EDA
System
IP
SolvNetPlus
Training & Education
Behind the world’s fastest supercomputers is an enormous amount of unseen engineering mastery. Chips must be designed, verified, and delivered on tight schedules. Expertise and resources are often limited. And complexity continues to rise.
For Hewlett Packard Enterprise (HPE), that challenge is especially acute within the Slingshot organization, which develops the high‑speed interconnect fabric that powers leading high‑performance computing (HPC) systems.
David Lacey, a distinguished technologist at HPE, sits at the intersection of engineering, methodology, and technology strategy. Trained as a verification engineer, he now works across teams to improve productivity, evaluate new technologies, and ensure that Slingshot silicon meets both quality and schedule goals.
“My role gives me the opportunity to work across pretty much all of our teams,” Lacey says. “That includes guiding how we improve our methodologies, integrate new technologies, and ultimately deliver new generations of Slingshot chips.”
Evaluating technology through a productivity lens
New tools and techniques arrive constantly in semiconductor design, but Lacey’s team takes a disciplined approach when deciding what to adopt.
“We look at what the technology is going to do for us,” he explains. “But we also look at what it’s going to take to integrate it into our workflows.”
That integration effort can be significant. Tools often need to be automated, connected to existing systems, and made usable across large engineering teams. Cost, effort, and expected return all factor into the final decision.
“If it’s a good return on investment and it helps us drive quality into our chips or pull in our schedules, then it’s something we look at very seriously,” Lacey says.
A longstanding collaboration with Synopsys
Many of these decisions are made in collaboration with Synopsys, which has been part of HPE’s design environment since the days of Cray supercomputers. That history spans multiple generations of Slingshot development.
“We’ve used Synopsys tools for many years,” Lacey says. “That continuity creates a very strong working relationship.”
Today, the Slingshot team uses Synopsys tools across the entire design flow, from front‑end development through back‑end implementation. For verification, simulation remains the workhorse, complemented by formal verification and emulation to form a comprehensive methodology focused on first‑pass silicon success.
“All of those pieces together are essential to getting high‑quality chips out the door,” Lacey notes.
Image courtesy HPE
Bringing cloud-like flexibility on-premises
One recent outcome of HPE’s collaboration with Synopsys is the certification of HPE GreenLake Flex for use with Synopsys FlexEDA Pay‑Per‑Use (PPU) licensing. Previously, that specific deployment option was limited to public cloud environments, which can pose challenges for teams with strict security and intellectual property requirements.
The newly certified solution gives engineering teams elastic, cloud‑like compute capacity while allowing them to keep workloads on-premises.
“Many teams can’t use the public cloud,” Lacey says. “This approach lets them take advantage of pay‑per‑use licensing and elastic compute, while still controlling all the security around their IP.”
The most immediate benefit shows up in turnaround time. Faster access to compute means results come back sooner, enabling engineers to respond more quickly.
“We might get regression results overnight instead of waiting 24 hours,” he explains. “That allows engineers to act right away instead of switching to something else.”
Turning faster feedback into schedule gains
Shorter turnaround times do more than save hours, Lacey says.
“When results take a full day, people naturally move on to other tasks,” he explains. “When the results finally arrive, you have to mentally switch back.”
Reducing that friction keeps engineers focused and helps quality improvements happen earlier in the design cycle. Over time, those gains can add up to meaningful schedule acceleration, especially in environments where staffing is tight and timelines are aggressive.
“Our challenges aren’t unique,” Lacey says. “Time to market and limited resources are things a lot of teams face. The question is how tools help engineers get more done in the time they have.”
Image courtesy HPE
Leaning into AI-driven engineering
As EDA tools continue to evolve, Lacey sees artificial intelligence playing an increasingly important role. He describes the industry as being at an early stage, with much more potential ahead.
“We’re just scratching the surface of what AI technologies can do,” he says, pointing to advances in generative and agent‑based systems that can reason through complex workflows.
The goal is not to replace engineers, Lacey emphasizes, but to change how they spend their time.
“We want computers handling the more mundane tasks,” he says. “That frees engineers to focus on the things only humans can do.”
Powering systems with real-world impact
The stakes are high. Slingshot fabrics power some of the fastest computers in the world — systems used for weather forecasting, healthcare research, automotive safety analysis, and other applications that directly affect daily life.
“That’s what makes this work meaningful,” Lacey says. “These systems enable research that helps humanity.”
Faster design cycles and higher-quality silicon are means to an end. For the HPE Slingshot team, that work ultimately supports the researchers, scientists, and engineers using some of the world's most powerful computing systems.
Continue Reading
