On Thu, 16 Aug 2012 09:34:58 -0400, John Gilmore wrote: >Edward Jaffe wrote (about the use of particular padding bytes): > >| It can make a *HUGE* difference! > >When I made my measurements 13 months ago I did not find large >differences, and if EJ can find the time to elaborate a little on this >comment I will be very interested to learn what he has found.
I can't speak for Ed, but he might have been referring to the fact that large movement of full pages can be performed by the data mover hardware in memory without using cache. The difference in the speed of performing the move that way may not be so huge (I don't know what the difference might be), but the fact that it does not use cache can make a big difference on other workloads, whose data has not been displaced from the cache. These kinds of effects are difficult to test. -- Tom Marchant ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
