When I was in high school and then in college, I used my mother’s Smith-Corona MANUAL typewriter to write all my English and journalism papers. (Yes, electric typewriters already existed, but we were “slow adopters” — aka “cheap” — heck, we didn’t even have a color TV until 1981) Those were the days of white-out, carbon paper, and lots and lots of revisions. Watching my high school-age daughters whip up their homework on their laptops, I can’t help but think how easy they have it.The same can be said for EDA tools designated for 3D TSV designs. As Mark Santoro, CEO of MicroMagic, developers of the world’s first layout editor for 3D IC, said to me, “You can use a manual typewriter to write your dissertation. It will take longer than using a word processor and you won’t end up with as good a product, but it can be done.”
At the RTI 3D ASIP Conference 2011 last week in Burlingame, CA, we heard from two major EDA tool vendors – Mentor Graphics and Synopsys – about their respective companies’ strategies in addressing the need for a dedicated 3D IC toolset.To get a third perspective, I also spoke with Santoro, whose company’s tool is reportedly in use by six different organizations to produce 3D TSV designs. Perception of industry need varies and seems to depend upon the level of risk the vendor is willing to take before 3D ICs are a commercialized reality rather than the pilot line reality that they are today.
Joe Sawicki, Mentor Graphics, started his presentation commenting on how unusual it is to have an EDA presentation at a semiconductor event. “We don’t design chips, packages, or systems,” he said, “But what we do is try to listen really hard, because we have to put the infrastructure in place that allows everyone to do the chip design.” He went on to outline the EDA challenges introduced by TSVs such as predicting impact on performance, power, reliability and cost; thermal prediction and management; 3D power and clock delivery; and 3D IC/interposer co-design. Sawicki identified a need for “TSV-aware” partitioning and placement, and design-for-manufacturability. Today, Sawicki reports, Mentor’s Calibre tool has been adapted and supports physical verification for both 2.5D and 3D IC. The company also offers tools for modeling stress as thermal management, as well as design for test. Future plans are underway for the required changes to support full 3D IC design flows. Sawicki said extensive work is being done, mostly at the university level, but he is confident that the company will have them fully developed by the time they are needed in a few years.
Last year, Synopsis stood firmly behind the conservative approach, promoting an evolutionary rather than revolutionary approach to 3D, and this year was pretty much the same. Michael Jackson, Ph.D., Synopsis noted that 3D IC is great, but the runway is still very long, and that the best launchpad is 2.5D IC for all parties involved that will provide a smooth transition to 3D IC. “We’re devoted to focus on 2.5D design with our partners,” he stated. “2.5D is technically and economically viable and makes the most sense while we wait for 3D IC to take off and mature.” He added that the company’s plan is to leverage and build off of existing tools, and that there is no need to build 3D IC tools from the ground up. Cautioning that blind investments often lead to dead ends, Jackson stated “At Synopsys we will continue to partner with our leading customers to ensure a successful long-term roadmap.”
And then there’s MicroMagic, who embarked on developing tools for 3D IC specific designs 7 years ago, before any customers were clamoring for them. Santoro said the mission was to be ahead of the curve. He agrees that while it would be foolish not to leverage what you have — MicroMagic leveraged their own 2D layout editor to develop the 3D editor — but if you don’t have good 2D tools you won’t have good 3D tools. He said even starting with their 2D editor as a basis; it still took 5 years to develop their 3D editor. “There are lots of issues that are non-trivial,” said Santoro. “It’s not that easy to slap a 3D interface on a 2D tool.” Initially, they added a 3D viewer to the 2D tool, knowing they’d need it. Viewing is complicated especially when you’re dealing with 3 or 4 stacks. With 3D floorplanning for TSVs, you’re looking for holes, he said, and it’s not just a place and route issue. “There are problems in 3D that we haven’t faced in 2D.” he noted.
The question is, will your company be the one to settle for the Smith Corona and get the job done, or will you be that kid down the hall who had the first generation Mac PC and made a fortune typing papers for all the kids who weren’t so lucky? Something to think about. ~ F.v.T.