Chips & SupplyUnited States

Solidigm eyes US NAND fab as funding questions linger

Credited to DigiTimes$ · digitimes.com4h ago

Useful

Solidigm, a subsidiary of SK Hynix, is considering building a NAND flash wafer fab in the US. Funding and the investment timeline have yet to be settled, but DIGITIMES Intelligence believes the project is highly likely to move forward, given AI-driven storage demand and the…

View the article · $DigiTimes

Front

Chips & Supply · United States

IBM, AMD, And Oak Ridge Unveil Quantum Error Correction Platform

Credited to The Next Platform · nextplatform.com · 18h ago

Over the past few years, quantum computing companies have made remarkable strides in addressing the central challenge of correcting errors the systems make, a hurdle that needs to be cleared before practical, commercial fault-tolerant quantum computing can fully be realized.

We have written about a range of approaches to error correction that are being worked on, from running an error-correction code with a quantum processing unit (QPU) created with a different grid to using a decoder that can run on a single CPU to leveraging AI reinforcement learning.

Each effort is made to reach the same basic goals, including reducing the number of errors quantum systems make and lowering the technical overhead needed to make this happen to help them scale and keep the costs down. Logical qubits – the group of many physical qubits – is a key technique for reducing errors, but more needs to be done. Quantum systems typically now have gate error rates ranging from one error per 100 to 1,000 operations. The error rate needs to improve significantly for fault-tolerant systems to become a reality.

The problem is that qubits are extremely fragile and can fall apart – or decohere – and lose their quantum nature through noise, which can range from light and sound to the activity of other qubits.

IBM, AMD, and the Quantum Science Center at the Oak Ridge National Laboratory this week unveiled a co-design platform for error correction (below) that integrates quantum computing with classical HPC systems and, importantly, puts the systems next to each other in the same location. This rarely happens, according to IBM, which means that connectivity needs to be made through the cloud with the help of APIs, with can lead to greater complexity and delays.

“In order to experiment with these techniques and build algorithms, quantum and classical processors need to be able to work closely together,” Seetharami Seelam, Distinguished Engineer and chief architect of classical system co-design for quantum-centric supercomputing, tells The Next Platform.

Quantum systems will work closely with traditional supercomputers in hybrid, quantum-classical environments, with quantum processing units (QPUs) essentially acting as another accelerator, alongside GPUs, TPUs, and other chips. AI will work as a connective tissue between the systems, with a lot of work being done – including by Oak Ridge scientists – to determine the roles generative and agentic AI will have in these environments.

“The collaboration will also focus on exploring how AI can develop more effective algorithms and workflows, as well as advancing error correction through both new methods and improvements to existing approaches,” Seelam says. “The GPUs that are part of the co-design platform will play a key role in running the AI models or algorithms that scientists plan to develop and improve upon.”

The co-design platform will be housed at IBM’s facility in Poughkeepsie, New York, and include IBM quantum computers, in particular the company’s 120-bit Nighthawk r2 quantum chip and Oak Ridge’s Umbra HPC system (below), which is powered by AMD’s fifth generation Epyc “Turin” Epyc 9005 datacenter CPUs and “Antares” Instinct MI355X GPUs. The AMD chips will power the classical system’s research into quantum error correction, decoding optimization, and software orchestration, according to the organizations.

Researchers will be able to use the tightly integrated platform to test and develop quantum-classical workflows, test quantum algorithms, and learn more about what error-corrected systems will need without having to pull together the infrastructure themselves. The goal is to use the platform as a jumping-off point for creating quantum systems and software in the years ahead.

IBM officials put an emphasis on Nighthawk’s square lattice connectivity, a change from Big Blue’s previous “heavy-hex”-based chip architecture. Square lattice connectivity offers a two-dimensional grid topology where most qubits connect directly to four nearest neighbors, about a 60 percent increase in connectivity over heavy-hex. It also leads to shorter execution times and compatibility with standard surface code for error correction. Combined high speed in maximum circuits per second and real-time pulse control, researchers will be able to better test error correction techniques and compare them in terms of qubits, time, and classical compute resources to get a better handle how GPUs can best facilitate quantum computing.

Through IBM’s quantum API, researchers can access deep pulse capabilities that will allow them to measure qubits mid-calculation and then deal with the results in real time, including set the exact timing of signals that can protect qubits from noise.

IBM pointed to hierarchical error correction as an example of what can be further explored with the platform. With the hierarchical approached, two codes are layered together. One is an inner code for protecting qubits within a module, with the other an outer code adding a second layer of protection. An advantage is that this approach keeps the hardware and connectivity demands simpler. According to IBM research, the approach could reach a trillion logical operations at a physical qubit error rate of one in 1,000.

It also relies on classical computing, and the platform will let researchers use GPUs to test such idea in an environment with low latency.

It also dovetails with the goal set by the US Energy Department’s Quantum Genesis Mission for developing fault-tolerant quantum systems. IBM, in conjunction with the University of Chicago, in July demonstrated experiments that involved up to 70 logical qubits and 438 hard operations, according to IBM’s Seelam.

“The Quantum Genesis Mission’s Q Competition targets 100 logical qubits running 100,000 such operations by late 2028,” he says. “To reach this goal, the broader quantum community needs to achieve a 238-fold increase in hard operation count, and we need to do it fast. We believe this goal is feasible, but achieving it will require progress across several areas and close collaboration among industry, national laboratories and government researchers.”

The tightly integrated quantum-classical platform will offer the resources needed to test systems to reach the Genesis Mission goal, Seelam says.

IBM also will take advantage of the work done on the platform as it develops its Quantum Starling system, which the company says will be a fault-tolerant computer that is expected to be delivered by 2029.