A real-time high-throughput low-latency quantum error correction accelerator
ORAL
Abstract
Recent experimental results in quantum error correction (QEC) have demonstrated real-time decoding systems achieving the required throughput to avoid an exponentially increasing backlog of syndrome data to be decoded. While maintaining this throughput, it is desirable to also minimise the full decoding response time; the time taken to send readout data, transform data into syndromes and return corrections from the decoder. This response latency will impact the logical clock-rate of a quantum computer. Current demonstrations of these systems have either been tightly-coupled to particular quantum hardware or have displayed undesirably large latencies.
We present a system that achieves high throughput and low latency decoding for quantum error correction. The work aims to create a system that can be integrated into a variety of qubit control hardware and support several decoders. A key component being the efficient transformation of qubit readout to syndromes in the format necessitated by the target decoder.
We present a system that achieves high throughput and low latency decoding for quantum error correction. The work aims to create a system that can be integrated into a variety of qubit control hardware and support several decoders. A key component being the efficient transformation of qubit readout to syndromes in the format necessitated by the target decoder.
–
Presenters
-
Marco Ghibaudi
Riverlane
Authors
-
Kenton M Barnes
Riverlane, Riverlane Ltd
-
Abbas B Ziad
Riverlane
-
Aniket Datta
Riverlane, Riverlane Ltd
-
Dogancan Davutoglu
Riverlane
-
Andrea Husseiniova
Riverlane, Riverlane Ltd
-
Samin Ishtiaq
Riverlane, Riverlane Ltd
-
Kauser Johar
Riverlane, Riverlane Ltd
-
Shaun Leeks
Riverlane
-
Philip McBrien
Riverlane
-
Asher Pearl
Riverlane
-
Gargi Sunil
Riverlane
-
Canberk Topal
Riverlane, Riverlane Ltd
-
Joseph Yates
Riverlane
-
Ankit Zalawadiya
Riverlane
-
Alex Zeffertt
Riverlane, Riverlane Ltd
-
Marco Ghibaudi
Riverlane