Cadence® system design and verification solutions, integrated under our System Development Suite, provide the simulation, acceleration, emulation, and management capabilities.
System Development Suite Related Products A-Z
Cadence® digital design and signoff solutions provide a fast path to design closure and better predictability, helping you meet your power, performance, and area (PPA) targets.
Full-Flow Digital Solution Related Products A-Z
Cadence® custom, analog, and RF design solutions can help you save time by automating many routine tasks, from block-level and mixed-signal simulation to routing and library characterization.
Overview Related Products A-Z
Driving efficiency and accuracy in advanced packaging, system planning, and multi-fabric interoperability, Cadence® package implementation products deliver the automation and accuracy.
Cadence® PCB design solutions enable shorter, more predictable design cycles with greater integration of component design and system-level simulation for a constraint-driven flow.
An open IP platform for you to customize your app-driven SoC design.
Comprehensive solutions and methodologies.
Helping you meet your broader business goals.
A global customer support infrastructure with around-the-clock help.
24/7 Support - Cadence Online Support
Locate the latest software updates, service request, technical documentation, solutions and more in your personalized environment.
Cadence offers various software services for download. This page describes our offerings, including the Allegro FREE Physical Viewer.
The Cadence Academic Network helps build strong relationships between academia and industry, and promotes the proliferation of leading-edge technologies and methodologies at universities renowned for their engineering and design excellence.
Participate in CDNLive
A huge knowledge exchange platform for academia to network with industry. We are looking for academic speakers to talk about their research to the industry attendees at the Academic Track at CDNLive EMEA and Silicon Valley.
Come & Meet Us @ Events
A huge knowledge exchange platform for academia. We are looking for academic speakers to talk about their research to industry attendees.
Americas University Software Program
Join the 250+ qualified Americas member universities who have already incorporated Cadence EDA software into their classrooms and academic research projects.
EMEA University Software Program
In EMEA, Cadence works with EUROPRACTICE to ensure cost-effective availability of our extensive electronic design automation (EDA) tools for non-commercial activities.
Apply Now For Jobs
If you are a recent college graduate or a student looking for internship. Visit our exclusive job search page for interns and recent college graduate jobs.
Cadence is a Great Place to do great work
Learn more about our internship program and visit our careers page to do meaningful work and make a great impact.
Get the most out of your investment in Cadence technologies through a wide range of training offerings.
Overview All Courses Asia Pacific EMEANorth America
Instructor-led training [ILT] are live classes that are offered in our state-of-the-art classrooms at our worldwide training centers, at your site, or as a Virtual classroom.
Online Training is delivered over the web to let you proceed at your own pace, anytime and anywhere.
Exchange ideas, news, technical information, and best practices.
The community is open to everyone, and to provide the most value, we require participants to follow our Community Guidelines that facilitate a quality exchange of ideas and information.
It's not all about the technlogy. Here we exchange ideas on the Cadence Academic Network and other subjects of general interest.
Cadence is a leading provider of system design tools, software, IP, and services.
Get email delivery of the Cadence blog featured here
Last week Dr. Andrew Kahng came to town. He was at CDNLive, where his presentation Toward New Synergies Between Academic Research and Commercial EDA won the best paper award for the academic track.Then the following day, he presented at the (internal) Cadence Distinguished Speakers Series, where he talked about PPAC Scaling at 7nm and Below. I first met Andrew back when I was at Cadence around 2000, when we were both on the Cadence Technology Advisory Board. He hasn't presented since that era since he opened by saying the last time he was in the River Oaks Cafeteria with ice cream, instead of the building 10 auditorium with pizza. At various times my office was in buildings 1 and 2, which for those of you too recent to know, were tilt-up buildings on the corner of River Oaks and Seely where condominiums now stand.There were four buildings there, which is why buildings on the current campus start from building 5, and, like Spinal Tap's amplifiers, go up to 11.
By the way, PPAC stands for power-performance-area-cost. The industry has talked about PPA for a long time, with the A for area also being a surrogate for cost. But with different process choices, multiple patterning vs. EUV someday and other options, area alone is not the only parameter that feeds into cost.
Andrew sees two megatrends that are driving all the issues.
The first megatrend is what he calls "the race to the end of the roadmap." This is advancing Moore's Law to the end of what we know as fast as we can. Despite economic challenges, the technical challenges are being addressed despite a few major issues: the lack of EUV and the lack of a replacement back-end technology to replace copper, restricted design rules, reliability limitations. The result is volume production of 7nm in 2018. Another major issue is guard-banding with excessive pessimism, making design excessively hard and impacting yields.
The second megatrend is keeping power under control. Low power is essential in all markets, from mobile to big data to cloud. We have done a lot of the easy stuff in previous process generations and now need more extreme approaches.
One major tradeoff is optimization versus schedule. Moore's Law is 1% per week, meaning that schedule trades off with PPAC. There are other rules of thumb, such as a mV reduction in voltage margin translates into another 5MHz of operating frequency.
Going deeper into the details, the first area that Andrew has been working is on adaptive voltage scaling. There are a number of forms of this, but the basic idea is the same. Instead of fixing the supply voltage based on pre-tapeout analysis, take measurements off the actual silicon and lower the supply voltage to reflect the actual margin. This requires an on-chip monitor, typically a ring oscillator (RO). Andrew's group at UCSD has been looking at ways to do this such that all sample chips meet timing. Since threshold voltages change with aging, voltage scaling can also be used to compensate over time, with the design signed off using aged parameters but performance improved for most of the life (which also slows aging).
The next area Andrew looked at is that combining worst-case corners derived from 3σ distributions can end up with corners that are much more pessimistic than necessary. In essence, two 3σ distributions combine to form a square when the true 3σ limit is a circle. Since most critical nets are routed on multiple metal layers, and variation in those layers is largely uncorrelated, this creates a lot of opportunity for tightening the corners.
There is "free" margin that can be recovered. Libraries are characterized a certain way, but the reality is that there are tradeoffs between, for example, clock-to-Q, setup, and hold times for a flop. The right tradeoff is not constant and so each FF can use its own "best" set of values compatible with how the silicon actually behaves. There is also a lot of free margin to be picked up when different timing engines produce different results. For example, in one experiment done at UCSD there was as much as 123ps slack divergence resulting in 20% performance difference, which is a whole node of Moore's Law scaling. Applying big data machine-learning approaches can up modeling, reducing that 123ps divergence to just 31ps, a 4X reduction.
Most foundries have multiple height libraries: large height with better timing but, obviously, more area and more power. Small height has smaller power and area but longer delays and often a requirement for more buffers. Andrew's group has done work on mixing cell height, which normally is not done, and getting better results. Of course, cells cannot be mixed arbitrarily due to things like power supply architecture, but can be partitioned and legalized to end up with mixed areas with better overall results.
At the 7nm and 5nm nodes, layers are one-dimensional grids with divisions done using separate cut masks. This produces more controllable layout that attempting to use few masks and much larger spacing (assuming no EUV for now). The\ interconnect isn't all that needs to be colored; the vias and the cut masks do, too. This creates more opportunity for further optimization when trading off timing against metal density rules and resolution enhancement technology (RET) rules. Optimizing end-of-line extension can produce better tradeoffs.
Andrew's final call to arms was for a massive "moonshot" to predict tool outcomes, find the sweet spot for different tools and flows, and thus design in specific tool and flow knobs to the overall methodology. This would combine all the ideas already discussed (and others that I haven't had space to cover) and so end up with a fully predictive, one-pass flow with optimal tool usage. With modern massively parallel, big data architectures, it is not unreasonable to use tens of thousands of machines if it could "get us to the moon" of a non-iterative flow.
Previous: Phil Moorby and the History of Verilog
Next: Patents and Standards, Managing the Challenge