Alienware 17r3 6820HK CPU problem ?

I have a problem with my alienware 17r3 Spec (6820HK 980m) and my friend have it too (15r1 4710HQ 17r4 6700HQ) all have problem with CPU and frequency drop when i play some game in normal it my cpu speed is 3.30 mhz but when i playing game it drop into 800mhz and GPU not load max it only load about 50 60% and fps drop so much Dota2 normal 120FPS when it drop all way 20 30FPS....
Here about my stresstest after 15 minute CPU frequency go drop all to 800Mhz
http://imgur.com/a/lVEZB

 

My guess here is you are thermal throttling from the 91c it is hitting.

 

What are your temperatures like?

 

I do not have your laptop, but the posts you have are pretty conclusive. There is another thread on the CPU you have generally having this overheating issue on varied cores.

http://forum.notebookreview.com/thr...ven-core-temps-due-to-uneven-heatsink.797477/

 

No when it begin stresstest my cpu temp go high like that but when i playing game like witcher 3 my cpu temp about 7x
http://imgur.com/a/mJtHe max temp 74 after that fps go down and CPU frequency go 800mhz.....

 

if you go to throttling at 74c that is bad but understand if it throttles the temp will go down once it hits the 800MHz.

Also remember that those readers see an average temp. The CPU will react instantaneously to a temp spike. This is to save the CPU before even a short overtemp run kills it.

 

yah i know what it mean when it throttles temp go down about 60-65
here: http://imgur.com/a/nFQ3u

 

when playing game temp CPU about 7x and GPU about 68 69

 

Those temps shouldn't be causing throttling. The CPU inside your machine throttles at 94-95*C

 

When you say 7x do you mean between 70c-79c?

 

yah u can see my picture there ...... cpu only 74 75 when playing witcher 3

 

i know that i ask alot of my friend because i am Admin 1 group about alienware in my country and some one have this problem too.... i dont know how to fix it -> update bios, reinstall window, repast with liquid ultra..... i test every thing and cant fix it :'(

 

Well in the below you show 74C for the GPU but 77C for the CPU. Still not enough for thermal throttling but your charts running furmark and prime 95 do show throttling.

Another cause could be not enough juice from the power brick. Try lowering the overclock on the video card for that.

http://imgur.com/a/mJtHe

 

how to do that ? i didnt overclock my GPU

 

help.....

 

I would hope unoverclocked it is not getting throttled from lack of power, then again everything has tolerances and it is possible (hopefully unlikely) it could still be an issue. You could try MSI afterburner and underclock. Alienware may have their own utility.

 

Run Throttlestop 8.48 and check the Limit Reasons box.
When it drops to 800 mhz, are you sure it is still saying the AC Adapter is plugged in?

You can also try unchecking "BD Prochot". Some posts saying that BD Prochot will force the CPU to 800 mhz.

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-530

 

On the above it could be the adapter is plugged in but is folding under the power drain. I did not know TS has that now, cool.........

 

What happens if you uncheck BD Prochot in throttlestop?

 

Sorry in my country at midnight now. So i will try tomorrow..... Thank all

 

i test with witcher3 and after 20 minute it not drop awesome i will test with another game late and feedback late

 

it's work for me but i cant stresstest ..... i stresstest with prime95 and furmark my lap go blue screen after 5 minute but i playing game 2-3 hour not thing happen no blue screen no drop FPS ..... cool

 

Yeah, the BD Prochot should cure the issue as it is telling the CPU not to throttle by ignoring the MB sensor telling the CPU it is overheated. Not a fix but a cure of the systems as it could still be overheating in very small spikes.

 

I doubt it's having heat-spikes as if that were the case, a P95 load test for an hour should show higher thermals.

 

I am not sure then what trips the BD_Prochot. If it is a defective sensor then so be it, if not it could be an issue. If the heat for some reason for a MS or two spikes but then the CPU cools this could do it, but that could also be a sign of a chip that will fail down the line. Prior to setting BD_Prochot it looks like he was seeing 91c (probably 96c is the 5c offset were in) on a single core and this again is the issue I linked to in the other thread with uneven core temps.

One way or another just keep a close eye on it.

 

There's definitely something odd happening. I've seen my fair share of defective thermal sensors (mostly on AM3+ Vishera) where temps were either 10*C under load or 150*C idle. Here I guess mentioning that software CAN be inaccurate is also important - for example, HWiNFO reports my GPU temps about 5*C cooler than GPU-Z does which in turn reports them as being 2*C cooler than HWMonitor. All set to 0.1s tracking, not spikes but just a stable 75*C on one, stable 80*C on the other and stable 82*C on the last.

 

Just so you understand, when BD PROCHOT is not checked, your CPU will still be able to throttle if your CPU core temperatures get too hot. This safety mechanism cannot be turned off by using ThrottleStop. Turning off BD PROCHOT only blocks outside signals to your CPU. I think BD PROCHOT signals can be either temperature or power consumption related. When stress testing, it might be the voltage regulator that is overheating or drawing too much power which is causing the throttling.

I think the combination of Prime95 and Furmark is too much stress for a laptop. If you are only gaming, you should be OK turning off BD PROCHOT but you could damage something if you run those 2 programs with your CPU running at its full rated speed.

Without being part of the engineering team, it is impossible to know what Dell is trying to protect. There might be some weak component that they are afraid will fail. Definitely bad design when it is only letting you run your high performance CPU at 800 MHz while gaming. These BD PROCHOT related throttling issues are way too common. Dell is not the only manufacturer with these sort of issues.

@don_svetlio : If you are bored, can you also do some testing with ThrottleStop? It would be interesting to see the temperature differences between programs. ThrottleStop runs at a higher Windows priority so for CPU tracking, it should be competitive with the other programs. You can use the More Data option so it samples the CPU more often. ThrottleStop avoids sampling the GPU too often because there is a lot more overhead reading temperature data from a GPU compared to an Intel CPU. ThrottleStop and some other monitoring programs get GPU temperature data directly from the driver. I think GPU-Z reads the temperature sensor directly.

 

Thanks, that is why I mentioned sensor issues. Now I did not know it also can see power issues, so that means again the brick could be suspect. So I will again iterate that this may be stopping the declock but it is not curing the primary cause.

Edit; I also have to ask is the sensor for heat and power driven by the chipset where there is a errant that the data is pushed to BD_Prochot where as almost all the programs are doing a read of the sensor data? If this is the case it is easy for us to miss an errata from just polling.

 

No 1. Many of these first Alienware Echo models 15r1 and Alienware 17R2 were known for full throttle down to 799MHz. There is no known cure for this. @VICKYGAMEBOY and @judal57 can chime in. From first AW Echo models Alienware 15 R1 Throttles to 0.78 GHz - Alienware 15 R1 CPU ISSUE *WARNING*


Nr2. I would have take a second look at PCH temperatures for all AFFECTED AW models. Well-known that some Alienware models has throttling problems related to this temp sensor(from the first AW Echo models to the newest). Even with ok processor temp, the processor will throttle down no matter. I would assume that the PC chips temp may be one of the culprits for this MESS.

Nr3. And please do not use Furmark in your testing(You will kill your graphics). Rather install and use AIDA64 stability test for all your testing as shown in this link.

 

It sounds like disabling BD PROCHOT was enough to cure his 800 MHz throttling problem, at least while gaming. The Dell engineers probably never considered what would happen if this laptop was running Prime95 + Furmark while unthrottled. A BSOD during this test is not too surprising. That is an unrealistic and excessive load for a BGA laptop.

ThrottleStop cannot solve all of these throttling issues but it is great to hear that it has solved this one.

@TANWare - I don't know enough about what manufacturers are dreaming up for throttling schemes or what sensors or triggers they are using. All I know is that if you disable BD PROCHOT and your CPU starts running at its rated speed, this usually means that a sensor has failed or has been set way too conservatively. Trying to get manufacturers to own up to these problems let alone fix them usually ends up being a frustrating waste of time. The people on the front lines will give you a blank stare if you start telling them that disabling BD PROCHOT fixed your throttling problem. They will go to their script, step 1, reinstall Windows.

If something on the motherboard is overheating, turning off BD PROCHOT is probably not 100% safe but I do not recall a single user contacting me to tell me that their laptop completely failed because BD PROCHOT was disabled.

 

@Papusan There are ways to keep from throttling less atleast, one is TS BD prochot, which works till some extend, first time when i found that a year back on on my Haswell 4720HQ, people denied me its because of that sensor, but after recording temps, and graphs along with so many things, that was the main culprit, not any other sensors, then judal confirmed it too from his side.. if BD prochot doesnt help people, i kindly suggest them to a do a repaste with LM, so that the overall system cooling gets better, and next do the PCH mod, by buying a 28X28X3mm aluminium heatsink, so that atleast it can spread the heat better thn the original small die area of PCH.. i dont know why dell doesnt check the sensor and increase the limit, as PCH fail safe is way over 105, atleast 90C should do.. and the problem is PCH temps have major spikes when there is heavy load from ur BUS lanes or like when there is sudden load of NVMe drives.. keep ur system cool.. and keep a manual fan curve.. before u start gaming.. thats all i can suggest for now.. if not remove the back panal access door and play, and keep a notepad cooler, sounds crazy, but these are the solutions..

 

Yeah, you get that blank stare as 99% of those tech people have no idea there is low level system code running unaffected by OS and program code. I casn even remember my P79 with a Q8200 the sensor at TSZ1 would trigger flooded for no reason to 99c. Again TS was to the rescue there as well.

 

so is there any other alternative ?? BIOS mod ?

 

Usually this code is extremely low level and an integrated part of the chipset. While even more low level think of it as an interrupt on the bus for your PCIe channel. This code may, or may not, be programmable by a new BIOS. The other issue is hardware design of the sensor could be defective causing a false errata push.

If you can't get them to admit there is even an issue good luck getting the white papers, original source code, required hardware design data and the engineers that can find the issue let alone solve the problem.

 

They will never admit there could be a flaw. Even with several complaints during the years starting with the first "Echo" model they continue in exactly the same way. Latest models continue at +83-84C on PCH chips and no physical heat spreader. Sooner or later, it will fail as always.
However, you have the offer of up to 4 years Premium support at premium price. And their Customers have to pay for a screwed design + screeed firmware. Oh'well.

 

@Papusan hi, the 800mhz throttling is not a problem anymore. the 47W limit is. i can play games at 3,5ghz on all cores without throttling, but i cant surpass the 47W while doing synthetic test's like AIDA64 with FULL boxes checked, once i start the max frequency is 3,1ghz for all cores. the excessive heat is another problem of Hotwell. i managed it with LM paste. but that is not an easy solution for a regular user.

 

If this is a 6820HK or 7820HK, either:
1) Increase CPU VR limit to 100 amps in the BIOS (Default is 70 amps). If it is called ICCMAX, that's the same thing.
2) Increase ICCMAX to 100 amps in Intel XTU.

 

I just installed an OEM i7-3940xm using ICD in my Samsung. I get 3.7 GHZ on all 4 cores under full load doing Video encoding. I can see spikes to 3.9 GHz from turbo boost before the heat eventually settles in and 3.6 GHz is more attainable with 3.7 spikes. At the later point 92c is about where the temps settle in.

I have no Bios access to XTU so I do not get to play much and I do not have to disable BD_Prochot but if I do with TS it seems I get 0.1 GHZ better but that could be a placebo effect.

I should note that my UEFI table was modified a while back to add the +400 MHz for the i7-3820 QS that was installed.

I note this as not every system seems to be tripped by those sensors.

 

As yoo can see. No changes!!

You could compare and see if t's placebo effect or not with wPRIME (Run as adm, select 8 threads in advanced options. Fire up 1024M stress test). You could even test with TS bench in Throttlestop.

 

@Papusan the problem is not what a program will see as a temp. The problem lies in that the sensors are so setup as a push at any showing of an errata temp spike. Even one so short and fast it is not caught during one of those programs polling the sensor data. A larger heat sink may help to disperse the spikes but may not fully stop them.

I just love when these techs say, "the monitoring program shows the temps are good" what a laugh. Either temps or electrical spikes happen way faster than any simple polling program will catch. You also have to watch, just a degrading CAP or two can cause enough unclean electrical signal to cause the temp spikes that will throw a system fits.

Edit; I should note a laptop power system is unlike a desktop ATX. A power brick may deliver 245 watts but it is all at 19v. The voltage rails for power distribution are broken down in the laptop itself. You may have enough total power from the brick but the rail breakout system may be failing through electrical degradation.

 

I thought more about that Dell does not do anything about the known PCH temp problem. You do not always need put heat shield on the PCH chips... But then must the internal airflow work more optimally(a problem in thin chassis). Regarding temp spikes... We was bothered with this with Alienware socket models (Haswell). The fans did not start until the temp was close to thermal shutdown. Remember we could opt for Extreme processors in their models. We asked several times if Dell could implent manuel fan control in main AW software or fix the crippled fan profiles in EC. But No... They wasn't interested. Even today... Nothing have changed. Dell's engineers said by them selves that we could use 3rd party fan control like Hwinfo.

Spoiler: See also my posting to several chef higher up in the AW system... Dell is Dell!!

 

Tried that, placebo effect. I do have to say this systems has been rock solid even if not the bleeding edge fast.

 

So Weird some friend told me that the model alienware 17r4 (6700HQ) GTX1060 have it same problem like me and they dont know how to solve them.....only change mainboard to solve it :'(

 

I will try and feedback for you

 

From the New AW models!! And other problems as well!! Yeah, I know.

 

i try repaste many time use 3 different types of thermal compound like MX4 IC diamond and my last choise is Liquid Ultra my temp realy good as well but they throtting when temp CPU only 7x u know that :'( i cant belive it :'( also try UV too..... temp go down when i stresstest full 20min only 88 degree

 

i can fix overheat in 17r4 too.... I try to repaste it and some model need to "curling" the heatsink "tight" with CPU........ (sorry my english very bad)