Pages: pp. 6
While reading "Back to Thin-Core Massively Parallel Processors" (A. Marowka, Dec. 2011, pp. 49-54), I noticed an easily correctable error.
The article incorrectly states that the BlueGene/P computer is not cache-coherent: "The Blue Gene/P supercomputer (1-PFLOPS) is an example of a contemporary large-scale system design (294,912 processors) based-on a many-thin (850 MHz, 1GB) processing core approach. The cores are not cache-coherent, and each node uses a thin version of the POSIX operating system."
While BlueGene/L's two cores on a chip were not cache-coherent, BlueGene/P's four cores are coherent. That was one of the main changes from /L to /P (V. Salapura, M. Blumrich, and A Gara, "Design and Implementation of the BlueGene/P Snoop Filter," Proc. IEEE 14th Int'l Symp. High Performance Computer Architecture [HPCA 2008], IEEE, 2008, pp. 5-14).
The author responds:
After reading the HPCA article, I agree that the four cores in a Blue Gene/P node are cache-coherent. Unfortunately, it seems that other papers led me to an incorrect conclusion.
I thank Kim Yates for bringing this information to my attention.
An article in Computer's December issue ("Peter W. Staecker Chosen as IEEE President-Elect for 2012," Computer Society Connection, p. 70) incorrectly states that "Division directors represent IEEE societies on the IEEE Board of Directors and Technical Activities Board."
Each year, the IEEE attorneys emphasize that members of the IEEE Board do not represent the constituency they come from but must act in the broad interests of IEEE as a whole. Thus, IEEE directors can take no instructions from their constituency as to the position they should take in any matter. The same principle applies to the Technical Activities Board and, indeed, the Computer Society Board of Governors, who, according to the IEEE Constitution, serve the same constituencies as the IEEE Board: IEEE members worldwide and the general public.
We appreciate Peter Clout's assistance in pointing out this error. The sentence he quotes should read, "Division directors serve on the IEEE Board of Directors and Technical Activities Board."
I will miss the The Profession column in Computer. Even though I didn't always agree with the content, I looked forward to reading it each month, and I appreciated the fact that Neville Holmes always offered good reasons for his comments.
Regarding the reference to computer-induced autism in the December column ("The Profession and Digital Technology," pp. 116, 114-115), I think Holmes reversed the analogy. Those who walk around staring at their iPods or talking on their iPhones are actually communicating with other people in real time.
In fact, modern technology now makes it possible to interact with others more frequently than at any time since humans adopted agriculture. Tending the fields and watching the sheep were quite solitary occupations. The introduction of craft and manufacturing occupations did not improve things since people had to pay attention to what they were doing. Hunter-gatherers, by contrast, were in constant contact with members of their group.
What's different about this new form of social networking is that you can now choose the people with whom you wish to associate. There has always been some ability to choose, but your schoolmates, coworkers, and neighbors were determined by factors other than personal choice. With social networking, that's not the case. You can choose to be friends or not with someone on the Web.
One implication of this ability to choose is that it will be easier to avoid people who are different. When this happens, it becomes more difficult to have empathy for people in different circumstances simply because you don't know personally anyone who is in that situation. For the same reason, achieving consensus on public policy becomes more difficult as individuals view their situation as typical.
I'm sure there are many more implication to this new technology of social networking. I'm also sure that sociologists will have a wonderful time exploring all of them.
In "Toward More Precise Radiotherapy Treatment of Lung Tumors" (S.S. Iyengar et al., Jan. 2012, pp. 59-65), the information in the author credits for S.S. Iyengar and Puneeth Iyengar (p. 65) was incorrect. The correct information is: S.S. Iyengar is the Ryder professor of computer science and director of the School of Computing Science and Information Computing at Florida International University. Puneeth Iyengar is an assistant professor in the Department of Radiation Oncology at the University of Texas Southwestern Medical Center.