The business use cases for high performance computing (HPC) are rapidly expanding. One of the most promising and powerful new ways to use HPC is in what are called “cognitive” applications.
Cognitive computing creates a “human and machine” partnership to understand, reason, and learn. It helps make sense of the massive amounts of data created by sources such as instrument telemetry, the Internet of Things (IoT), and everyday business and manufacturing operations. And it goes further – making it easier to find the answers to complex questions. It can amplify the benefits of analytics and automation with greater flexibility and speed.
Manufacturing is one business sector where cognitive computing is already having a significant impact. Within this sector, electronics manufacturing is leveraging the benefits of machine learning and other cognitive computing techniques to improve and enhance design, testing, and manufacturing efficiency while lowering costs.
When we think about manufacturing in electronics we usually think about the assembly of the final product, rather than the manufacture of all of the individual components. Right at the start of the supply chain are the microchips themselves. As components, they are relatively cheap per unit, but the design complexity and manufacturing setup costs are staggering. For example, last year Intel announced it was investing $7 billion dollars over the next four years to complete its Fab42 facility in Arizona to produce 7nm chips, while in December 2016 TSMC (Taiwan Semiconductor) announced plans to invest roughly $16 billion dollars in a fab facility for 5nm and 3nm chips.
At such a small scale, the density of components can lead to unpredictable behavior with electrostatic and quantum effects that cannot be adequately modeled using existing tools and techniques. Designers must often make compromises to try and minimize these effects.
More importantly, it also means that previously insignificant imperfections in the silicon substrate or within the manufacturing processes itself can significantly impact the wafer yield or manufacturing failure rate. While manufacturing yields are a closely guarded secret, broadly speaking, as the density and clock speed goes up, the yield decreases. Thus, as chips shrink to 5nm and 3nm, yield rates decrease, which impacts profitability. Something needs to change.
Semiconductor manufacturing is already well advanced in autonomous operation and process automation, but there is always room for improvement. In a recent IBM study, “Why cognitive manufacturing matters in electronics,” 92% of electronics executives indicated that they were planning or implementing big data analytics, with 83% seeing moderate to significant ROI. Meanwhile 100% indicated that they were planning for or implementing Artificial Intelligence (AI)/cognitive computing to aid in both chip design and manufacturing.
While these numbers seem high, they are not that surprising. For example, AMD has been using Hadoop to improve yield predictions for several years, and at the International Solid State Circuits Conference in 2017, TSMC illustrated that the speed of a chip could be increased by 40Mhz through the use of cognitive (machine learning) techniques to predict congestion during place and route. In addition, the EE Times article AI Tapped to Improve Design discusses a research program IBM has launched along with eight other companies and three universities to investigate the application of machine learning in electronics design.
With today’s complex designs, millions of verifications can be required. But what happens if you find a serious bug a few days before manufacturing configuration or tape out? Delaying tape out can have significant revenue impact. This is one area where cognitive systems may be able to help. If during the design process we can input knowledge about the various verification runs, we might be able to determine which of those millions of tests we really do need to rerun to test the fix, and which have no relevance – reducing the retest cycle and minimizing the market impact.
There is significant interest in hybrid cloud where specific design/big data/cognitive workloads are sent off-premises. The design and verification of chips takes significant compute power, and the desire to apply cognitive techniques will continue to drive the demand for compute resources which may be satisfied by the cloud. In the same IBM study noted above, 100% of respondents indicated they were implementing cloud computing. Security is often cited as a barrier to cloud adoption, and in an industry dominated by design intellectual property (IP), such a high rate of adoption seems strange. However, many semiconductor companies run large compute farms that many classify as private cloud.
Many of these private/hybrid cloud environments utilize automation solutions from IBM Spectrum Computing.
Hybrid cloud may be powerful, but it’s not necessarily simple by nature. Cloud compute resources, unlike cloud storage, can be more expensive than on-premises resources. Security is always an issue. And you need to have apps that are cloud ready. The IBM Spectrum Computing family of workload solutions is designed to address the wide range of management challenges found in today’s HPC, AI, and data analytics environments, including hybrid cloud. For users, it can help you take advantage of technologies such as accelerators designed to speed results. For the infrastructure, it can help get the most from additional compute capacity available in the cloud during peaks in workloads.
With such large investments in manufacturing, the combination of analytics, automation, and cognitive offers significant competitive advantage to chip manufacturers – and a growing list of other industries and organizations. Learn about how IBM Spectrum Computing solutions can help enhance agility and productivity in your enterprise here.
 IBM Institute for Business Value: Why cognitive manufacturing matters in electronics, February 2017 https://public.dhe.ibm.com/common/ssi/ecm/gb/en/gbe03806usen/global-business-services-global-business-services-gb-executive-brief-gbe03806usen-20180205.pdf
 IBM IT Infrastructure blog: Bill McMillan – Improving chip yield rates with cognitive manufacturing, March 2017 https://www.ibm.com/blogs/systems/improving-chip-yield-rates-with-cognitive-manufacturing/