Retina MBP 15" aTnd Windows - some questions

I just bought myself a 15" rMBP with 2.6 GHz CPU - I am switching from Sony Vaio SVZ (Ivy Bridge Z2), and so far everything is more or less smooth - except one thing:

I installed Windows (Server 2012, but it is more or less Windows 8 for the discussion purposes) via BootCamp, so I am running under Apple's BIOS emulation (CSM), and the CPU is not entering turbo frequencies no matter what (it tops at 2.6 GHz, which is the maximum non-turbo P-state).

Now, the funny thing is - If i just start ThrottleStop, and then exit it, the P-state switching works as it should, with 4-core turbo going all the way up to 3.3 GHz.

Mind you, ThrottleStop is not even active, it just needs to be >started<.

I guess, what happens is that ThrottleStop does poke the IA32_PERF_CTRL MSR register, and Windows power manager or UEFI then starts using all P-states again.

Problem comes back after sleep/wake cycle or reset. So, I need to start and exit ThrottleStop every time after wake from standby.

--

So, my question is, did anybody else observe such unusual behavior? And, if yes, is there a fix?

Note, the problem was present before and after application of Apple's official EFI update - I tried updating EFI to fix the issue, but it did not help.

Problem also persists in pure UEFI mode - I booted Windows Installer in UEFI mode, and looking at HWInfo64, the CPU is also not entering turbo frequencies, indicating that the problem is not CSM bound.

 

will try to see if it happens here. but reset the scm

 

The lack of turbo boost might be another side effect of the EFI 1.0 update that's been giving RMBP users headaches over the last several weeks. No word from Apple about any fix.

Some Retina MacBook Pro users complain of graphics issues after EFI update

 

Hmm, it is not EFI 1.0 update as i did it precisely because I was getting this "turbo" problem.

I think I saw a post somewhere (can't remember which forum) where another person had the same issue - but he/she claimed that modifying the Windows Power Management settings (maximum CPU frequency) solved it - unfortunately, it did not work for me.

 

Is the clock frequency correct after a cold boot of the rMBP? In other words, does the issue only manifest after sleep/resume?

 

Nope, it is always capped at @2.6 GHz, even after cold boot.

This happens in Windows only.

 

There are a couple of registers than control whether Turbo Boost works or not. It might be bit[38] in register 0x1A0 that is set that shouldn't be set. Setting this one bit will disable Intel Turbo Boost.

MSR Tool
MSR.zip

http://i.imgur.com/PN8D9.png

You can play with my MSR Tool if you want to have a look. MSR Tool doesn't support the hyper threaded Core i CPUs but it might be useful enough.

Some bios versions do not set up IA32_MISC_ENABLE 0x1A0 and the IA32_PERF_CTRL register consistently. Even in monitoring mode, ThrottleStop initially makes sure that these 2 registers are in agreement. If a CPU has Turbo Boost available, I could not think of a reason why a user would not want to use that feature.

Maybe with the above info you can find someone to fix this problem so you don't need to use ThrottleStop. Maybe Turbo Boost has been deliberately disabled to reduce power consumption and heat. The majority of users would never notice that their CPU is not performing as Intel intended it to perform.