Tom Marchant's benchmark test disclosed, unsurprisingly but conclusively,
that there are very significant overheads associated with repeated AMODE
changes.
One moral is obvious: Avoid AMODE changes in loops like that which figures,
legitimately, in this test.
In such situations my own practice is to make sometimes extended
"unnecessary" use of AMODE(64), which is anyway faster, in situations in
which not doing so would entail many changes into and back out of it.
Intuition can fail anyone. What seems obvious to me may be quite wrong.
Still, I have found this thread disconcerting. That AMODE changes must
necessarily have significant overheads associated with them is, I still
think, obvious.
John Gilmore
Ashland, MA 01721-1817
USA
_________________________________________________________________
PC Magazines 2007 editors choice for best Web mailaward-winning Windows
Live Hotmail.
http://imagine-windowslive.com/hotmail/?locale=en-us&ocid=TXT_TAGHM_migration_HM_mini_pcmag_0507
----------------------------------------------------------------------
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
