HyperLynx What's new in VX.2.13

Q&A

Q: It looked like the simulation cluster was running on an Amazon AWS instance.  Is using commercial cloud servers allowed by the HyperLynx license?

A:  The computational cluster used for the results we presented is called a “HyperLynx Cluster”, where the HyperLynx Job Distribution manager manages the 3D simulation process directly by transferring data files, launching solver jobs and pulling results back once an individual solve is complete. This is ideal for situations where you have spare computing capacity that isn’t part of a managed compute farm (LSF, SGE or other). HyperLynx Job Distribution also works with a number of different compute farm managers and also supports cloud-based simulations with AWS, where the simulation licenses are pulled from the user’s local license manager to authorize simulations in the cloud.

Q: How does HyperLynx automatically detect where 3D modeling is required?

A: HyperLynx uses the DRC engine to analyze serial links and determine areas where Transverse ElectroMagnetic (TEM) Wave propagation will be disrupted, which is where 3D EM modeling is required. The DRC engine can detect those areas, determine where TEM flow ends on one end of the structure and begins on the other end (the minimum area that needs to be modeled), and also determines the extent of the signal’s return path, automatically including in the resulting structure for EM modeling. 
 

Q: For this large design example, how much time was required with model setup and were the models from Intel?

A:  For the PCIe Compliance Analysis we showed, no time was spent obtaining and setting up simulation models, because we weren’t running IBIS-AMI simulation – we were analyzing the serial links themselves for compliance with the PCIe Gen5 standard. This is part of the HyperLynx Progressive Verification methodology – find as many problems as you can, as quickly as you can, with as little effort as possible. There’s no point in going to the time and effort of running IBIS-AMI simulations if the serial links themselves don’t meet the spec requirements – and that’s what Compliance Analysis does. IBIS-AMI analysis would normally follow the Compliance Analysis step.

Q: Does HL provide 3D signal path and return visualization as you showed earlier in the presentation?

A:  Yes! In fact, the animation we showed was generated using the HyperLynx Advanced Full Wave Solver, plotting current behavior at the PCIe Gen 5 Nyquist frequency (16 GHz). Here is the link to the Support Kit that will help you create current animations: MG621713 .

Q:  Are you planning to use Nvidia GPU's on computational card (or even high end graphics card) as accelerators for Hyperlynx?

A:  We don’t use CPU cores for simulation acceleration at present. This has been an on again / off again topic in EDA for many years, but it never seems to have caught on widely. We suspect there’s a mismatch between what graphics cores are designed to do (highly repetitive, pipelines processing) and the nature of problems solved by EDA simulators. For that reason, we scale by using an array of general purpose processors, using tools like HyperLynx Job Distribution.

Q:  Are the cable simulator models only coax or are there other models available?

A:  There are Bundled, Bundled Twisted Pair, Flat Flexible, Flat Ribbon, and Twinax Cable types-  in addition to Coaxial Cables.

Q: For the Cable modeling, how does that work? Does the user select the cable and it will create an S-parameter that the user stitches in using LineSim? Or some other way?

A:  You can use the 3D Explorer parameters to specify the cable structure and design as well as set figures of merit to let 3D Explorer optimize a cable design. The tool will create experiments to find the optimal cable design. The resulting S-Parameter model can then be used in Time Domain simulations.

Q:  I may have missed it but are there any hl-pi vx2.13 updates?

A:  VX.2.13 was just released, so there were no updates at the time of the webcast. VX.2.12 has 3 updates available on SupportCenter.