Ah, but unless you implemented the CICS fences for the storage pieces, it must 
have been a bear when the apps started over-running memory.

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Roberts, John J
Sent: Monday, October 22, 2012 4:23 PM
To: [email protected]
Subject: Re: Donald Knuth and The Art of Computer Programming - was RE: Book 
Enquiry

Errr ???  I'm skeptical.  Suppose the address is 0, relative to the base of the 
managed storage pool.  What's its size?  Which is the buddy?

I may be a bit hazy on the details.  See 
http://en.wikipedia.org/wiki/Buddy_memory_allocation for something a bit more 
accurate.

What I am not hazy on is the performance boost we achieved by implementing a 
scheme with Knuth's algorithm at its core.  It cut the path length of a CICS 
GETMAIN or FREEMAIN from 2000+ instructions down to about 250.  And doubled the 
transactional capacity of our app.

John


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

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

Reply via email to