As my colleagues Adam
Sherer and Joe
Hupcey reported last week, Cadence has just published "A Practical
Guide to Adopting the Universal Verification Methodology (UVM).” It is the
world’s first book on the UVM. Or is it? The definition of "book" has
become rather fuzzy these days. Could the UVM Reference Manual and User Guide
published in PDF form by Accellera qualify as books? Would we still say that we
had "published a book" if this "Practical Guide" were
available only as PDF? What if it were available only as an e-book? What if we
had tweeted it out 140 characters at a time?
I've been involved at some level or another in more than a dozen book
projects over the years, and it's always interesting to hear the differences of
opinion on how to go about it. I did a lot of the logistical work for
publishing the Verification Methodology Manual (VMM) for SystemVerilog
when I worked at Synopsys. I specifically wanted a book that would sit on
every verification engineer's desk, and so decided to use a traditional
publisher (Kluwer) and made sure that the Field had lots of copies to hand out to
customers. The book sold quite well in its time, and still pops up now and then
although it's somewhat outdated.
After I joined Cadence and we embarked on the Open Verification Methodology
(OVM) we specifically wanted an independent, hardcopy book to sit alongside of,
and eventually displace, all those VMM copies I had distributed in my
previous job. So I worked with an independent author, Sasan Iman of SiMantis,
linking him up with our OVM experts to help him complete his book. And yes,
Cadence ordered a bunch of copies for the field to distribute to customers. Step-by-Step Functional Verification with SystemVerilog and OVM
is a very good book, but also a bit outdated since the OVM evolved so rapidly.
So here we are today, with the Accellera Universal Verification Methodology
(UVM) standard available in Early Adopter mode, and again we are opting to
publish a hardcopy (although softcover) book and order lots of copies to spread
around. So what did we learn from our past experiences? Won't the UVM also
evolve quickly and outdate this book? Why did we not just do an e-book, for
which updates would presumably be easier? Why not just distribute a free PDF
file rather than charging for a book? Are we just stuck in an old mode of
thinking? Do hardcopy books still have value?
Actually, a lot of things are different for the UVM book. We
self-published with Lulu, so we can sell the book for less than other
methodology books on the market. Lulu provides on-demand publishing, so we
don't have to figure out how big a press run is needed or risk getting stuck
with outdated inventory when we want to update the book. Accellera has
committed to a high degree of backward compatibility for the UVM going forward,
so the initial book will still have value even when production versions of the
UVM are released. Finally, we will have e-book versions available for sale as
The one thing that hasn't changed is that many engineers prefer to have a
hardcopy book by their side. Browsing, flipping, highlighting and marking pages
are more satisfying that way. Our Field team also loves books; they make great
gifts for customers and giveaways at raffles. The book announcement has
generated a buzz of excitement that I don't think could have been matched by an
e-book or online documentation. Our Field and our customers are clamoring for real,
physical books, and that's why we made the choice to publish the world's first
UVM book in a traditional format with some modern twists.
The truth is out there...sometimes it's in a blog.
Fair enough, but beyond definitions I'm interested in how the state of information transmission is evolving. With all due respect to all the other formats available, it is very interesting to see how much value is still placed on a physical book.
Tom: A book is a book if it has an ISBN number, at least for the agency that tracks these matters. It's ISO standard 2108, of all things. Of course, a book is a book even without an ISBN number, but it won't have the same credence as one with that number.