Server Products
Data Center Products including boards, integrated systems, Intel® Xeon® Processors, RAID Storage, and Intel® Xeon® Processors
4616 Discussions

Dissimilar CPU stepping/revision in one dual proc IBM 8853 Blade


Dear Expert.... we have several IBM 8853 Blade servers that have dissimilar CPU specs in the two CPUs installed. Here are the specs that are alike and those that differ:

FamilyModelSteppingRevisionCPU TypeEnhanced Halt StateExecute Disable BitLevel 2 cacheLevel 1 cacheLevel 1 instruction cachepchipset IDPROC 16F6C90YesNo4MB2 X 32KB2 X 32KB2670PROC 26FBB70YesNo4MB2 X 32KB2 X 32KB2670


and here is the system event log that brought this to my attention:

We have three servers with a similar issue. The only indication of any problem is in the system event log, and from what I have read here on the forums thus far, it doesn't pose an issue if just the revision is different, but no word on what the issues are if the Stepping is different. If you could answer this question, I would appreciate it. Thanks!

2 Replies
Valued Contributor III

This question would be better posied to IBM as the answer relates to how the BIOS configures the system

In general, the newer stepping of the CPU will contain fixes for errata as out lined in the processor spec update.

The BIOS likly contains work-arounds for these errata that may be impacting system preformance.

Based on the note in the event log, IBM's BIOS just treats both processors as if they were the older version and continues to load the work-arounds.

It is best to run with both processors at the same stepping. You may want to consider moving the processors so that the blades contain match pairs just to head off any potential issues.


Seems you're using two Xeon 5100 series processors, mixing B-2 (CPU ID 06F6) and G-0 stepping (CPU ID 06FB). On the processor side this is supported, but you may want to check with IBM if it's a validated/supported configuration on their server. Also see AG111 in Intel® Xeon® Processor 5100 Series Specification Update.

0 Kudos