WP_Term Object
    [term_id] => 64
    [name] => Solido
    [slug] => solido
    [term_group] => 0
    [term_taxonomy_id] => 64
    [taxonomy] => category
    [description] => 
    [parent] => 157
    [count] => 58
    [filter] => raw
    [cat_ID] => 64
    [category_count] => 58
    [category_description] => 
    [cat_name] => Solido
    [category_nicename] => solido
    [category_parent] => 157

Machine Learning and EDA!

Machine Learning and EDA!
by Daniel Nenni on 04-21-2017 at 7:00 am

Semiconductor design is littered with complex, data-driven challenges where the cost of error is high. Solido’s new ML (machine learning) Labs, based on Solido’s ML technologies developed over the last 12 years, allows semiconductor companies to collaboratively work with Solido in developing new ML-based EDA products.

Data acquisition is expensive; brute force methods are time and resource intensive. Large amounts of data require a high level of expertise to provide valuable insights. Many EDA teams don’t have the expertise or resources to quickly and successfully parse this overwhelming amount of data, which can also be hard to visualize and interpret. Additionally, there is an important need to seamlessly integrate solutions into current design flows. Overlooking one of these elements can lead to poor designs, limited scalability, delays, or even more.

Solido has developed proven machine learning technologies for engineering applications. Engineering challenges are unique, as users are making expensive decisions where cost of errors is high. Results from ML technologies must not be estimations, but rather production accurate and verifiable. Solido’s ML technologies, developed over the last 12 years, form the basis of Solido Variation Designer and produce production-accurate results—not estimations. Variation Designer’s ML technologies build adaptive, self-verifying models that detect and correct errors automatically. Results are verifiable, and can be trusted by users. Solido’s ML technologies are scalable to 100K+ input variables, parallelizable to large clusters, and capture high-order interactions, non-linearities, and discontinuities (eg bi-modal, n-modal, binary, n-ary).

Overall, Solido’s ML technologies create large speedups, accuracy boosts, increases in coverage, and reductions in computing resources and license usage. All this resulting in faster time-to-market, improved designs and reduced engineering costs.

Solido is introducing Machine Learning (ML) Labs to make their machine learning technologies more accessible in solving an expanded range of data-intensive problems, improving access in applying their ML expertise and technology to EDA’s biggest challenges.

Here’s how ML Labs works:

  • Bring your EDA design challenges to Solido
  • Solido’s experts will work with you and your designers on how these challenges could be solved with either their existing ML technologies, or if new ML technologies are required, using Solido’s team of ML experts
  • Solido will work with you as a lead customer to bring the technology to a production EDA software product

Solido has the industry’s top EDA and ML experts, who develop innovative ML solutions, effective rapid prototypes, and conclusive proof-of-concepts. Their product integration team will make the solution work with your tools, in your environment. Solido already has experience in integrating new technologies with top EDA tools, which they can leverage to accelerate time-to-solution and make it work in any design flow. Their usability experts make their solutions easy to learn and use for your designers, providing support throughout the deployment and production use. With ML Labs, Solido’s high-quality team will be with you at every step along the way, to make it “just work” in production.

The first two products to come out of Solido ML Labs are ML Characterization Suite’s Predictor and Statistical Characterizer. Predictor uses machine learning to model the full library space using data from existing characterized library models. This reduces library characterization time by 30-70%, while saving on characterization licenses, simulation licenses, CPUs, disk, and time. Statistical Characterizer generates statistical timing data >1000x faster than brute force while maintaining Monte Carlo accuracy. It does this by adaptively selecting simulations to meet accuracy requirements while minimizing runtime for all cells, corners, arcs, and slew-load combinations.

You can find more information about ML Labs at http://www.solidodesign.com/ml-labs or by contacting Solido at mllabs@solidodesign.com.

About Solido Design Automation
Solido Design Automation Inc. is a leading provider of variation-aware design software for high yield and performance IP and systems-on-a-chip (SOCs). Solido plays an essential role in de-risking the variation impacts associated with the move to advanced and low-power processes, providing design teams improved power, performance, area and yield for memory, standard cell, analog/RF, and custom digital design. Solido’s efficient software solutions address the exponentially increasing analysis required without compromising time-to-market. The privately held company is venture capital funded and has offices in the USA, Canada, Asia and Europe. For further information, visit www.solidodesign.com or call 306-382-4100.

Share this post via:


0 Replies to “Machine Learning and EDA!”

You must register or log in to view/post comments.