VMware Cloud Community
pjdf
Contributor
Contributor

VM guest CPU capacity comparison question

Hi,

I have a guest on a host Proliant BL685c G5 (4x Quad-Core AMD OpteronTM Processor 8354 (16 Cores) 2.1 GHz) and another guest on a host Proliant BL685c G1 (4x Dual-Core AMD OpteronTM Processor 8220 (8 Cores) 2.6 GHz). Both guests have 2 virtual CPUs allocated. If I make the assumption that every other guest on both hosts are virtually idle, which guest has the highest capacity from a CPU standpoint? In this case I would think that the higher class cpu host/guest combination (G5/8354) would only have 80% of the cpu capacity of the lower class cpu host/guest combination (G1/8220) from the guests perspective.

0 Kudos
6 Replies
AndreTheGiant
Immortal
Immortal

Remember that 1 vCPU = 1 core.

Probably the guest will work better on Processor 8220 (cause has more MHz per core).

Andre

Andrew | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
pjdf
Contributor
Contributor

Thats what I am thinking ... 5.2 total GHz as opposed to 4.2 total GHz ... thx. Not really a vm guy, a performance tester trying to explain to people why their DEV environment actually has more CPU capacity than STG and PRD from a guest perspective

0 Kudos
AndreTheGiant
Immortal
Immortal

If you have a CPU bound program than MHz could make a big difference (with similar CPU architecture).

Usual L1, L2 (, L3) cache could be more important.

Andre

Andrew | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
0 Kudos
pjdf
Contributor
Contributor

thx again ... I'll keep that in mind as well

0 Kudos
Ken_Cline
Champion
Champion

Remember that 1 vCPU = 1 core.

Probably the guest will work better on Processor 8220 (cause has more MHz per core).

That is potentially true; however, remember that a MHz doesn't equal a MHz! If you compare older generation clock speeds to current generation clock speeds, there is a distinct possibility that you're comparing apples to oranges. The CPU vendors don't change ONLY clock speed when introducting a new processor - there are lots of other architectural changes that get thrown in for good measure. Things like deeper pipelines, bigger caches, more parallelism, etc. All these things contribute to the "speed" of the chip (not to mention differences in the speed of the FSB and RAM, better interconnects, etc.)

So - I'd say it's impossible to answer your question with any degree of certainty other than by testing your specific workload on each system and see which is faster Smiley Happy

Ken Cline

VMware vExpert 2009

VMware Communities User Moderator

Blogging at: http://KensVirtualReality.wordpress.com/

Ken Cline VMware vExpert 2009 VMware Communities User Moderator Blogging at: http://KensVirtualReality.wordpress.com/
0 Kudos
AndreTheGiant
Immortal
Immortal

however, remember that a MHz doesn't equal a MHz!

Right, in fact I have write "with similar CPU architecture" Smiley Wink

And with similar I mean the similar number of pipelines, similar instructions, ...

Andre

Andrew | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
0 Kudos