[rescue] [Semi-OT] 700MHz Xeons in a Proliant 6400R?

[Marc Gutschner]

Curtis H. Wilbar Jr. schrieb:
> I have a ProLiant 1600.... tops out at 2xP3-600 Katmai.... even though
> it is a BX and I should be able to take it to 2x(fastest 100mhz Tualatin
> P3 at 100mhz meant for dual cpu operation) with the right slot1 to
> socket 370 converters.
Do you know for sure that an "older" BIOS will allow you to upgrade to 
more recent P3s? Maybe the (still active) COMPAQ FTP site has an older 
Firmware image. Although I don't know if it's worth your time to plough 
through an 9 MB ASCII file to look up the "SoftPaq" number :-)
> I'd be interested to hear if this works. I have a 2xP3 Xeon box
> that I upgraded to the fastest proc supported (with 2M cache
> each). It's not a compaq, but a Dell. I did not try going beyond
> the rated mhz becase I saw the voltage changed.
That was a valuable hint - but to my dismay it led to the conclusion 
that my "flavour" of 700 MHz Xeons will not work in a ProLiant 6400R. 
According to the information in the "Pentium III Xeon Processor System 
Compatibility Guide" 
(http://download.intel.com/design/PentiumIII/xeon/applnots/24524803.pdf) 
newer Xeons (i.e. clock rates beyond 550 MHz) differ sufficently from 
the older ones. Quote from said document:

<begin quote>
7.0 Core Voltage Identification and HV_EN# Pin

The 2.8V Pentium. III Xeon processor uses the legacy VID scheme as per 
the VRM 8.3 DCDC Converter Design Guidelines (please refer to the 
processor datasheet for further details). The expected operating voltage 
at the input of the OCVR for the 2.8V Pentium III Xeon processor is 2.8 
V. Therefore, 2.8V Pentium III Xeon processor VID_CORE[4:0] inputs will 
encode the value for 2.8 V, VID[4:0] = 10111. The 5/12V Pentium III Xeon 
processor will encode the value for No Core, VID[4:0] = 11111 since 
this processor type is meant to eliminate the need for VRMs.
<end quote>

Further on it says that "proper baseboard logic" will "disable the 5/12V 
Pentium III Xeon" if 5/12V are not available. So I guess that's what 
happens here - if the 700 MHz bricks are plugged in the systems stalls 
during very early stages of POST...

*Sigh* So I'll stick with the 500 MHz Xeons to have a "SNMP baby" ;-) 
Lesson learned: *NEVER* let a cat near any piece of hardware that you 
want to keep. Cats (esp. males) have a vary nasty habit of peeing on 
anything they don't like!

regards,
Marc