I think IFLs run about 10% of a full speed CP processor. So it doesn't take too much to make them cost effective.
On Wed, Oct 30, 2013 at 10:02 PM, adarsh khanna <adarshkha...@yahoo.com> wrote: > Thanks Timothy.I get it. > IFL costing across models is more to do with marketing and product > positioning. > > > > On Tuesday, 29 October 2013 2:45 PM, Timothy Sipples <sipp...@sg.ibm.com> > wrote: > > Bob Shannon writes: >>Each book has to have at least one configured PU, so >>the minimum configured PUs for the CEC would be four. > > Is that true uniquely for the 2827-HA1? It's not true for, say, the > 2827-H43 as far as I know. Single engine 2827-H43 machines are reasonably > common. Could you be thinking of memory? There is a minimum installed > memory required in every processor book. I think it's 16 GB on current > models. > > Agreed, this topic is about 99% academic, though it's sometimes fun to > explore the corner solutions. > > Adarsh Khanna asks: >>Does the cost of adding an IFL different on different machines >>e.g. 2817 compared to 2098. If yes why? as it is just characterization >>of a core. > > Not exactly. IFLs are no longer available for purchase on machine type > 2098. Said another way, the price of IFLs on 2098 machines is infinite. :-) > > From time to time these sorts of questions come up. I don't know why, > because the answer is obvious if you've ever bought anything in your life. > The simplest answer is that you should rapidly disabuse yourself of the > idea that price must equal either marginal or average cost (plus perhaps > some fixed percentage). I don't know why people think mainframes are > expected to violate what is so completely common in the world, even > (especially?) in the computing world, in terms of pricing. As an example, > what's the difference between Microsoft Windows 8 and Microsoft Windows 8 > Pro? Answer: A flipped entitlement bit and many dollars in price. > > As another example, a few hours ago I stepped off an airplane. The price of > the airline ticket was about US$1300, which was a high price and which > generated a hefty profit for the airline. At the same time, although I > don't know what price she paid, the passenger sitting to my right > undoubtedly paid a much lower price. (My ticket price would have been the > top price on the plane in economy class. It was in fare class Y and booked > mere hours before the flight. It had to be done.) > > Anyway, unless you work for a government agency -- and even then there's an > argument! -- you most probably work for a company that seeks to maximize > profit. Your company is (usually) constrained in that effort by various > forces, including competition. (Yes, there is *vigorous* competition for > workloads that could or do run on mainframes.) The airline I flew had the > only nonstop flight, had a seat, and US$1300 was the price they set -- and > that they ended up receiving. > > It is true that the price of IFLs (and mainframe computing capacity in > general) has been steadily trending downward in both nominal and real > terms. When IBM reports "mainframe capacity shipments up XX% and mainframe > hardware revenues up YY%" where XX > YY then unit prices are obviously > declining -- that's just basic math. And that's good for both IBM and its > customers. > > -------------------------------------------------------------------------------------------------------- > Timothy Sipples > GMU VCT Architect Executive (Based in Singapore) > E-Mail: sipp...@sg.ibm.com > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN -- Mike A Schwab, Springfield IL USA Where do Forest Rangers go to get away from it all? ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN