New Parallel SPICE Start-up Company

posted in: Chip Design Mag | 4

SPICE is a wonderful tool for transistor-level circuit designers to predict speed, timing and power of their sensitive designs across Process, Voltage and Temperature regions or corners. One problem is simply run-time of SPICE, it can be slow enough to take days or even weeks to complete a simulation.

cyberedaCybereda is the latest start-up EDA company to offer parallel SPICE technology with a tool named PCSIM. The claims sound reasonable, about a linear speed-up as you add more cores or CPUs to your design.

An 8-core version of PCSIM leaves commercial SPICE in the dust:

Right now the company web site talks about how great their tool is however we don’t hear from any customers yet, so stay tuned to learn how this parallel SPICE tool will compete with other earlier entrants, like:

Berkeley DA is hiring a parallel software developer in India, so expect another competitor in 2010.

4 Responses

  1. Daniel Payne

    Good point. All start-ups are very enthusiastic about their technology however business success comes when you start to announce the names of your clients and the projects they are working on.

  2. Colin Warwick

    Hi Daniel,

    Thanks! This is an important class of tool. But how about a shout out for the version of SPICE in Agilent ADS which has had multicore support (at no extra charge) for almost two years!

    (It used to be called HF SPICE but we call it ADS Transient Convolution Element now because we added supports for s-parameters via a patented convolution algorithm that ensures causal behavior.)

    We also support massively parallel BSIM4 model evaluation via NVIDIA Tesla GPGPUs and an add-on product called ADS Transient Convolution GT.

  3. Daniel Payne

    Thanks for the info, I’ve updated my post to reflect your SPICE product offerings. Agilent has mostly been used by SI engineers instead of IC designers, so I forgot about your tools in my original post.

    Nascentric was the only other SPICE company to use the GPU approach and saw about a 4X speed up. How is your GPU approach doing in the market place?

Leave a Reply

Your email address will not be published. Required fields are marked *