ABO is going to make a V3/V4 program faster, in general, because it has access 
to ARCH(up-to-10), in your case, instructions which the V3/V4 compiler does not 
have access to.

It is also able to use some optimisation techniques not available with OPT in 
V3/V4 (learning for the optimisations in V5, which in turn learned from 
optimisations for Java).

A program may be "flat" (before = after) but that would seem fairly unlikely, 
and any such program would likely not be a CPU heavy-hitter.

Caveat. It would be possible that the ABO has initial "overhead" larger than 
the original program that affects CALLed programs. I don't know. Not seen it. 
Just sayin'.

The ABO people are very confident that CPU savings will pay for the product. 
Not considered in that is costs to get the ABO'ing up and running.

I think there is a role for your type of program, done in a different way, in 
establishing some of the things which will reduce CPU, perhaps depending on how 
you are going to implement ABO.

If you are going to ABO everything in one shot (or similar), then there is less 
need. If you want targeted implementation, then you can do something very 
useful (establish the most likely target-programs).

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

Reply via email to