A correction:  You can allocate any size of memory above the bar; for small
areas (up to 64KB, iirc) you'd use IARST64.  This is a version of cell pool
services that handles the set-up and management of the cell pool for you.
So it should be much more efficient for getting & freeing lots of pieces of
storage than using almost anything else, especially considering both
programming time & run time.

Charles is right in the sense that VSM (or whatever its a-t-b analogue is)
allocates memory in 1mb chunks, and with IARV64, the low-level memory
service, that is the minimum request size.

For some reason, there is yet another macro (IARCP64) for mid-sized chunks
(64K-1mb), that is also cell-pool based.

IBM might have made this simpler, but once you get going, it's pretty easy.

sas

On Tue, May 18, 2021 at 9:19 AM Charles Mills <[email protected]> wrote:

> ...
> +1 to what Timothy says about bytes: particularly if an application can
> exploit above the 2GB bar storage, bytes have quit being something to worry
> about. Heck, the above-the-bar "GETMAIN" functions only work in increments
> of a megabyte -- you can't allocate 10K or 100K. Everyone still worries
> very
> much about CPU cycles, so if you can trade a couple of megabytes of storage
> allocated for a couple of seconds of CPU time saved it is well worth it.
>
>

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to