IBM

In an annual Hot Chips conference, IBM disclosed details of their upcoming new IBM Telum Processor that is designed to get in deep learning to enterprise workloads to help address fraud in real-time. Telum is the first IBM processor that has on-chip acceleration for Artificial Intelligence inferencing during the transaction. 

After three years of research and development, this new on-chip hardware acceleration is created to aid customers to improve business insights across finance, banking, trading, insurance applications, and customer interactions. This Telum based system is planned for the early months of 2022.  

According to the latest research, 90% of the respondents stated that being able to build and run Artificial Intelligence projects whenever their data resides is vital. And the IBM Telum is created to enable applications to run effectively where the data resides, lending hand to sail through the traditional enterprise Artificial Intelligence approaches that need significant memory and data movement capabilities to tackle inferencing. With Telum, clients can build and train Artificial Intelligence models off-platform, infer and deploy on a Telum-enabled IBM system for analysis. 

Telum is the first IBM chip with technology that was created by the IBM Research Artificial Intelligence Hardware Center. And Samsung is also IBM’s partner in the development of the technology for the Telum processor.

Innovations across FinTech sectors 

Since today’s businesses use detection techniques to identify fraud when it occurs, this process can be time-consuming and compute-intensive due to the drawbacks of the technology when looking into the fraud analysis and detection is taken over in terms of mission-critical transactions and data. The major issue is that due to latency requirements, critical frauds cannot be completed in real-time. 

When we take a look at the Federal Trade Commission's 2020 Consumer Sentinel Network Databook, more than $3.3 billion has been lost by the consumer in 2020 up from $1.8 billion in 2019. The advantage that clients can get from Telum is that they can think towards fraud prevention posture rather than only thinking of the fraud detection posture. 

The new chip has a unique centralized design that allows clients to use the full power of the Artificial Intelligence processor for A-specific workloads making it best suitable for the FinTech sector. These new services clients can enhance their existing rules based on fraud detection or use machine learning, improve credit approval processes, better customer service, and attain profitability aiding to propose solutions to create more efficient processes. 

Full Stack Approach to Chip Design 

IBM's innovative design and engineering has been of great heritage value and this is also followed by the Telum. The chip is designed with 8 processor cores with a deep super-scale out-of-order instruction pipeline, that runs with more than 5GHz clock frequency level, that is optimized for the demands of diverse enterprise-class workloads. The cache which is remodified and chip-interconnection infrastructure can increase the scale to 32 IBM Telum chips. And the dual-chip design also contains 22 billion transistors and 19 miles of wire on 17 metal layers.