IBM Launches a Fraud Fighting Chip and the Outer Space Threat to Quantum Computers

This Week in Engineering explores the latest in engineering from academia, government and industry.


Episode Summary:

IBM has launched a novel processor that is optimized for a special application: finance. The new Telum processor is built for AI integration, which with the right software, can perform real-time analyses of financial transactions and stop fraud as it happens. IBM predicts that the technology will similarly be useful for real-time credit approval for online banking and transaction settlement with very little latency. The processor is built around eight 5 GHz cores and contains an incredible 22 billion transistors. 

A global team of university and industry researchers have discovered the root cause of a serious problem in the development of quantum computers: correlation. Detecting and correcting errors is a major roadblock in quantum computer development, and the team’s discovery that single errors can affect a large number of qubits simultaneously is a significant step forward in developing practical quantum computers. Just as significant is the discovery that a major source of these errors are cosmic rays, highly energetic and difficult to screen. Future quantum computers may require heavy radiation shielding for optimum performance.  

Access all episodes of  This Week in Engineering on engineering.com TV along with all of our other series. 

Transcript of this week’s show:

To see any graphs, charts, graphics, images, and/or videos to which the transcript may be referring, watch the above video.

Segment 1: Industry stalwart IBM have announced a new integrated circuit with a difference. The company’s upcoming Telum processor is designed with a special application: transaction fraud. According to IBM, the new processor is IBM’s first that uses on-chip acceleration for AI inferencing while a transaction is taking place. Current systems use detection techniques that catch fraud after it occurs, and due to latency requirements very sophisticated fraud detection frequently can’t happen in real time. For example, criminals may successfully purchase goods with a stolen credit card before the transaction can be flagged or stopped. 

The Telum chip contains eight 5 GHz processor cores with a new cash and chip interconnection infrastructure that allows systems that can be scaled to 32 Telum chips. Like all modern integrated circuit technology, the statistics are almost inconceivable: 22 billion transistors and 19 miles of wire arranged on 17 metal layers. The chip is one of a new generation of integrated circuits that are optimized for specific tasks demanded by economic activity in the Cloud. 
IBM designed the chip for AI specific workloads, including fraud detection, loan processing, clearing and settlement of trades, anti-money laundering and risk analysis. While helpful for detecting fraud, the system may also be used to create real-time credit approval processes and use AI to modify or optimize settlement processes with minimal latency. The market potential for the new device is huge. According to the Federal Trade Commission, in 2020 consumers reported $3.3 billion in losses to fraud, up from $1.8 billion in 2019.  

Segment 2: Quantum computing promises great devices with problem-solving capabilities that are orders of magnitude better than the fastest supercomputers today. While the payoff is considerable, engineering quantum computers is extraordinarily difficult. They operate at cryogenic temperatures and in quantum realms where conventional material science and traditional binary logic simply don’t apply. A major problem is error correction. 

So far, quantum computing is severely constrained by the need to detect and correct errors in real time, and new research by a large team from the University of Wisconsin Madison, the Fermi National Accelerator Laboratory, Stanford University, INFN Sezione di Roma, and Google Inc. suggests that there is a new and previously unforeseen cause for quantum computing errors and how quantum circuits react to them. The team built a quantum testbed device and discovered an unusual property: fluctuations in the electrical charge of multiple quantum bits or qubits, the fundamental unit of quantum computer, are not random and independent. 

If a random burst of energy enters from outside the system, it affects every qubit in the vicinity of that event simultaneously, a phenomenon the team calls “correlation”.  This creates errors that can instantly span the entire system. Previous error correction techniques have assumed that errors can be tracked to a point source, but the new theory suggests that entire system can be simultaneously disrupted, complicating the error correction process. Just as interesting is the analysis of the source of many errors: gamma radiation from cosmic rays. 

Cosmic rays are highly energetic atomic nuclei that come from multiple sources outside our solar system. Most are deflected by the Earth’s magnetic field, but some get through and on hitting the Earth’s atmosphere create a secondary emission of other particles that scientists call an “air shower”. These charged particles have been known to create errors in conventional integrated circuit electronics. In 1990, an IBM study showed that conventional computers experience one cosmic ray induced error per 256 MB of RAM per month on average. 

Error correction code or ECC memory was developed to address this issue, but the unusual behavior of quantum state devices, namely correlation, makes the radiation challenge much more difficult to solve. 6 inches or more of dense radiation shielding might prove to be the easiest technique to mitigate cosmic ray induced errors in quantum computers, but some particles are highly energetic and it’s unlikely that a truly radiation proof processor we will be developed anytime soon. Error detection and correction through software will be with us for a long time to come. 

Written by

James Anderton

Jim Anderton is the Director of Content for ENGINEERING.com. Mr. Anderton was formerly editor of Canadian Metalworking Magazine and has contributed to a wide range of print and on-line publications, including Design Engineering, Canadian Plastics, Service Station and Garage Management, Autovision, and the National Post. He also brings prior industry experience in quality and part design for a Tier One automotive supplier.