Cisco software includes a compiler that enables quantum workloads to run in distributed environments, including multivendor quantum compute nodes.
Cisco has announced a package of prototype software it says will facilitate distributed quantum computing networks and support real-time applications.
Cisco’s Quantum Labs designed a software stack that includes three layers: an application layer with a network-aware distributed quantum computing compiler that supports quantum algorithm execution in a networked quantum data center; a control layer with quantum networking protocols and algorithms that support the applications as well as manage the devices (hardware and software) that make up a quantum network through northbound and southbound APIs; and a third layer for device support, consisting of an SDK and APIs to physical devices as well as a library of emulated and simulated ones.
[ Related: More Cisco news and insights ]
Essentially, Cisco has developed the software to enable distributed quantum computing in the quantum data center, Vijoy Pandey, senior vice president and general manager of Outshift by Cisco, told Network World.
“We have always believed in a full-stack approach to building robust, high-performing systems. For classical networking, we design and build this stack from the ground up – custom silicon, integrated hardware systems, and all the layers of software that control and manage everything,” Pandey wrote in a blog about the new software. “It’s the philosophy that enabled us to pioneer many of the foundational technologies of the classical internet, from the core of the data center to the edge of wireless networks and all the connectivity in between.”
Some of the first devices running the prototype software are based on the quantum entanglement chip announced by Outshift in May that generates pairs of entangled photons that can instantly transmit quantum state between each other, regardless of the distance between them. The entanglement chip generates 200 million entangled pairs per second. The chip operates at room temperature, uses minimal power, and functions using existing telecom frequencies, Panday said.
The chip is designed to work with existing infrastructure, meaning it can send photons over existing fiber, and it operates at standard telecom frequencies, so there’s no need to rip and replace anything to support it, Pandey said. In addition, because of these properties, customers could deploy gear supporting the chip alongside an existing classical computer infrastructure, Pandey said.
The software stack supports three other prototype applications to help enable quantum networking and the data center. The first is what Pandey describes as a network-aware distributed quantum compiler that lets quantum algorithms run across multiple networked processors.
“The compiler is the piece of technology you need to enable practical, pragmatic, distributed quantum computing. It takes a quantum workload, a quantum circuit, and it partitions it so that it runs in a distributed environment, in a connected set of qubits or quantum compute nodes,” Pandey said. Significantly, it’s multivendor; the quantum compute nodes can be from the same vendor or from other vendors, such as IBM: “It could be as messy a brownfield, heterogeneous environment as you want. It doesn’t matter to the compiler, which will take an algorithm, partition it across any heterogeneous, brownfield environment,” Pandey said.
“What makes it unique, and an industry-first, is that it accounts for quantum interconnect requirements between processors and supports distributed quantum error correction. Existing compilers target circuits for only single computers,” Pandey stated. “Ours compiles circuits for network-connected computers potentially made of heterogeneous quantum compute technologies and can distribute that partitioned circuit across an entire data center of processors, all connected through a quantum network.”
The distributed quantum error correction is a key feature of the software. Error correction ensures the accuracy and reliability of quantum computations and is a challenge for any distributed or standalone network.
The Cisco software in this case understands the error correction intricacies of each of the quantum computing modalities in the network, and “we can ensure that those are carried over from node to node, giving us a distributed or a holistic view of the entire distributed environment and result,” Pandey said.
In addition, “we are developing our own algorithms [to determine] the best way, using our network, to do a distributed architecture and error correction,” said Reza Nejabati, chief of quantum research and the Cisco Quantum Lab. “For our purposes, when we demonstrate the technology, we use surface code, which is a type of quantum error-correcting code, and we will support more advanced correction code, like bicycle code, in the future,” Nejabati said.
Other research prototypes being announced by Cisco include Quantum Alert and Quantum Sync.
Quantum Alert can tell two endpoints in a quantum link if there is an intruder in the system. “So, the way we do it, you use it in an entanglement-based network, and if there is an intruder in your network, they will destroy that entanglement property. Because, by the very motion of tapping into a network, they have destroyed entanglement that exists between those two endpoints. If you deploy this, we can tell you that somebody is actually in your network and snooping and tapping into your network,” Pandey said.
Quantum Synch uses quantum entanglement to enable distributed systems to make correlated decisions without sending messages, Paney wrote.
“Think of it as two nodes, Alice and Bob, getting a special pre-correlated coin in a box each, where if Alice opens it and finds a heads, Bob is guaranteed to find a tails. An example use case is in high-frequency trading where milliseconds equal millions and financial institutions could lose money to coordination problems every day. We demonstrate this using our quantum network simulator with real protocols,” Pandey stated.
