Customer Reviews

System Design with SystemC

5 star:		(0)
4 star:		(2)
3 star:		(0)
2 star:		(1)
1 star:		(2)

 

 

I am a active user of SystemC, so had great hopes for either a real reference or even an introductory book. Unfortunately, this book provides neither a introduction to the language, real world help for coders, or even reference material. The free manual with SystemC better then this. Worse, it badly needs editing to clean up awkward sentances.

Generally I expense technical books, but I feel too guilty to bear the company with this overpriced book. Back it goes.

This book is not meant to be a introductory book or
about the SystemC language or coding guidelines.
For that please refer to other elementary or primer
book or the user guide available for download from
systemc website.

This book is mainly useful for researchers to understand
the philosophy behind the SystemC language and
interesting problems addressed by the language.

A very good book, if it is used as a reference material, specially for TLM modeling, it helps along with the code for simple_bus available at www.systemc.org. For starters, I will recommend to add more examples in the text of the book as the systemC library is huge. It is sometimes very hard to remove the bugs, if sufficient examples are not available. A suggestion will be to add a functional model example of a CPU based system design, so that the ideas can be extracted out of the text easily.

I read this whole book trying to figure out who it was written for, and failed.

To all outward appearance, it seems to address the C++ programmer who, for reasons unknown, has been thrown into the task of creating executable specifications for complex HW/SW systems. Well, it happens. That poor slob's problems are never really addressed.

The authors spend significant time discussing bit-accurate, cycle-accurate, both-accurate, and neither-accurate simulations. Those topics all deserve attention, as does optimization of the model for faster simulation time. The typical programmer, however, just won't be ready for concepts of sensitivity and some of the weirder forms of assignment. Straight-up hardware guys won't get the classing structures, and won't see all the discussion they need to translate statements into cycles; software guys won't get enough of the class hierarchy to follow along, and will need to sit down with someone for a long time to understand why all those fixed-point formats deserve to exist. (You wouldn't believe some of the questions I've gotten. For example, "0.56 isn't a fraction with power-of-two denominator - so why does that scaling logic assume I can just shift?" Lawd lumme, what kids don't learn these days.)

There's not enough here of hardware system analysis for the software guys, and not enough of the C++/SystemC strangeness for a logic designer to grab hold of. On the whole, it looks like a fair answer to a question that I didn't hear asked.

-- wiredweird

SystemC is my bread and butter and I bought this book with great hopes to understand both the language and the underlying concepts of the language, unfortunately the book provides neither. I think it is a grossly overpriced useless book.