Re: Paging file

["Mark Waterbury" <mark.s.waterbury@xxxxxxxxxxxxxxxx>]

Hi, Steve:

I am not trying to "pick on you" here, I just want to clear up some
misconceptions, for the benefit of everyone on the list.

Regarding your comment (excerpt below):

> ...  The expansion of the page size is what caused objects to
> increase dramatically in size when going from V3R2 to V3R6.
> ...

I think this idea is based on some false assumptions.

Simply changing the page size does not generally affect the actual
object size; if an object required 128 512-byte pages on CISC, it
would now require only 16 4096 byte pages on RISC. Do the math;
the total object size remains the same.  Each object could grow at
most by just under 4096 bytes, due to "rounding up" to the nearest
4K, for the last few bytes of any object.

It was the conversion from CISC hardware to RISC that required
conversion from an IMPI (CISC) instruction set to PowerPC based
RISC instructions, and this is precisely the difference between CISC
and RISC, that is, the instructions are simpler, and so, you need
more of them, to do the equivalent job.  So, for example, where you
had a single instruction of 2, 4 or 6 bytes in the IMPI machine could
now require a sequence of several RISC instructions, all of which are
4-bytes each.  This is what caused the huge "growth" in object sizes
seen during the CISC-to-RISC conversions, V3R1 to V3R6, or
V3R2 to V3R7, etc.  Typically, *PGM objects doubled in size,
on average, due to this expansion of the "encapsulated" portion of
*PGM objects, which contains the actual instruction stream, in
addition to the MI program creation template (the virtual instructions
for the MI virtual machine).

I hope this helps to clear up some of these misconceptions.

Regards,

Mark