CPU sizing - Applied weight | Nutanix Community
Skip to main content

Hello, guys

Please, help me with these two points:

  1. I’m searching for the spec int value that the sizer uses. For example, I see that a default processor is E5 2680v2, 42.31 specInt. But I searched on the website of Spec.org and I couldn’t find that and the other serverals values that use Nutanix sizer, as is shown in the picture. Maybe I search on the incorrect section of Spec.org, 2006? 2017? Please help.

 

  1. I was wondering if the applied weight value is published in any website of Intel. I figured out that it is a fix value: 7.3 per processor for Xeon Gold 6244, for example. I wrote down the applied weights for 3,4,5 and 6-node configuration (dual-processor) and the calculations works this way:

Applied wieight in 3-node cluster: 43.8 (7.3x6)

Applied wieight in 4-node cluster: 54.8(7.3x8)

Applied wieight in 5-node cluster: 72.99(7.3x10)

Applied wieight in 6-node cluster: 87.59(7.3x12)

 

So I would like to know if these values (7.3 and others) are associated to the processor itself or is a value that Nutanix calculates based in another value shown in Spec.org or other metric given by Intel.

Sorry if maybe the questions are so tricky, but I feel so curious about the way that Nutanix works.

 

Thanks!

Hi Rolando0308

I am working on getting the answers to your questions and I will post a reply soon.


Great, ShvetaD! I appreciate your help.


Hi Rolando0308,  

According to the sizer guide below , we use SPECInt 2006 which is now retired and replaced by the SPEC CPU2017 suite.

https://sizer.nutanix.com/#/help/articles/155

“Applied weight” is a Nutanix concept, so you may not find a lot about it out on the web. It’s an attempt to allow a certain workload to be modeled on many different CPUs with varying clock speeds - not all CPU cores are created equally.

I would like to request you to please reach out to your Account team and have a conversation about this with the aligned SE who can help you understand this concept in detail.

 

Thanks

Shveta Dave