ManHands' Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

[manHands]

skirk08's Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

I've used all types of programs, MSRDumper.kext, ACPIMI, DCPIManager, HWMonitor all the same issue. Like ACPIMI "shows" the other P-states but for some reason it never goes to those states.

Have you looked through this thread? What does IPG look like?

"What changed between Mountain Lion and Mavericks power management? The existing processor frequency/power state reporting tools (i.e., MSRDumper, HWMonitor) stopped reporting power states between idle and max non turbo. This led to an incorrect conclusion that native Mavericks power management was not working. A new tool, Intel Power Gadget, shows Mavericks power management is working as designed."

 

[keenerz]

skirk08's Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

yeah Ive been looking at it. the IA jumps up and down but the logs point to it jumping between 800 and 4000 at idle mainly. I dont know if this is now how haswell works or am I just not getting intermediate P-States, I'm asking Ambelin cause he has the exact same setup mobo and cpu wise to see if I'm the only once getting that (I wish we had Private Messages for this). I dont know if anyone else gets that same exact states.

I mean its not a game breaker everything else on this computer is working great, I thought I'd try to polish it by getting the extra p-states working. but I just want to know so I can stop this crazy perfectionist ambition of mine

 

[Amblin]

skirk08's Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

Hi,

I've only just seen these messages (and in fact just woken up!) I may well be half a world away from you.

I'll be back in a while and will leave the IPG running a while whilst I work and post my graph.

I am fairly sure that if you look closely when the SSDT creation script is running you can see all the states being generated (I may be wrong and am yet to have a coffee this morning) and in fact I thought I saw them when I read the script before running it, Hmmmm. Too early to see, much less read a script !

Do not believe HW Monitor though, I'm sure you know that, It's very out of whack on my machine and often looks as though 800 and x46 or 17.9Thz are the only states ( ) but it's wrong.

I did post a picture in this thread somewhere in the this thread with an IGP graph on it that appeared to suggest proper steps were in effect as I did various things. I havent re-checked since I OC'd though as I have been working on my vcore. I will have a look for you later, and report back.

Good luck for now,

Amblin

 

[keenerz]

skirk08's Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

yeah I saw the script making the P-States and also I saw them show up in AppleIntelCPUpowerManagementInfo.kext however I never see the the p-states ever being selected. So I dont know if its just how haswell does things or if I'm just not getting the full spectrum of p-states. Maybe haswell does interpolate between 8 and 40 to simulate intermediate states. I would love to know but no one seems to answer the question directly.

I do appreciate you taking the time and helping me figure this out.

 

[Amblin]

skirk08's Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

Ok so I could not get IPG to log to a file no matter what I tried, even setting a different path for it to log to and then rebooting....Which is odd.

Sorry I cannot post a log, only a pic from earlier today. Now I take this to show that multiple steps are being used. Whilst not in this particular image, if I benchmark, it flatlines at 4.6 before resorting to the kind of behaviour while opening apps and playing music, browsing and so on as shown in the picture below which indicates to me, multipliers are being applied correctly dependent on load.

I hope someone with more experience can confirm this is correct P (or C?) state useage, I have no P state errors in my console (now !) nor any issues with my build at all other than an IPG log!

It's certainly not using only 800mhz and 4.6ghz. I'll test completely idle for you tomorrow as I have to off!

Also, I may be well wrong here, but might the SMBios impact this somehow? I'm sure it used to on older chips but then as Skirk said Mavericks changed PM....

Cheers

 

[keenerz]

skirk08's Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

Ok so I could not get IPG to log to a file no matter what I tried, even setting a different path for it to log to and then rebooting....Which is odd.

I had the same issue I ended up going to preferences and changing the log placement to the desktop and apparently that worked for some reason. So if you want to give it a shot that might work out. See the thing is mine does the same thing where it'll go up and down on IPG but apparently the graphs and numbers on IA are "averages" from some other thread I was reading where the logs show actual numbers. See if that desktop thing works for you. Thanks again for helping me solve this issue.

 

[manHands]

skirk08's Haswell Build: i7-4770K - GA-Z87X-UD5H - 32GB RAM - GTX 760

yeah I saw the script making the P-States and also I saw them show up in AppleIntelCPUpowerManagementInfo.kext however I never see the the p-states ever being selected. So I dont know if its just how haswell does things or if I'm just not getting the full spectrum of p-states. Maybe haswell does interpolate between 8 and 40 to simulate intermediate states. I would love to know but no one seems to answer the question directly.

You may need to head over to the Mavericks: Native CPU/IGPU Power Management thread to get your questions answered directly from toleda. I know Amblin has a 4790K. However I don't, and PM works perfectly with this board and the 4770K as you can see documented here. So my recommendation would be to get some assistance over in the other thread where there may be more traffic, and possibly a direct response from toleda. There's a lot of good info Pike's site too, including his latest PM post.

 

[keenerz]

yeah sorry didnt mean to hijack your thread, just trying to solve this issue. And yeah I've been to that thread and toleda hasn't been quite the most helpful person especially with something that doesn't seem like anything wrong. Also that pikeralpha last post was great. I've been on his site before and never came across that. Thanks!

 

[ChesBiscuit]

Hey,

Just followed your guide to update my build from 10.9.1 - 10.9.4 - all working smoothly. Have PM working, and system seems stable.

One query - I was previously on a macpro system definition and now over to iMac 14.2 as recommended. The downside seems to be that I get audio popping as OSX puts the audio driver to sleep. Do you have this issue? If so, what did you do to fix it? The macpro definitions don't get it AFAIK as they don't sleep the audio driver.

 

[manHands]

One query - I was previously on a macpro system definition and now over to iMac 14.2 as recommended. The downside seems to be that I get audio popping as OSX puts the audio driver to sleep. Do you have this issue? If so, what did you do to fix it? The macpro definitions don't get it AFAIK as they don't sleep the audio driver.

My bad, I'll update the OP. Some things have been known for so long, I forget to include them. In reviewing the original documentation, there are a few different ways around this. Download the M-Realtek ALC AppleHDA Capabilities.pdf and search for Audio Popping Solutions. I simply just use the black port, and make the change in System preferences.

This is all courtesy of toleda in his Mavericks: Audio - Realtek ALC AppleHDA [Guide].