Re: % temp addresses = 96.800

["Leif Svalgaard" <leif@xxxxxxxx>]

From: <D.BALE@handleman.com>
>
> So, I guess we'll do an IPL this evening.  Can anybody tell me what causes
the
> temp addresses to get this high?  BTW, I understand that this problem goes
> away with RISC, and we are close to upgrading to new boxes.
>

The S/38 supported a 64-bit address space with a 48-bit hardware address
(like putting a 10 pound load in a 5-pound sack) by treating the upper 16
bits of the 64-byte address in a special way. These bits were called the
SID extender. For the value to put there they used the IPL number.
Every time the system was IPLed this number was incremented by 1.
That gave them a new 48-bit address space at each IPL. Furthermore,
the design called for not reusing the 48-bit addresses except at an
IPL boundary. The S/38 was designed to be IPLed once a day. At
that rate it would last 200 years before the IPL number rolled around.
The top two bits of the 48-bit address was a "quadrant" number.
The first quadrant was reserved for permanent objects. The next
quadrant was used for temporary addresses. The third was
used for temporary objects in access groups, while the last quadrant
was not used (thus belying IBM's claim of 256TB of virtual memory,
when it was only 192TB). Now, a segment is 16Mb and there are
therefore only 4 million temporary segments available per IPL.
So when you have used that many temporary addresses (and the
system uses these up internally a lot faster than you do) you are
out of temporary addresses and must IPL. That simple ! ! ! :-)



+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---