blazinsmokey Posted July 28, 2015 Share Posted July 28, 2015 Hardware* Asus Maximus VII Hero (Z97)* i7- 4790K (Manually clocked at 4.4GHZ @ 1.1V)* G.Skill Sniper 16GB 1866MHZ* MSI GTX 970 4G Gaming Summary After a post on reddit a while back about which SMBIOS people are using I got into a discussion about my setup and using iMac 15,1 instead of MacPro 5,1. In short apparently it doesn't even really matter at least from my testing. Below you will see I tested on the two mentioned and MacPro 6,1 for added {censored} and giggles. I passed on 14,2 because I initially started with it but it was missing P-States that are again missing in 15,1. Surprise! I am now using MacPro 6,1. I wanted to post here with what I did and get feedback and thoughts from this great community. I generated my SSDT with Pikers tool, ssdtPRGen. The same SSDT is used in all 3 SMBIOS. Everything remained the same on my board and setup. All USB, sound, wake/sleep functions did not change and worked. For graphics to work though and not to boot into a black screen I had to edit deep inside AppleGraphicsControl.kext to make it work on iMac 15,1 and MacPro 6,1. I made both adjustments and no ill effects. You can see only the MacPro takes advantage of additional P-States 41-43. iMac does not use these. I assume in reality that means very little but ideally and this may be my overclockers mentality kick why would you want to use more when you can use less if it's available. Another thing I noticed (placebo effect?) is that using as iMac 15,1 and MacPro 6,1 is that the UI seemed a bit snappier in real time use. The benchmarks reveal the SMBIOS doesn't mean much between my hardware and these options. ScreenshotsGeekbenchCinebench iMac 15,1 Console/AICPUPMI All C-States recognized P-States not recognized (9-11) DPCIManager P States used (8,40,44) Geekbench 3.3.4 (64 bit) Single 4517 Multi 17985 Cinebench R15 OpenGL 113.62fps CPU 884 cb MacPro5,1Console/AICPUPMI All C-States and P-States recognized DPCIManager P States used (8,40,41,42,43,44) Geekbench 3.3.4 (64 bit) Single 4529 Multi 17939 Cinebench R15 OpenGL 114.35 fps CPU 880 cb MacPro 6,1 Console/AICPUPMI All C-States and P-States recognized DPCIManager P States used (8,40,41,42,43,44) Geekbench 3.3.4 (64 bit) Single 4542 Multi 17931 Cinebench R15 OpenGL 114.04 fps CPU 883 cb Edit: I totally forgot to mention. Switching back and forth between SMBIOS is really easy with Clover. The additional plists are saved in the same location as the original but named config0.plist and config1.plist. It is as easy as going into options in the boot screen, then add a 0 or 1 to the first option which will already say config. Very handy if you guys are going to test around and try to find your best compatible SMBIOS. Also make sure you sign out of iCloud, Messages, Appstore, and Facetime when your switching back and forth. Link to comment Share on other sites More sharing options...
OffCornerDev Posted July 28, 2015 Share Posted July 28, 2015 Hi, I am using SMBIOS iMac 15,1 right now with the same processor but AICPUPMI only recognises p-states 8-40-41-42-43-44 and nothing in between. Do you think there's a way to solve this? Link to comment Share on other sites More sharing options...
MacUser2525 Posted July 28, 2015 Share Posted July 28, 2015 Hi, I am using SMBIOS iMac 15,1 right now with the same processor but AICPUPMI only recognises p-states 8-40-41-42-43-44 and nothing in between. Do you think there's a way to solve this? No Apple does an all or nothing approach to cpu power you get the lowest state and the high ones with nothing in between. Well that is unless you make your own SSDT then you will see other states being used in the HWMonitor.app, but I am not sure if that is nothing but cosmetic as if you have the DCPIManager p-state option open at the same time they do not show up in it. BTW interesting to see that system definition used is effectively useless for processor power of the cpu. Link to comment Share on other sites More sharing options...
OffCornerDev Posted July 29, 2015 Share Posted July 29, 2015 Seems like I got the p-states working. I made an SSDT before but once I downloaded the latest AppleIntelCPUPowerManagementInfo.kext the p-states started showing up. My Hackintosh is running perfectly, except for sleep. Sometimes it reboots while it's sleeping. I looked on the forums and it seems to be a common issue with my setup but also on native macs... The System definition can change the way that the computer manages the p-states. I think the iMac 15,1 definition is not really using many intermediate p-states. Will try some other ones as well. Thanks for your quick reply though! Link to comment Share on other sites More sharing options...
toleda Posted July 29, 2015 Share Posted July 29, 2015 No Apple does an all or nothing approach to cpu power you get the lowest state and the high ones with nothing in between. Not true. DPCIManager, MSRDumper and HWMonitor do not report Ivy Bridge or Haswell pstates properly. For accurate Pstates, use Intel® Power Gadget. In short apparently it doesn't even really matter at least from my testing. From the information provided, power management may not be properly configured in any test case. Link to comment Share on other sites More sharing options...
blazinsmokey Posted July 30, 2015 Author Share Posted July 30, 2015 Not true. DPCIManager, MSRDumper and HWMonitor do not report Ivy Bridge or Haswell pstates properly. For accurate Pstates, use Intel® Power Gadget. From the information provided, power management may not be properly configured in any test case. So Piker's AICPUPMI is inaccurate? I do use Intel Power Gadget and it does seem to be working correctly. Some people said Intel Power Gadget is not accurate b/c it's averaging the frequency of all cores. Please enlighten me. I will provide proof or what is necessary to see if I am incorrectly doing something. I'd rather be called a idiot than spread false information. Here's my Intel Power Gadget screen shot. This is on MacPro 6,1. I recently went into testing some overclocks and to my surprise Clover generates it just fine, so I am currently running SSDT free. Link to comment Share on other sites More sharing options...
toleda Posted August 2, 2015 Share Posted August 2, 2015 So Piker's AICPUPMI is inaccurate? Some people said Intel Power Gadget is not accurate b/c it's averaging the frequency of all cores. No AICPUPMI presented in Post #1 Power Gadget, most accurate tool available for OS X. Attach IOReg, AICPUPMI output and config.plist. Link to comment Share on other sites More sharing options...
Recommended Posts