better answers, faster."

A three-course learning path

The new Applied Quantum Optimization path teaches developers to identify promising use cases, formulate mathematical models and build applications on Stride. It is designed for learners with varying levels of experience, and select courses include access to Leap and office hours. The courses are available now through Leap.

The company said the benchmarks show the kind of improvement it expects across a wide range of optimization problems.

Sources

More on D-Wave Quantum

All D-Wave Quantum (QBTS) coverage →