Jim Mulder had a superb analysis of this. Basically VSM goes through the DQE chain for 4K pages and bypasses it for 8K pages. In addition, with each GETMAIN of a 4K page, the DQE chain is one entry longer.
I do not try it with page boundaries and I did not try STORAGE.
Shmuel Metz (Seymour J.) wrote:
In <[EMAIL PROTECTED]>, on 05/11/2005 at 01:03 PM, Steve Arnett <[EMAIL PROTECTED]> said:
When bbbb is equal to 8192(8KiB), the job runs in less than one tenth
of a minute, using .03 minutes of CPU.
When bbbb is equal to 4096(4KiB), the job takes 5.4 minutes and uses 4.12 minutes of CPU.
Odd. Both the 4 KiB and the 8 KiB request should get full pages. Does the timing change if you explicitly request page boundaries?
BTW, what are the timings if you use STORAGE instead of GETMAIN?
---------------------------------------------------------------------- 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
