I have a late 2014 Mac Mini, with the i7 option, 3 Ghz turbo boosting to 3.5 Ghz according to Apple. To give the system some work to do I am running a ray trace:
$ time povray --benchmark +WT1
This should max out one core with some intense floating point calculation. But when I check as it runs:
$ sysctl hw.cpufrequency
hw.cpufrequency: 3000000000
Why isn't Turbo Boost kicking in? The machine is otherwise idle apart from the Safari window in which I am typing this now, this is the scenario as I understand it that Turbo Boost was designed for?
Just for interest, this machine completes this benchmark in 13m25s and our iMac 3.2Ghz i5 completes in 12m58s – I'm not seeing that paying the premium for i7 was worth it 🙁
Update: Well having done some more research, I am getting some boost, according to the Intel Power Gadget:
But I still can't match a 3.2 Ghz i5.
Best Answer
When you run the command
sysctl hw.cpufrequencyyou are getting a constant that was given to the OS at startup. If you want to get the CPU frequency "on-the-fly" you will need to poll the CPU. That's what the Intel Power Gadget is doing.As for why you are not seeing a performance increase with your specific application, it could be from any number of factors including whether or not the application is optimized to work on this particular architecture.
Disclaimer: Personally, I don't know much about ray tracing software, but from the little bit that I did read up on. I don't see why the calculations for the creation of the trace wouldn't be done on the GPU and not offloaded from the CPU. This makes me wonder if this is not what's happening here
My suggestion here is to use something that is consistent across all platforms like Geekbench
Finally, Intel has this disclaimer about the Turbo Boost technology...
and...
So, as you stated it appears that your CPU is "boosting" but as to why it's not getting up to the full 3.5 could depend on any/all of these factors that Intel noted.