Plunify: Unifying Programmable Logic with Machine Learning Technology
The global chip shortage dominates the news and the semiconductor industry worries. Even consumers who thought nothing about what powers their mobile devices understand the next model may be years away due to the shortage.
One workable way out of the shortage is to reuse existing field programmable gate array (FPGA) designs with today’s technology and Lanza techVentures portfolio company Plunify could just be part of that solution. Its design software, in use at various systems companies developing FPGAs, optimizes a design’s performance and eliminates many performance deficiencies.
Plunify, the name a combination of programmable logic (PL) and unify, was founded in 2009 by Harnhua Ng and Kirvy Teo who understand the challenges confronting FPGA engineers using vendor-specific design tools. They identified a way to compile and optimize register transfer level (RTL) code for an FPGA application using powerful, machine learning algorithms. The tool called InTime applies synthesis and place and route (P&R) technology to analyze design data and optimize it for better performing FPGA designs.
In the early 2010s, that concept was unheard of. Plunify was way ahead of the market and one of the lone providers touting machine learning-based chip design software. Today, machine learning, deep learning and artificial intelligence are frequently used terms in the semiconductor lexicon. InTime can be used as a Software-as-a-Service in the cloud as well. Even in the semiconductor industry, cloud-based design solutions are commonplace and driving many new application segments.
A Friendship Started in Junior High
Stepping back a few years, Harnhua and Kirvy are friends who met at a Singapore junior high school. They took different routes through college — Harnhua went to Carnegie Mellon University and earned a Bachelor of Science degree in Electrical and Computer Engineering and a Master of Science degree in Electrical Engineering. Kirvy majored in Computing at the National University of Singapore.

HarnHua Ng, Founder and CEO
They reconnected when Harnhua was living in Japan and working at AMD as a platform engineer after doing a stretch as a Xilinx application engineer. Kirvy was at a mobile communications startup. He and his new wife honeymooned in Japan and prevailed upon Harnhua to show them around. As they caught up, they realized they both had a desire to do a startup and agreed to combine skills to meet their ambition.
The FPGA space, they decided, was changing and expanding, which meant new challenges and an opportunity for a startup. A data science approach to FPGA design software was a natural extension of what they both knew and brought their two disciplines together.
Bootstrapping, Funding and Advisors
Harnhua and Kirvy bootstrapped Plunify for several years while they built a prototype. Funding came through Singapore startup grants and Kirvy explored opportunities afforded by the National University’s alma mater network.

Kirvy Teo, Founder
Fundraising began in earnest in 2011. Then, Rick Carlson, vice president of sales at Verific (see more about Rick below), introduced them to Mark Templeton, a seed investor from Silicon Valley and noted executive who founded Artisan Components (now Arm), in 2014. Harnhua and Kirvy met Mark at a conference in Northern California and learned Mark and Lucio Lanza, managing partner of Lanza techVentures, were investment partners.
Lucio invested in Plunify at Mark’s recommendation. “I trusted Mark for his business acumen and his intellectual honesty.” (Mark died in a kayaking accident in 2016.) Mark believed FPGAs were the center of the semiconductor world because the number of changes a chip through in its lifetime makes programmability a critical requirement. Equally important, chips need to be consistent with industry trends. As well, a chip controls power and performance. Mark saw Plunify as a conduit to ensuring FPGA designs meet specs and are optimized for high performance. Its InTime tool could make the design last longer.
In their final assessment, Mark and Lucio concluded Plunify could help more individuals become FPGA designers. The relationship between Plunify and Lanza techVentures developed quickly and Plunify became a portfolio company.
A brief interlude here to formally introduce Rick, another important figure in the Plunify story. Rick read about Plunify in a Synopsys blog post back in 2010 or so and thought it had an exciting idea. Harnhua smiles as he remembers, “Rick called across the ocean late one night from his home in Colorado to learn more.” Rick has been an advisor ever since, offering advice and counsel, helping make connections and introductions wherever he can.
Now we come to Plunify’s Norman Yeung phase. Norman was a founder and CEO of Sandcraft, developer of embedded processors (now part of Broadcom) and another Lanza techVentures portfolio company. Norman moved from C-suite management to engineering management with Silicon Valley household names, including Cadence Design Systems in 2014, and stayed in touch with Lucio. At Cadence, Norman was vice president of R&D for the Hardware System Verification Group where he oversaw the development of the Palladium emulation platform.
Norman left Cadence in 2018 and joined Lanza techVentures as an investment partner. From his early days as an investor and advisor, Norman was intrigued by Plunify and helps out anywhere he can. According to Harnhua, he offers practical, hands-on advice. He’s a big picture thinker and can also help with tactical details. Norman, in many ways, assumed the role Mark played to nurture Plunify and mentor Harnhua and Kirvy.
And nurture, he has. As Harnhua explains, Plunify began as a vision of these two passionate engineers. They knew data analytics combined with FPGA domain expertise could resolve complex design challenges and significantly accelerate a product’s time to market. They firmly believed Plunify’s approach to design optimization would offer faster results and reduced Capex to FPGA users and potentially to ASIC designers also.
Harnhua offers assurance that the mission hasn’t changed and Plunify is out front leading the pack. Awareness through partnerships with Xilinx, Amazon Web Services and several OEM agreements has helped. Plunify today has a reputation for delivering results in markets as diverse as communications, test and measurement and audio/video processing. Emerging markets for Plunify’s software include financial trading, a popular FPGA-based application, and military and aerospace electronics.
“Plunify today has a reputation for delivering results in markets as diverse as communications, test and measurement and audio/video processing. Emerging markets for Plunify’s software include financial trading, a popular FPGA-based application, and military and aerospace electronics.”
With momentum building, the next step is going vertical. Plunify’s InTime software is horizontal, according to Harnhua. FPGA designers use it to optimize a design at the synthesis and P&R level. The data generated teaches the tool about the characteristics of test and measurement as compared to those of telecommunications, for example. Vertical use gives Plunify insights gained from the data for specific markets. Eventually, Plunify expects to expand to power applications. It can because InTime knows FPGA architectures and its learning algorithms enable the software to get smarter the more designs it runs. InTime can be uniquely designed for each user and builds on itself. As Harnhua noted, “It’s a learning process and the highest-achieving application for specialized FPGA targets.” Just consider a few results. An engineer working on a design can achieve the target performance in a matter of hours, reducing the turnaround time by 50% or more.
Plunify went from a lone provider of machine learning-based FPGA design software to being a technology leader solving complex chip design timing and performance problems through these techniques. It also is a good fit within the Lanza techVentures investment portfolio of innovative design technologies and advanced AI-based, edge computing applications.