-------------------------<snip>------------------------------
Could you elaborate on that? I understand that there is a savings if
reentrant code can be reused without reloading (such as if it is
resident in the LPA) but why would reentrant code be inherently faster
than non-reentrant? There is certainly an additional overhead for
GETMAIN, storage initialization, often an extra level of indirection,
etc. Reentrant design often burns one more register, which may in turn
lead to additional register save and restore operations. Reentrant code
is typically more scattered in its storage references, which increases
paging overhead (at least in theory).
It's an academic question, I admit. 99% of all assembler code is so fast
it does not matter, and the 1% that matters can always be optimized
despite any considerations for reentrance. I'm just curious about your
assertion.
-----------------------<unsnip>------------------------------------
It's a function of how the instruction fetch and data fetch caches are
used inside the processor.
----------------------------------------------------------------------
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