RP4440 CPU version

[EMCD]

We have an RP4440 server, which we configured with two CPUs (1Ghz Dual core).
When booted to BCH there are warnings:

WARNING: The system contains an incompatible mix of processors. Processor is booting from the first available device.

CPU's look identical apart from the CVERSION
PA 8900 CPU Module 3.2
PA 8900 CPU Module 3.2
PA 8900 CPU Module 4.1
PA 8900 CPU Module 4.1

Does CVERSION point to FW? Will a FW update resolve this issue?

Thanks

[cnb]

Hi Ewan,

Can you post the system revision information from the MP?

"Determining Current Revision
Establish a telnet session with the Management Processor.
Logon to the Management Processor using the Admin password.
Type CM to enter the Command Menu.
Type SYSREV to display system revision information."

You may want to verify that you have the latest critical code fixes:
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=15351&prodSeriesId=2512015&prodNameId=401769&swEnvOID=7&swLang=13&mode=2&taskId=135&swItem=pf-57813-1

[EMCD]

I don't have access to the server at the moment, I've been searching google, but can't find what the "cversion" relates to?
Can you explain?

Cheers,

[david d_2 · Accepted solution]

Ewan,

No it does not have anything to do with the FW version. It is the Chip Version . When there is a change to the core the rev the chip. Have you tried running with just one processor at a time to verify they are both good?

-Dave

[EMCD]

Yup, the 3.2's work on their own and the 4.1's also work on their own. It's mixing the two that cause the problem.

[Stefan Stechemesser]

Hi,

CVERSION means the internal chip version of the dual core PARisc CPU.
You should be able to mix different CVERSIONs.
You cannot mix PA8900 and PA8800 CPUs (which may have the same PARisc chips but different L3 cache sizes).
PA8900 and PA8800 CPUs have identical CPU Cores, but PA8900 has 64 MB L3 cache, PA8800 only 32 MB.
You can check what you have installed with the "df" command on the MP (CPU PIROM).
Here an example of a PA8900 CPU:
FRU info N/AFRU NAME: CPU 0 PIROM ID:20

PROCESSOR DATA
S-spec/QDF: 9.13
Sample/Prod: 03

CORE DATA
Arch Revision : 127
Core Family : 240
Core Model : 88
Core Stepping : 50
Max Core Frequency : 1000 MHZ <===
Max SysBus Frequency : 200 MHZ
Core Voltage : 1300 mV
Core Voltage Tolerance,High : 65 mV
Core Voltage Tolerance,Low : 65 mV

CACHE DATA
Cache Size : 65536 KB <=== PA8800 has only 32 MB !

PACKAGE DATA
Package Revision : A3
Substrate Revision: 00

PROC PART NUMBER DATA
Part Number : PA8900 <===
Electronic Signature : 07013F646D8FC444

The value that is desplayed in the BCH information menu is not the "real" CPU model, but that what the firmware expects based on the model type ("HVERSION").
Make sure that you have really only PA8900 CPUs with 1GHz Core frequency. The chip revision does not matter.

I hope this helps

Stefan

[Sameer_Nirmal]

Most likely the processors are not really identical in terms of speed. Maybe one of them is 800MHz and other is 1GHz. Mixing of these 2 type of CPU on a system is not supported and that's why they are not working together at the same time. Verify the CPU speed using MP>DF as mentioned by Stefan for each of them in place. Also check the HVERSION using BCH>IN PR.

http://h18000.www1.hp.com/products/quickspecs/12034_div/12034_div.HTML

[EMCD]

Hi, I've attached a txt file with all the CPU information.

There is also an error pointing to CPU mismatch, but apart from the C.version, these are identical.
The product numbers are the same, they look identical and they show up as the same CPU speed.

Thanks for any further help.

Regards

[Stefan Stechemesser]

Hi Ewan,

you information does not contain the "df" output from the MP.

The CPUs in your output have different cache size:
3.2-Module: 67108864
4.2-Module: 786432

Hm, the first is clearly a PA8900 module with 64 MB L3 cache, but strange is the 2nd.
Only the L2 cache is displayed here, maybe the firmware does not handle the current mismatch correctly.
Either this is a PA8800 CPU or else it is simply defect.
Please check the info from the MP command "df -nc -all"

best regards

Stefan

[EMCD]

Thanks for the info