> From: John McKown <[EMAIL PROTECTED]> > > I found the original machine: a 9672-R55. That is ancient. And trying to say > "If I get ... result on the R55, then I will get ... result on a z9" seems > unlikely to me. They play around with LSPR values to do their extrapolation. > But they then use that general to make a statement about a specific, single, > application.
Hmm. I'd have to read more about their extrapolation, but I don't see how it is possible to extrapolate z9 or z10 performance and power consumption based on a 10+ year old machine that isn't even a zSeries, much less a reasonably current one. So I guess the study proves that if you use the latest and greatest Windows hardware and software, you can get better performance and power consumption than you would with a 10+ year old mainframe. Go figure. Offhand, I wouldn't be surprised if you could buy a dozen 9672-R55 machines and run them in parallel to equal the performance numbers and _still_ come out ahead of the cost for the Windows systems, even with power consumption. Let's see a real test. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
