Testers show MacBook Pro Turbo Boost oddness
It works! Oh no it doesn't! Oh yes it does! etc, etc.
Some unexpected benchmark results may show how Apple is implementing - or not - a key feature of Intel's second-generation Core i platform, 'Sandy Bridge'.
Tests performed by dead-tree title PC Pro show that Sandy Bridge's Turbo Boost technology, which dynamically overclocks one, two or four cores based if the CPU load could use it but holds back if the thermals show this would be a bad idea, isn't always enabled.
The mag tested Core i5- and Core i7-based - both two-core chips - 13in MacBook Pros running Windows 7 with Apple's Bootcamp multi-OS system, and found that Turbo Boost works on the former, but doesn't on the latter.
Website Anandtech provides corroborating evidence that the Core i5-based 13in MBP uses Turbo Boost when possible, in both Windows 7 and Mac OS X.
The odd thing: the Core i7 Turbos under Mac OS X too, so it's only under Windows that the feature is seemingly disabled.
Conspiracy seems unlikely - if Apple was trying to ensure Windows performed less well on these machines, it would surely disable Turbo Boost on the Core i5 MBP too.
In any case, delivering good Windows performance is a sensible move for Apple: it helps convince PC owners to migrate to Mac by showing they won't lose horsepower if they do.
If the reason was related to system thermals, you'd expect the quad-core MBP to stop Turbo Boost in both OS X and Windows 7, not just the latter.
The 15in MBP, all versions of which use the Core i7 - and a quad-core one, to boot - and there's no reduction in Turbo Boost performance, Anandtech's tests show.
The question is, is PC Pro's odd result the sign of a one-off machine problem, an issue with that particular CPU - the dual-core 2.7GHz Intel Core i7-2620M - or a glitch in the Core i7 family?
Recent reports of 15in and 17in MBP freezes under heavy CPU and/or GPU Load show some machines are thermally challenged - though where through excessive use of thermal paste, problematic logic boards or some other reason, isn't yet clear. ®
I suspect it's simple.
Windows is more multicore friendly and keeps all four occupied. (You can read this as "Windows has loads of unnecessary cruft running in the background" if you prefer).
Using Windows on the hotter i7, it doesn't get the thermal leeway to boost with all four cores in use. MacOS keeps the thread count down, allowing it to turn a core or two off so it can.
That's my guess.
Has anyone considered that it could simply be a problem with windows, either a software bug or simply a case of the OS being more bloated and putting enough strain on the cpu that it raises the temperature enough to cause turbo boost to become disabled?
Has anyone tried benchmarking it under linux, where it would be easier to see at a lower level exactly whats going on?
Even a VM shouldn't need to do anything much when the Guest is idle, so I don't see what you are getting at.
Do you have a Mac OS VM to compare to? The load that you see is likely the overhead that is reuired to run a VM. I see pretty much the same load on my ESXi box for each OS, while it's doing nothing.
Bootcamp doesn'T actually do anything..
PC-Style booting is part of the EFI firmware.. Bootcamp is a GUI for the partition resizer.