IBM's new Nighthawk r2 processor claims a 25-fold jump in circuit speed

Started by Inland Aidan, Yesterday at 09:02 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Topic: IBM's new Nighthawk r2 processor claims a 25-fold jump in circuit speed   Views(Read 76 times)
Active members in this topic:
Inland Aidan(1) RuntimeCandle(1)

Inland Aidan

IBM released Nighthawk r2, which it calls its fastest quantum processor to date, capable of executing more than 100,000 circuits per second, a 25 times speedup over IBM's current Heron fleet, which runs roughly 4,000 circuits per second. The processor keeps the same 120 programmable qubits as its r1 predecessor but introduces a new dissipative reset mechanism that replaces the older conditional reset approach, pulling a qubit's effective T1 relaxation time from a median of about 200 microseconds down to roughly 25 nanoseconds and cutting idle time between circuit runs to as little as a single microsecond

That faster reset also reduces initialization error by around 25 times while maintaining Heron class gate fidelity, meaning the speed gains don't come at the cost of computational quality. Nighthawk r2 packs a total of 458 physical quantum elements, its 120 programmable qubits, 218 dedicated couplers, and 120 independent reset elements, making it the most complex quantum processor IBM has put into production. The chip has already demonstrated accurate observable estimation on circuits containing more than 7,500 gates, hitting a key 2026 milestone on IBM's public roadmap, and researchers used it to run doped Clifford sampling quantum advantage experiments alongside a 12 times speedup on neutron scattering simulations that could be directly compared against real laboratory data in about 60 seconds

IBM specifically designed Nighthawk r2's reset capability to work during circuit execution as well as between circuits, making it well suited for dynamic circuits and quantum error correction research that requires repeated mid-circuit qubit measurements. The processor's square lattice architecture connects most qubits to four nearest neighbors rather than the two or three typical of earlier designs, enabling more complex circuit connectivity. Curious what people think matters more for genuinely useful quantum computing progress right now, raw qubit counts, or this kind of speed and throughput improvement that lets researchers actually run far more experiments within the same allocation of time


I read every reply. Even the bad ones.

RuntimeCandle

25 nanoseconds down from 200 microseconds for reset time is such a dramatic engineering jump that it's honestly a little hard to intuitively grasp just how much faster that actually is in practice

Save money on everyday spending Free cashback on thousands of retailers
View offer