You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
New research posted yesterday in the journal Science claims that testing accelerates the learning process, and appears to be more effective than other techniques such as concept mapping and straightforward study.
As an EDA user, the most effective training was that which offered the shortest path to productivity (and even provided the best catering ). Now, as an EDA provider, I'm interested in approaches that have the most significant impact. Incorporation of testing into training material is potentially an approach worth considering, along with self-paced/online vs. classroom-based learning etc. etc.
In an ideal world the products we use wouldn't even require training (or documentation for that matter), but in the fast paced world of chip design it will probably be unavoidable for the foreseeable future.
If you are a 'receiver' of EDA training (or any other type of training for that matter), what techniques in your experience have been most effective? Alternatively, if you are a 'provider' of EDA training, what has worked best for you and where do you see the future of customer education?
Ron,
I've taken EDA training and delivered EDA training, both at Intel and with EDA companies. My personal learning style is to learn by doing, so that means that I get the most out of the lab exercises. During lecture I enjoy the short pop quizes to keep me on my toes. If I cannot attend a class, then the next best thing is a self-paced tutorial with some pop quizes.
The last class I taught was on Verilog for beginners at a semiconductor company and we had good feedback on the lecture/lab approach. Students also appreciated being able to ask any question during lab or lecture to clarify their understanding.
Thanks for the feedback. The consensus these days is clearly that 'death by slides' as an approach to training doesn't work. The trick for us in EDA with our limited resources is going to be to figure out how to manage training classes where the emphasis on a hands-on approach lead to much more interaction.
I think the effective use of animated training materials has a substantial role to play in EDA as well as other semiconductor related training. For example a visual represntation of how a complex computing application is splitted and executed in a parallel computing environment or how a PCB inspection machine works will make the learning quicker and easier for students.
I am a firm believer in the "learn by doing" approach. We make a tool (Breeze) which automatically documents and visulises programs and processes so that new team members can see what their system is doing and get to grips with their area more quickly. After all, a picture speaks a thousand words. Breeze can be found at www.ellexus.com and can be used to maintain in-house code as well as train up new members of your team more quickky. I'd love your feedback on it.
One of the problems with "learn by doing" for the EDA vendor is providing non-proprietary data for class use that still illustrates the tool features that support the latest technology. Generic or open-source libraries don't do that very well. If the number of customers is small, I find the best way is to do the training at the customer site using their data. It may take more time to collect the data and get it set up, but the attendees get the most out of it that way, IMO. For a large number of customers, or for training at the EDA vendor's location, using generic data is the only way, and that has its limits.
Simon,
I agree with you on the training data.
I usually bring along generic data for a training class, then invite the students to try it during lab time on their own proprietary designs and libraries.
@harpoon, it's not a matter of "not being good enough." Sometimes the issue is tool features that support a specific technology or foundry. It's difficult to represent that with a generic library. We are using generic examples in a lot of cases, and they are good enough to train people on the generic capabilities of the tools. When it comes to something like TSMC 28nm layout rules or model-based DFM, there is no way to do that without the foundry specific data, which can't be done in an open class that may include TSMC's competitors. EDA vendors have been walking this tightrope for a long time.
). Now, as an EDA provider, I'm interested in approaches that have the most significant impact. Incorporation of testing into training material is potentially an approach worth considering, along with self-paced/online vs. classroom-based learning etc. etc.