Turbo Boost in VMware ESXI? Does it work?

J

John_TX

n00b
Joined
Sep 17, 2009
Messages
19
Has anyone been able to verify whether turbo boost actually works and is beneficial on ESXi? Given a moderately loaded ESXi host (less than 40% CPU utilization), will the Xeon E5-2600 v3 series chips actually raise their clock speed similar to how turbo boost works on in Windows on a desktop?
 
Last edited:
This is a weird question. Turbo boost isn't an OS specific feature. Are you experiencing something unexpected?
 
John_TX said:
Has anyone been able to verify whether turbo boost actually works and is beneficial on ESXi? Given a moderately loaded ESXi host (less than 40% CPU utilization), will the Xeon E5-2600 v3 series chips actually raise their clock speed similar to how turbo boost works on in Windows on a desktop?
Click to expand...

Yeah it does work. Though it won't benefit ESXi if other people are accessing VMs on the same host. Turbo boost speeds up a single core at the expense of other cores (afaik). or let's say 2 cores at the expense of 6 if you have an 8core processor
If you're the only user on the host and the other VMs are idle. I'd say leave it enabled.
Also disabled the CPU power management features on the server's bios as the kernel takes care of that. That should also speed things up a bit
 
OP: You really shouldn't cross / double post. Your question was answered in the other subforum that you posted in already.
 
Sorry for the double post. This seemed more like a processor question after I initially posted.

I'm trying to determine whether the E5-2630v3 or E5-2618Lv3 would be better for running ESXi.
E5-2630v3 [8 core, 2.4ghz (3.2ghz max turbo 2 cores), 20MB, 85w]
E5-2618Lv3 [8 core, 2.3ghz (3.4ghz max turbo 2 cores), 20MB, 75w]

The 2618Lv3 has a slightly lower base clock-rate, but also turbo's higher than the 2630v3, all the while using 10w less power at max TDP. Any thoughts?
 
Depending on the workload your system is not always going to be running at the highest clock rate.

I think the performance difference is negligible. You would pick the low voltage CPU if that was the prerogative of your data center (low voltage). I think the LV chip cost quite a bit more and the savings are only a few dollars per year.
 
The price difference is actually only about $35. Irrelevant given the total cost of E5 systems IMHO. As a general rule, I count any system power savings and actually double it to account for PSU inefficiencies as well as decreases in needed cooling capacity. I would think the low voltage chip would also be from a higher quality bin since it can basically hit the same speeds with 12% less amperage/voltage.
 
John_TX said:
The price difference is actually only about $35. Irrelevant given the total cost of E5 systems IMHO. As a general rule, I count any system power savings and actually double it to account for PSU inefficiencies as well as decreases in needed cooling capacity. I would think the low voltage chip would also be from a higher quality bin since it can basically hit the same speeds with 12% less amperage/voltage.
Click to expand...

makes perfect sense to me if you're building a data center
 
John_TX said:
The price difference is actually only about $35. Irrelevant given the total cost of E5 systems IMHO. As a general rule, I count any system power savings and actually double it to account for PSU inefficiencies as well as decreases in needed cooling capacity. I would think the low voltage chip would also be from a higher quality bin since it can basically hit the same speeds with 12% less amperage/voltage.
Click to expand...
Sounds like you have came to a conclusion all on your own.
 
You must log in or register to reply here.
Back
Top