Rated CPU speed is what the CPU can do at 100% load - ie 2820qm can do 2.3GHz when under full load. Mine runs at normally 2.7GHz when rendering movies (renders take 30+ minutes with all 8 threads at 100%, CPU speed at start of render at 3.4).
If your CPU can't maintain its rated maximum there is a problem with your system - software the most likely culprit. About the only hardware isues would be lack of heat transfer compound, or running with the lid shut - open it, the MBP is not designed to run with it shut.
I use the latest 'hacked' CCC & AMD driver for the 6750, and have used aTI tray tools to overclock the GPU 20% with no CPU slowdown past 2.3GHz when playing Cliffs of Dover, Driver 3, Skyrim etc.
I'm doing some renders on the weekend - I'll get some screenshots then, these were for some other thread somewhere about stock GPU speeds:
I don't post opinion, thanks. Good luck with fixing your problem.
Software aint an issue on my end, i've used over 5 clockrate software to cross verify the values.
They all are showing the same information.
how did you take these screenshots while playing your game?
Did you alt-tab to take them?
If you did then their basically junk, just looking over them they must be alttabbed.
I can see your CPU is on a small load (likely alt tabbed load with game running in background) you need to see CURRENT values, not the max values, thats why dual screens or alternative method is needed. I can see your GPU is idle clocked too, your not testing your system properly for the current readins the max values wont show you a cpu throttle.
All of those screenshots are showing idle GPU load clocks, thus the idle clockrates so the game window couldnt of even been running windowed mode either.
Your showing a 10%-5% cpu core load only, which is a joke of course it will run at rated speeds wheres your gpu load too? Its just idle clocks at current values.
Run a game like BF3 or stress both your CPU and GPU with OCCT tool and prime 95, then take a screenshot.
I can tell you right now I know you did not take those ingame, you alt-tabbed its not a stressed system. Minor load is all im seeing.
We need to record the data while ingame, not while alt tabbed.
If you dont have dual screens to view this, then windows can suffice (but poor testing this method)
Here is what we can do as a workaround if you have only one monitor...
You should run Furmark (at highest settings/res/aa) and Prime 95 at max (8 threads, max stress).
Furmark to 100% gpu load (all modern games will max this GPU transistor count).
Prime 95 can be 100% but realistically the most stressful games will put 50-75% load on the cpu (ie bf3).
Then screenshot your 'CURRENT' values after 15-30min.
Take a screenshot when you have those both running, i'd put money that you get throttled (dont crop the furmark or prime95 out for me).
Proper testing is needed to verify this, you current have borked data that is not showing the needed data. Do this test for the current values, which will give us your throttled cpu speed. But a GPU/CPU CO-LOAD is always needed, just a cpu load will not show the throttle.
Make sure you run this test for a half hour, come back then look at the values / printscreen for me.
Prepared to be throttled.
These are my baseline settings ie stock (although the screengrab software was causing the cores to run at near-max freq) - something to compare my next round of screenshots with.
Using Sony Vegas Pro to render HD video, using GPU effects, will use 100% CPU, and up to 100% GPU.
I'll try and run Counter Strike simultaneously in a window to bring the GPU up to 100%, and take some screens with the same openhardwaremonitor gadget, and CPU load and temp gadget. The vegas render will show how long the system has been under load.
I look forward to seeing if this causes any throttling - it hasn't in the past.
I'd vote that you push for furmark on your gpu load and even prime 95 for your cpu these are tried and tests methods for finding these problems, they put a solid stress level; I have no experiance with sony vegas being a good debugging tool for this kind of thing. I'm looking fowards to see what you come up with.
I used vegas to render a 1080p movie with 7 GPU accellerated effects on each clip.
I had Counter Strike running in a window (for extra CPU & GPU usage)
I had ATI Tray Tools open with its GPU renderer running for GUU load
I had winrar archiving 22GB of video - for CPU load.
The screen area of interest is top right - shows core/thread loading & temps in top gadget, and core speeds/loads & GPU speeds & load in the lower gadget (openhardware monitor).
I had to overclock the 6750M GPU to 725/932 (stock limits 600/800) but then I finally did see some thermal throttling - instead of the i7 2820QM running at its rated 2.3 GHz it was clocked back to 2.0. The lowest I saw it go was 1.9GHz, which is 400MHz slower than its rated speed.
So, I have to admit - the cooling design of the macbook pro is a comprimise, and under extreme load you may see CPU thermal throttling, but to get that throttling to reduce the i7 to its minimum speed of 800MHz I believe there must be a problem with your individual macbook pro - I had my system fairly loaded both CPU & GPU, and the GPU was overclocked ~20%, and I only briefly saw 1900 MHz, it average 2000 to 2100 MHz which is onyl just under its rated 2300 MHz.
I hope this helps.
Hey just a heads up, i've found the throttling on three units, two i returned and the current with a new Logicboard, fan and thermal paste.
It isnt the system, even the display units in the Apple store do this when i run the tests off my External TB SSD.
So lets move on to getting you the correct baseline stressing.
The only problem i see is your GPU core is only 75-80% loaded, and using rendering as a subsitute for cpu stressing (less consistant vs more consistant).
ATI tray tools 3d rendering is outdated and far aged for its time, as is CS:S (almost decade old), we need to get your GPU v-core to sit at the 1.0 volt load power usage (stock v-core), its currently being undervolted and/or under-utilized, that extra 25% gpu load is extremely important we need to see it @ 100%.
Can you run Furmark with maxium stress/resolution/depth detail in windowed mode. If your messing with your clocks and GPU volts can you set them to default too, this will give us a baseline since v-core voltage will toss off the stock throttling. Also were not completely maxing your cpu (97-98% or so might as well run it @ 100%) so run Prime 95 with 8 threads maxium heat ftt. When you run it make sure furmark is at the top of the window stack (currently selected).
When you run them both, i'd even run one or two CS:S windows just for good measure in the background, so that were pushing past stress testing like BF3 on the gpu.
What you've done for testing is pretty good, but i'm sure we can get the throttle far higher then 1.9ghz. We've got what we need, we know that its throttling, but just need to get the baseline testing correct and it should tank pretty quickly i think. That extra +25% gpu load might/should do it, but toss prime cpu stress into the mix for safe measure. This will give us a conclusive idea, rather then a more dyanmic one with non-stressing based software. It makes it harder to find these problems that way.
Can someone please figure this out? I upgraded from the MBP 13" platform to the 15" for a graphics card that I cant even use?... League of Legends is anywhere from 20-60FPS never consistant never enjoyable on this "mobile computing solution." so far all i'm really happy about with my mbp is how well it does itunes and browses the web in osx lion..
I tried to use catalyst control center to enable power play but the apply button is grayed out.
Please apple please tell me I didn't just **** away a bunch of money that I could have used on a quality PC gaming laptop.
So did you end up keeping the 2011?
I purchased mine from best buy during christmas.. stuck with it.
but my main question is.. Has anyone figured hout how to use catalyst control center? I am trying to access the powerplay options but it is grayed out when I go to apply any changed settings..
also if I make a change in the OSX part of the computer for power to help increase performance is that reflected in bootcamp? I figured no but just curious.
SOMEONE PLEASE FIND A SOLUTION