From:	CSBVAX::CSBVAX::MRGATE::"SMTP::CRVAX.SRI.COM::RELAY-INFO-VAX" 12-DEC-1988 00:03
To:	MRGATE::"ARISIA::EVERHART"
Subj:	Thoughts On Change Of Bit Definition For PRO Consoles Rev 8 And Up...


Received: From KL.SRI.COM by CRVAX.SRI.COM with TCP; Sun, 11 DEC 88 20:20:51 PDT
Received: from central.cis.upenn.edu by KL.SRI.COM with TCP; Sun, 11 Dec 88 20:00:21 PST
Received: from LINC.CIS.UPENN.EDU by central.cis.upenn.edu
	id AA25372; Sun, 11 Dec 88 22:58:39 EST
Received: from XRT.UPENN.EDU by linc.cis.upenn.edu
	id AA07348; Sun, 11 Dec 88 22:59:45 EST
Posted-Date: Sun, 11 Dec 88 23:00 EDT
Message-Id: <8812120359.AA07348@linc.cis.upenn.edu>
Date: Sun, 11 Dec 88 23:00 EDT
From: "Clayton, Paul D." <CLAYTON@xrt.upenn.edu>
Subject: Thoughts On Change Of Bit Definition For PRO Consoles Rev 8 And Up...
To: INFO-VAX@KL.SRI.COM
X-Vms-To: @INFOVAX,CLAYTON

During the same recent upgrade of VMS to 5.0-2, that caused the SID problem on 
the 8820, another interesting item came to light. 

I have had several 8XXX class systems in my checkered past, some of which use 
the PRO console as the boot medium and operator console. Until recently these 
consoles have been rather slow at booting the system, as some of my previous 
'The Saga Continues...' articles had detailed.

With the release of Rev. 8, the speed has once again taken a drastic step 
slower. This 'slowness' caused me considerable concern, that I had missed 
something somewhere, on both 4.7 and 5.0 of VMS to cause the delay.

After researching this item with several groups, it comes to light that a 
definition for a bit in R5, in the boot file on the PRO has changed. The bit 
is #7, and being set to a '0', which is the old typical value, now causes a 
FULL physical memory diagnostic to be run on the 8XXX processor BEFORE 
preceeding with the boot process. As the box being upgraded had 170+MB of 
memory, this amounted to about 10 minutes of delay in the process. 

If you want the old 'speed' back, then set bit 7 to a '1' and the memory check 
will not be performed. In other words set the TWO low order HEX digits to 
'80'. If you want a conversational bootstrap, then place a '81' in the TWO 
low order HEX digits. Note that this register has control of a considerable 
number of things, the biggest of which is the SYS root to boot into in the 
HIGH HEX digit.

You might ask what the drawback of this is. Needless to say, the memory is not 
checked to insure that it has no severe errors. Other then that, not much. 
Remember that the memory has ECC correction, to a certain degree, which can 
handle some errors. If its a total failure, you will know about it soon enough 
anyway. Do not get me started on the past, and current error logging rule of 
caching 16 memory errors BEFORE writing the information to disk. 

Anyway, if you suspect problems, it may be good to have different boot files, 
one with the memory check and one without. Or just leave it the way it is and 
accept the delay.

Your choice.

pdc

Paul D. Clayton 
Address - CLAYTON%XRT@RELAY.UPENN.EDU

Disclaimer:  All thoughts and statements here are my own and NOT those of my 
employer, and are also not based on, or contain, restricted information.