Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Help: rMBP 650M GPU Throttling on BootCamp

Hi to all. I have problem with lags, and after reading this thread still cant find solution. I installed win7 via bootcamp, latest drivers from nvidia 306.23. After few days of normal gameplay, suddenly i experienced lags - after playing mass effect 3 for 5 minutes I'm receiving lag tick every 1 sec, even after game closed and laptops cools down, it continues (i can see it on windows animations), only reboot helps, inspected through nvidia inspector - max temperature 72 degrease, 900/2500. What can i do? Or maybe its just my laptop have factory fault?
 
i am having the exact same problem however i am getting around 225 core clock when i run my game. this make my game unplayable most of the time.
i really need help
please help me
 
dusk007 said:
The Op seems to be under the impression that 900Mhz ist default and anything below is throttling.

725Mhz is the default clock speed of the 650M with GDDR5.
900 Mhz is the Boost aka Turbo and how much it boosts depends on the workload.



GTA4 apparently simply consumes too much. Also the quality of the DIE plays a part. As it measures outside the chip the real power consumption some that got a cold chip will get better Turbo out of it than those with slightly hotter chips. Much the same as some can overclock higher with out problems than others. Also since according to Anandtech power consumption is measured, better cooling won't change too much about it. Heat has some impact on power consumption but only some.

Read up on Trubo boost.
http://www.anandtech.com/show/5699/nvidia-geforce-gtx-680-review/4

What you are experiencing is not throttling just not as great Turbo speeds.


If you don't want to deal with all this just over clock the chip. Adjusting the power target would really be the way to go. I don't have a 650M notebook but maybe some notebook over clocking tools allow it already like on the Desktop. Otherwise just set some custom fixed over clock as suggested, as long as heat stays reasonable and the power supply + battery can handle it.
Click to expand...

Not the case with the 650M in the rMBP. Its base clocks are 900/2500 and it has no boost frequencies.
 
I've experienced the exact same "throttling" Problem...
SMC Reset only seems to cure this Thing for a short period of time, then after some days of usage, the Problem returns.
 
Just started to experience this odd behavior myself. So i did of course reset the SMC and all, but somehow its really a little weird. I tested both the CPU and GPU while playing Skyrim.

I force my clocks with NVInspector (1150/3060)

At first when i start my game it seems really nice, but at some point my gpu core clock goes as low as 724Mhz whereas my memory clock stays stable the entire time. Isnt the EFI/throttling issue causing both clocks do get lowered?

All of this happened after i updated my drivers from 306.23 to 306.79.
Even after going back to 306.23, the problem persists.

I will try to reset the power management mode in the control panel first, if that doesnt work, i will simply do a clean uninstall of my drivers.

Any other ideas? Anyone experiencing the same issue?

Before i forget, i set both the minimum and maximum processor states to 90%, so it doesnt use turbo mode. Also i never experienced any kind of overheating. Even in intensive scenes, my temperature is around 73°C..
 
Anyone?

This problem is seriously annoying for me. The gpu is throttling a lot, while my cpu stays stable. Should i maybe use throttlestop instead of the standard power profiles on Windows?

How do you go about resetting the SMC anyway?
It should be Left shift, control, alt and power button simultaneously, right?
Since the light has never ever changed its colors, i am not even sure if it worked right...
 
what is your gpu temp? have you clocked it to see if its not really throttling itself? the kepler ones have a much lower throttling threshold, its around 70-80c
 
Its about 72-74°C while being 100% utilized. For the moment i seem to have it under control as i had switched from my high performance profile (which had 90% for min/max processor power state) to the balanced profile. (5% min, 95% max)

I guess it might be a power distribution problem maybe?
Maybe all of what i described just know didnt even solve my problem, maybe it was just me resetting the smc and restarting my computer. I guess i will never really know, but i do hope Apple will finally fix this damn SMC problem soon.
Its much easier to do something about a power distribution problem than doing something about an always returning smc issue.
 
Hi there,

any news regarding this issue? I've got the same on my rMBP.
Resetting SMC helps but after 2 or 3 boots into Windows it comes back and the clocks will be locked to around 350 Mhz which is very annoying. You can't tell if throttling is active until you start a game.

Very annoying indeed and can also occur in OS X.
 
I just want to know what exactly the problem here is :

Apparantly people get 2 different issues. Either they are stuck at very low clocks (SMC Bug?) or they get occasional throttling -- am i right?

In my case, i get occasional throttling, that is that my clocks are lowered from 1150MHz to 792Mhz (Core clocks only, memory remains fine) -- very bad for performance and very annoying.

Is there any solution to this? Is it maybe possible that i just used NVInspector the wrong way? Here is my bat :

nvidiaInspector.exe -setBaseClockOffset:0,0,250 -setMemoryClockOffset:0,0,550 -setpowertarget:0,0 -setVoltageOffset:0,0,0 -setGpuClock:0,2,1150 -setMemoryClock:0,2,3058 -forcepstate:0,1

Anything wrong with it? Should i just try resetting the SMC? I did notice, that with Skyrim it seems almost ok for most of the time, but Borderlands 2 and Far Cry 3 really get this issue to appear.

My temperatures are alright by the way and my isssue occurs at maximum performance profiles and a very moderate balanced profile..
 
My problem has all along been that the gpu has been throttling like a bugger everytime i have started a game and when the gpu reaches 97 degrees. Reseting the SMC didin't work. What i did was deliver it to Apple for guarantee repair. They tried changing the motherboard twice, didin't work. But then a genuis suggested after 3 months that the heat problem may come from a fault on the heatsink. Incredible that it needs 2 repair guys from apple and 3 months of service to find out that a heat problem can come from a faulty heatsink. My rmbp will arrive back in a week or so. Hopefully they have fixed the problem.
 
Well my temperature is quite alright. 85°C on the CPU and 76°C on the GPU...during heavy gaming. So i guess it must be something else..

Edit : After a lot more testing it seems to be the SMC bug, because resetting it a number of times does seem to do the trick. Since this is a bug that occurs on Mac OS X and Windows, how is it possible that Apple didnt offer any solution yet, even after almost half a year?? :( If it was Windows specific, it would at least seem a little more plausible that Apple hasnt such a huge interest in fixing it...but this bug here happens universally...
 
Last edited:
When did throttling occur for you guys? Maybe while the MBPr was recharging? For some reason i get these stutters when i recharge it. Being on battery power only or being on a fully charged one that is plugged in wont trigger throttling apparently..
 
MagnusB said:
My problem has all along been that the gpu has been throttling like a bugger everytime i have started a game and when the gpu reaches 97 degrees. Reseting the SMC didin't work. What i did was deliver it to Apple for guarantee repair. They tried changing the motherboard twice, didin't work. But then a genuis suggested after 3 months that the heat problem may come from a fault on the heatsink. Incredible that it needs 2 repair guys from apple and 3 months of service to find out that a heat problem can come from a faulty heatsink. My rmbp will arrive back in a week or so. Hopefully they have fixed the problem.
Click to expand...

Magnus, did fixing the heatsink solve your problems? I'm having the same issues - smc reset doesn't fix anything and getting temps of 97degrees on the GPU.
 
Temps look rather fantastic, especially for the compact case... Now I'm tempted to get a '13 model when they come out...

But what's the wattage? If the power supply cannot provide enough power for CPU and GPU... my 2011 MBP needs 95+w and the power supply provides 85. As a result, the teeny tiny brick gets scalding hot to the touch (and if the stock model throttles, the BTO isn't going to be any better.) The CPU (and/or GPU) throttle to save themselves with.

That is also why Macrumors found the 17" to run faster, despite having the same CPU and GPU as the 15" model. The 17" has a 95w battery, compensating for the inadequacy of the 85w AC adapter. The 15" MBP did not have a 95 watt supply and could not provide the juice.

Don't the American engineers that design these things know these things? :( No wonder Steve Jobs was lambasting Americans... ironic... of course, Apple could be doing things on the cheap to score higher profit, in which case my apologies to the engineers* and my derision toward the marketers instead.

* including the engineer who was ignored when he warned the CEO of the time of the antenna problem, with the CEO deciding to blame customers and telling them "YOU'RE HOLDING IT WRONG!" instead...

----------
DTKblaster said:
Don't let the GPU hit 78 degrees.

Disable turbo (throttle stop).

Max out the fans manually with Lubbosfancontrol (this happens anyway in games, just retardedly slow, so just do it manually to stop early throttling).

Repaste the CPU/GPU with decent paste.

If you want, you can also force P1 in nvidia inspector by making a text file containing "nvidiaInspector.exe -forcepstate:0,1 " and then making it a .bat file and running it in the same folder as the nvidia inspector exe. (this won't get around the temperature based throttling)

(to make a .txt into a .bat in windows 7; open a folder, press alt to reveal the menu, then go tools -> folder options -> view tab -> uncheck "hide known file extentions". Now you can rename the file extention from .txt to .bat)
Click to expand...

A general rule of computing - always keep running temps as low as possible. The temps, high as they are (78C), are far better than previous models' higher-end readouts...

Still, I've been told "high temps don't matter" and "it just works" (so why take it all apart to fix it when it wasn't made properly to begin with?), so my cynicism isn't without underlying cause... :(
 
thermodynamic said:
A general rule of computing - always keep running temps as low as possible. The temps, high as they are (78C), are far better than previous models' higher-end readouts...

Still, I've been told "high temps don't matter" and "it just works" (so why take it all apart to fix it when it wasn't made properly to begin with?), so my cynicism isn't without underlying cause... :(
Click to expand...

This seems to have changed, probably in one of the firmware updates. Now it happily runs over 80 degrees without throttling down, even while overclocked.

And while you mention wattage, I just measured at the wall: 91w max draw (furmark + prime95), although this forced the cpu to run at 1.2ghz, even though its temps never got high. I would assume its limiting power draw so as not to drain the battery at the same time as maxing out the power adapter. Closing furmark lets the CPU jump to 3.1ghz.

While playing a GPU bound game (~99% gpu usage), it draws about 85W on medium brightness and volume.

Charging was drawing about 75w while at 50% battery, I noticed it later drawing 50w when almost full.

Another intersting point is that throttlestop lets me lower the cpu multiplier as low as x8, resulting in 800mhz clock rates under windows. This seems pretty new, as before I couldnt go lower than 1.2ghz.
 
So, no solution yet? My gpu throttling down to 725 mhz from 900 mhz when temperatures reach 80-85... And then temperature goes down at 75-78 celsius. So I have glitch in games (skip frames).
 
I am experiencing this kind of frame skip too and i am not entirely sure what is causing this. It might be temperature based and not SMC bug related after all :(

While playing Skyrim i notice it occasionally in dense forests. While playing Crysis 2/3 i notice it all the time...

Or is it still possible that all this is still related to the SMC bug?
 
imperialnavy said:
I am experiencing this kind of frame skip too and i am not entirely sure what is causing this. It might be temperature based and not SMC bug related after all :(

While playing Skyrim i notice it occasionally in dense forests. While playing Crysis 2/3 i notice it all the time...

Or is it still possible that all this is still related to the SMC bug?
Click to expand...

I've seen the odd frame skip issue you mentioned on my rMBP . Seems to happen often when the gpu hits 78c
I fixed it by disabling turbo on the CPU and I haven't seen it anymore.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back