These are unabridged comments from some of the students who took the CS/EE 6761 (Parallel Computer Architectures II) course from me in the Spring quarter, 1998, in which I had used the on-line (free!) version of the textbook by Culler, Singh, and Gupta. Kishore Ramachandran Associate Professor College of Computing Georgia Institute of Technology Atlanta, Ga 30332-0280 Ph: (404) 894-5136 ===================================================================== Dr. Ramachandran, Thanks for a great experience in CS 6760 and 6761. I feel like I've learned a lot about parallel computer architecture. You've done an excellent job teaching the courses. I thought I'd offer some input on the 6761 textbook. The book, on the whole, is quite good. The authors have made an impressive effort to present a subject that is complex and changing - a daunting task. So although the comments below are critical, I just want to point out that I did enjoy using the textbook. Hopefully the authors will find these comments useful in making their next revision. I guess the main suggestion for improvement of this book I have is to decrease the wordiness. Often a point is obscured by complex and unwieldy text. For example, on page 441 (section 7.3.1 - transaction ordering), the text reads, "Whereas a bus provides strong ordering properties among transactions...". Maybe I misunderstood the point, but I think this would read much better as, "Whereas a bus provides strong ordering of transactions...". The latter is much clearer to me. In many places in the book it appears as though the authors intentionally use big words to convey their ideas when simpler ones would suffice... not necessarily bad; I like for my vocabulary to be challenged! But sometimes the authors choose these big words which have a different shade of meaning than what is intended. This misuse of big words makes their text look impressive, but obfuscates their points. I've seen this effect when an author overuses the thesaurus in his/her word processor... they type what they mean using small words, and then search the thesaurus to find big words that supposedly mean the same thing. But often, these words have a slight shade of meaning that changes the nature of the sentence. The wordiness in the text takes on another form. Sometimes the text belabors a point with many sentences when only a few would suffice. When using this text, I found myself often writing in the margin, distilling a paragraph down to its core points, so I wouldn't have to search through the paragraph again to try to determine the author's point. My suggestion here is for the authors to think about each paragraph: 1. What am I trying to say? 2. What is the point of this paragraph? Perhaps making a detailed outline would help here. One thing I think would really help this text is if there were "key points" sections highlighted in the text. As a student, it is sometimes hard to differentiate between a key point, something you definitely want to remember, and an ancillary point, one that is not essential. If the essential points were highlighted or recapitulated in an obvious manner, I think that would improve the usability of the text. I know these comments above have been quite general. Here is a more specific comment: Sometimes the authors use equations without explaining the notation. For example, on page 711 (10.5.1), the authors describe a routing algorithm as a function R:NXN->C, where none of the variables are described. What are R, N, and C? Throughout the chapter, C is meant to represent capacity, but I don't think that's what is means in this equation. Without knowing what the variables represent, one can only speculate on the meaning of the equation. Well, that's my input! Again, I think the book is quite good. If I went into what I liked about this book, this meessage would probably be a novel! So although this email is critical, one should keep in mind that there are many excellent features in this book. My main suggestion is to put the book on a diet! :-] Regards, | Greg Slabaugh | Graduate Student, DSP | Center for Signal and Image Processing | School of Electrical and Computer Engineering | Georgia Institute of Technology | | e-mail: slabaugh@ece.gatech.edu | web: http://www.ece.gatech.edu/users/slabaugh/ ===================================================================== Respected Sir, I apologize for the late feedback on textbook that you had asked all of us to send in. It's just that I had to re-install windows three times in the last one week on my PC at home. I also had to replace the DVD decoder card in my PC. Also I was hunting for papers on Real Media, Quicktime and Microsoft Netshow. FEEDBACK 1. In general trim the whole textbook. 2. In particular trim the following chapters - 1, 5, 7, 8, 10, and 11 3. Page 440 : The first sentence in the paragraph on Protection should be changed. Thanking you, Edward. "Edward S. Rasalam" ===================================================================== I thought the book was very good. It was really thorough as mentioned in class. I found it hard to sit down and read for very long just because it was so verbose. The best way to read it was to read the first couple of pages about each new topic and skip a lot of the case studies and analysis of individual machines and implementations. Thanks, Jon Felten Jonathan Felten ===================================================================== Dear Dr. Ramachandran, Thank you for two very instructive quarters. My suggestion on improving the book, is to include several flow charts which describe the protocols of the different concepts. This would clarify some of the very wordy parts of the book. Other that and a couple of grammar mistakes, Its not really a bad book. Especially since in was free :) Thanks again. Sincerely, David Young "David Young" ===================================================================== Dr. Ramachandran, As has been stated in class, the textbook was rather "wordy", making it really annoying to read. This may have been in part due to the fact that we had it as a collection of postscript files rather than in actual book form. However, the text itself is mostly the problem. The book needs to cut to the chase and present all the necessary information in a more accessible manner. Once someone reaches graduate status and is half burned out, searching through extraneous information becomes bad. Thanks, Rick Presley Rick Presley Research Engineer I Georgia Tech Research Institute Information Technology and Telecommunications Lab Phone: (404)385-0288 Fax: (404)894-9081 =====================================================================