Chip Startup Ceremorphic Intend to Produce End Goods for Target Sectors

Founded in April 2020, Ceremorphic develops advanced silicon and software products for next-generation computing systems. Leveraging more than 100 patents and proven expertise in creating industry-leading silicon system products, Ceremorphic has built a new architecture that delivers the performance needed for applications such as AI model training, automotive processing, drug discovery, HPC, and metaverse processing with unprecedented energy efficiency and reliability. Designed in an advanced silicon technology node (TSMC 5nm), this architecture solves today’s high-performance computing problems in reliability, security and energy consumption across all performance-demanding market segments.  Dr. Venkat Mattela, Founder & CEO, Ceremorphic during an interaction with Nitisha elaborates Ceremorphic’s special offerings and his future strategies.

Kindly talk about Ceremorphic and its special offerings.

Ceremorphic was founded in April 2020, after Redpine Signals, a start-up that developed an ultra-low-power wireless solution that outperformed products from wireless industry giants by up to 26 times in energy efficiency, was sold to Silicon Laboratories, Inc., a US semiconductor company, for $308 million in March 2020. Redpine has been working on sophisticated wireless technologies for over 15 years, delivering industry-leading System-on-Chip (SoC) devices that are now being used in mainstream products.

Identifying the correct problem to tackle is one of the first obstacles that each firm must overcome. This is where the entrepreneur’s knowledge and experience come in handy. Some people study the competition and strive to replicate it in some way. Some businesses succeed in this manner as well. My company, on the other hand, is on a mission to change the industry by tackling the most difficult challenge in computing. We are fortunate to have identified a problem that is critical to the industry’s success and will necessitate significant effort to resolve. We’ve zeroed focus on reliable high-performance computing. We’ve already made a lot of progress in this thread. We’ll be one of the first in the industry to overcome the supercomputing reliability challenge with remarkable energy efficiency. It’s a huge undertaking. We’ve been researching for a long time, created over 100 patents and are currently developing a product based on our unique technology. By 2024, we expect to be able to produce this silicon in TSMC’s most advanced silicon node, 5nm.

Explain your journey and the idea to start this company.

We previously founded and sold Redpine Signals, Inc., a world-class wireless technology firm in Hyderabad to a well-known US semiconductor company. In Redpine Signals, Inc., we developed over 130 foundational patents in the wireless area, which we sold to SLAB for $308 million. With the success of Redpine in terms of innovation and achieving milestones previously thought to be unachievable, we decided to do the same in the AI training supercomputer space, resulting in the birth of Ceremorphic.

Ceremorphic has identified the most pressing requirement in today’s high-performance computing era: reliability and energy consumption, and we have already developed many ground-breaking innovations in response. Ceremorphic’s portfolio includes over 100 patents to date, and while the company was officially founded in April 2020, R&D efforts began three years earlier. Ceremorphic is a mature AI training supercomputing technology company with 150 employees and a highly seasoned and qualified engineering team. It has patented technology, deep technical expertise, and proven management. Product sustainability, according to the company, is only possible if the company owns and differentiates the underlying technology, hence undertook development of all the core technologies in-house.

How is Ceremorphic’s product likely to disrupt the semiconductor space? Is the Indian ecosystem ready for semiconductor manufacturing? Perspective.

The semiconductor technology roadmap includes tools and nodes for developing advanced silicon products, with the 5nm node being particularly efficient for producing high gate density (approximately 40M gates in 1 sq. mm area) and high performance. If a company has access to this type of technology node, it can create extremely efficient and high-performance processors in a manageable physical die size. Managing leakage power and reliability, on the other hand, is a challenge with these nodes. These two issues, reliability and energy efficiency, are now limiting the adaption of AI / ML applications all segments in a pervasive way. .

Ceremorphic’s current technology portfolio, as well as the design of QS1 (the first chip in the Hierarchical Learning Processor™  – HLP series), will address reliability and energy efficiency issues, paving the way for an architecture that will enable Exascale performance while maintaining a manageable power budget. This method will reduce the cost of high-performance computing systems, allowing AI and machine learning applications to become more widely used.

Now, when it comes to manufacturing in India, there are two factors to consider: talent and infrastructure. In India, talent is never in short supply, but infrastructure and previous proven experience are essential for achieving any meaningful outcome at scale. Both of these aspects of the ecosystem are improving in India in a noticeable way.

Kindly talk about the new reliability architecture adopted in its 5nm Hierarchical Learning Processor (HLP).

Ceremorphic intends to deliver a complete silicon system capable of supporting next-generation applications such as AI model training, HPC, automotive processing, drug discovery, and metaverse processing. This new architecture was designed from the ground up in advanced silicon geometry (TSMC 5nm node) to solve today’s high-performance computing problems in reliability, security, and energy consumption to serve all performance-demanding market segments. Ceremorphic is one of the few chip startups with access to TSMC’s 5-nanometer node, but its real differentiation will come from in-house technologies that provide competitive advantages in efficiency, reliability, and security for AI, high-performance computing, and beyond.

For applications such as model training or ADAS (automotive), the reliability is a paramount criterion.  Coupled with this, huge integration and multiple processing nodes (thousands of processing cores) makes the problem more noticeable.  Solution is very expensive if we use conventional methods of duplication that’s why the industry needs a new reliable architecture using machine learning approach to predict faults and correct them using optimal hardware-software approaches.

Kindly highlight the technology behind delivering Reliable Performance Computing in an Energy-Efficient AI Supercomputing Chip.

Ceremorphic uses its own patented multi-thread processor technology ThreadArch® in conjunction with cutting-edge new silicon algorithm technology. The team is using its extensive knowledge and patented technology to create an ultra-low-power training supercomputing chip. The challenges that this market faces with ‘reliable performance computing’ cannot be solved with existing architectures; rather, a completely new architecture built specifically to provide reliability, security, energy efficiency, and scalability is required. This industry desperately needs reliable performance computing, and Ceremorphic’s approach is a significant step in the right direction.

Hierarchical Learning Processor (HLP) deploys the right processing system for optimal power performance operation. Key features of the QS 1 include the following:

• Custom Machine Learning Processor (MLP) running at 2GHz
• Custom FPU running at 2GHz
• Patented Multi-thread processing macro-architecture, ThreadArch® based RISC –V® processor for proxy processing (2GHz)
• Custom video engines for Metaverse Processing (1GHz) along with M55 v1 core from ARM®
• Custom designed X16 PCIe 6. 0 / CXL 3.0 connectivity interface
• Open AI framework software support with optimized compiler and application libraries
• Soft error rate: (1000,000)-1

The Ceremorphic architecture is intended to scale across a wide range of compute-intensive markets and applications, including AI training supercomputing, data centre processing, automotive, metaverse processing, robotics, and life sciences.

How was the year 2020-21 in terms of business?

Inspite of 2021 being the year when the Covid-19 was still prevalent, we were fortunate to execute our plan efficiently. We create huge IP in the connectivity, machine learning and compiler technology. We also engaged with potential partners in 2021 and the same is being carried forward in this year. We were very successful in collecting critical requirements and folded them into to our final product specification.

Ceremorphic now employs 150 people, 140 of whom are based in India. We intend to increase this to 250 by the end of the year. We expect to have over 300 employees working from the Hyderabad location by the end of 2024. We have a very clean and efficient infrastructure of 35,000 SFT space in Hyderabad with all ultra-modern networks and computing systems in India. Since the company’s inception in April 2020, we have been operating from 26 cities in a remote manner with all our computing resources placed in Datacenters.

In summary, 2021 was a very fruitful year from both technology development and business traction. We started working from on-premises in April of this year and we are very satisfied with the results.

Kindly share your future market strategy for Ceremorphic in India.

A product of QS1’s complexity necessitates considerable effort. We are utilising the talent and knowledge of the team in India to efficiently create this product. Because of the importance of high-performance computing in the future of AI and machine learning, our Go to Market strategy will be global in scope, with India as a target market.

Through our OEM network, we intend to produce end goods for target sectors such as datacenter, AI training, automotive, robotics, metaverse processing, and life sciences. We will also be developing system solutions with our own hardware after-market products, workstations and servers. The India Development Centre will have a significant impact on R&D and product development, and it is already involved in the development of architecture, algorithms, semiconductor design, and software tools. However, the product will be available throughout the world.

Vertical integration has become key to the companies in the industry. Ceremorphic’s unique technology portfolio coupled with its qualified team makes it a perfect partner to many industry giants in their own field.