Mathematics of Electronic Design Automation
Just an observation but one of the biggest problems I see in EDA, Semiconductor IP, and life in general is poor math skills. Look at the sub prime loan fiasco, Wall Street and the crippling national debt, some people are just bad at math. One of the best applied math books I have read is “Mathematics of Poker”, by Bill Chen and Jerrod Ankenman. A simple card game is fueled by elementary math, probability, and advanced statistical analysis. It is a must read for EDA and Semiconductor IP executives.
Expected Value (EV) is a mathematical term that is relevant to both gambling and business. +/- EV is how much you expect to gain or lose as a result of decision you make, over time. For example, suppose you have a 50% chance of winning a $100 pot. You either win $100 or will win $0 but your expected value is $50. The way you calculate expected value is: (percentage chance of winning * pot size). The EV of the play is based on how much you wager to gain the $50.
Using the Design Automation Conference as an example, is exhibiting at DAC +EV or -EV? With the declining attendance numbers it is a difficult question for industry executives.
A simplified EDA company EV calculation for DAC would be D dollars spent versus R direct revenue recognized. Unfortunately very few of the EDA companies I know effectively track or manage this equation, and it requires too much overhead for the smaller ones. Top tier EDA companies also have alternate venues for customer interactions such as SNUG, CDN Live, and MUSIC. Even worse, some trade show plans are fueled more by ego than math, which makes the yes or no DAC question more of a gamble than before.
The first Design Automation Conference I attended was in 1984, Albuquerque New Mexico, which was one of the first ones to allow exhibitors. It was an interesting time, so much innovation, so much excitement, the Design Automation Conference was the cornerstone of EDA and was absolutely +EV. I have attended each DAC since and have witnessed the roller coaster of EDA year after year. EDA marketing/PR continues to outpace engineering and revenue streams are controlled by rabid discounting versus customer gain-share or proven value propositions. For those of you that foresee the demise of EDA, the most fitting place for the eulogy would be the Design Automation Conference.
Exhibition management could certainly reverse the DAC EV decline. One idea that I floated to MP Associates (manager of DAC) is to supplement the floor by adding villages: Analog Village, IP Village, ASIC Village, etc… These villages would have a scanner at the entrance, be populated by demo kiosks and AE’s, with a uniform format for signage and promotions, a level playing field. The (3) immediate EV benefits I see are:
- More efficient foot traffic. Attendees can focus and optimize time spent, see more product demos making their DAC experience +EV.
- The focus is on EDA technology and innovation not cancerous EDA Fear Uncertainty and Doubt (FUD).
- Top Tier EDA companies can trim their massive ego booths and go head-to-head technology wise with the other companies in the village. Emerging companies can spend less VC money on marketing and more on innovation.
Bottom line is: increased attendance, increased company participation, increased innovation, DAC would be +EV for EDA once again.
Nice Dan,
You make some good points.
“EDA marketing/PR continues to outpace engineering”
You hit the nail on the head. It’s hard to distiguish the true from the false.
Mike
Dan,
Nice to see you on the train. Keep it up. Your ideas about DAC have been pushed for some time but I never have seen them in print. Hope someone is listening.
Good ideas however the DAC management team has a long cycle to incorporate changes. I do recall the notion of Villages being tried at least once before. Having similar industries grouped together is certainly a rational request and would reduce the amount of travel time by visitors.
In the bygone glory days of EDA you could actually see real, live demos in the booths.
Today you are lucky to even see a Powerpoint presentation in the booth. All of the real, live products are hidden in the demo suites, and you have to pre-qualify to get let into the fortress. I’ve been allowed into demo suites only to be removed midway through because of “competitive concerns”. Not really an open society yet.
This is a thought provoking first post, I am glad to see you blogging.
My first DAC was in 1982 in Las Vegas, it was the first time they let vendors exhibit. There were a set of 6′ long tables in a hallway between the meeting rooms and the casino floor.
Unlike a poker pot, few EDA markets are winner take all. At a minimum it’s at least high-low game with two winners and there is typically room for at least one if not two other players. There are some exceptions (e.g. license management) but because the customers want to differentiate they are normally willing to support two or four vendors in a segment.
I think it’s a mistake to criticize larger EDA firms for purchasing larger booths. In the end it’s their choice. A trade show is a stag hunt, it’s an attempt by a collection of firms to create a larger shared opportunity that when divided yields them a larger benefit than if they prospected separately.
Any one hunter or set of hunters can defect from the larger hunt and either hunt on their own or form a smaller hunting party. I think it’s important to remember that EDA suppliers normally have little market power: the top 50 customers or so are larger than any of the EDA firms, and the once you leave the top 10 EDA firms the bottom 97% are smaller than probably 80% of the buyers (by purchase dollar).
I am not sure how much new business from non-customers CDN Live and Music will bring Cadence and Magma respectively. They are a great vehicle for servicing existing customers but it’s not clear that they even help them grow share of those customers.
But going back to your earlier argument. Let’s say that the top ten vendors pulled out and left the bottom 300 at DAC. The show would go on because the academics need a place to present, the IEEE and ACM need a place to conduct standards and other business, customers would come because they would be looking for innovative new startups that would allow them to differentiate from their competitors (few fims will bet all future designs on company, even if it’s Synopsys, and if one company does, it’s competitors lose nothing by at least evaluating, if not spending enough to keep one or two competitors alive in each segment).
300 or so firms each paying 5-10K is still enough money to fund a decent show, and certainly an innovative one.
Unfortunately, I think, I’ve been around longer than most or all of the previous commentators. The first DAC with exhibits was Minneapolis in 1981. CAE Systems was one of the exhibitors and had a puppet show mocking the other exhibitors which directly lead to the rather stringent show floor rules. Albuquerque (1984) was comical in that it rained in the middle of the outdoor DAC party at the Hilton Hotel – rain in June in Albuquerque? Who would have thought?
Whether or not a market is “winner take all” is irrelevant, it merely simplifies the math. Dan’s comments are still correct. But, part of what makes the equation more challenging to derive is that companies are also paying attention to other audiences such as VCs and the press – albeit a dwindling group for EDA. DAC is one place where small companies find it easier to get seen and heard by all audiences. There is also a DAC equation related to the show floor draft. Skip a year and your draft order position gets hammered.
I do not mind the big EDA companies taking large spaces at DAC and I think that could still happen in addition to the village concept. What I think we need from the big EDA companies is market leadership. Cadence pulling out from DAC in past years was a mistake in that it hurt the entire industry. Similarly, talking down the venture capital market for EDA back when Mike Fister took the reins at Cadence was also a mistake that hurt EDA overall. The model of start-ups developing technologies and getting acquired worked, and worked well, for the industry and for the customers. It was not planned, but evolved, as a capitalistic market with sufficient capital should. A much smaller collection of start-ups exists now simply because Fister claimed he was not going to buy more companies (so the VCs did not invest in very many start-ups). This, of course, turned out to be false. But in the wake of this statement were many ideas and companies that never got funded. Big companies need to help grow the industry – those with the largest market share should get the biggest benefit from this. The “all you can eat” licensing deals come from a fixed-pie model and that mind set leaves it nearly impossible to grow the market when it is the mind set of market leaders.
Dan – very nice (and funny) blog – I hope to read it all before I retire, which is way the hell out there. I think I started the first EDA design center in the Bay Area (Wyle Labs ASIC design center) – we had Daisy (ooh, aah, graphic schematic capture!), and the very powerful LSI Logic/IBM mainframe netlist simulator (all test vecotrs in ASCII characters) – 10K gates was a huge design. Selling EDA was all about – and still is – easing “engineering anxiety.” BTW, I coined that phrase. Anyway, now I am selling supercomputing interconnect (InfiniBand) – just moved higher up the food chain – it’s much easier and more predictable than EDA, and it really helps to be a sole-source supplier, so now I don’t have to worry about the 80-20 rule, and I can sell myself and still get a PO – because I don’t know what the hell I’m talking about . . .
LinkedIn Groups
* Subject: New comment (1) on “Mathematics of Electronic Design Automation”
Thanks Daniel for starting this blog.
We should all know that, as we all do, Electronics in all aspects of its existence, required an extensive knowledge of Mathematics and Physics.
This is a good start of this discussion and we can talk about the ways we could improve this importance.
Posted by Saeid Sarir