Archive for June, 2010

25 June, 2010

For customers who were unable to attend DAC this year, we are hosting Mentor @ DAC Extended. The registration is at:
What is Mentor @ DAC Extended?
Mentor @ DAC Extended is an online presentation of 22 of our DAC 2010 suite sessions. Each 50 minute presentation includes:
• presentation by technology area experts
• live Q&A
• an archive of the session emailed to you

The session for Calibre xACT 3D is on July 14th, 9 am US/Pacific
Here is the abstract for it:
When doing parasitic extraction, designers have long faced a performance versus accuracy dilemma. That is, to get high accuracy, they had to pay a prohibitive performance penalty. For high-performance, tradeoffs in capacitance accuracy had to be made. Calibre xACT 3D is a next generation field solver technology that solves the problem by delivering reference-level accuracy along with the performance and functionality necessary for production design flows.
We hope that you can attend!

8 June, 2010

Mentor Graphics is excited to announce the launch of their new fast field solver for IC design, Calibre xACT 3D This new tool is based Mentor Graphics’ acquisition of Pextra Corporation last year. This deterministic field solver has break-through performance, and excellent scalability, which enables the efficient use of multiple CPUs to achieve the fast turn-around-time. Since it is fully integrated into the Calibre environment, it works seamlessly with Calibre LVS, and with Calibre Interactive and RVE, which lets customers easily run the tool in their design environment using foundry rule decks.

GUI for Calibre xACT 3D

GUI for Calibre xACT 3D

In Mentor’s press release at , Stephen Fu of UMC states: “Our customers require reference-level extraction accuracy to ensure realistic simulation of complex devices manufactured at 45nm and beyond.” It is the combination of field solver accuracy, high performance, and modeling of advanced process effects and process variation that provides value for our customers.
The largest design that I have personally run on this field solver is a one million net SRAM, which ran in slightly more than 4 hours doing flat RCC extraction with 4 CPUs. This to me seems quite reasonable, considering you can also run in hierarchical mode, which would greatly speed up the extraction run. I think that field solvers are very useful, because you can attain a high level of accuracy, and with the accelerated performance and ability to use multiple CPUs, run time is less of an issue.
Do you currently use a field solver? Why or why not? What do you use it for?