Problems? See this thread at archive.org.
Yes, Core 0 and 2 are ~10°C hotter than Cores 1 and 3, screws could help on that and be enough.[/QUOTE]
Great! hope mine will be fixed
-
-
check it again
original was 2.6 but current is 3.1
-
-
-
-
The processor run at the rated max 45W TDP and can't hold Turbo boost
Spartan@HIDevolution likes this.
-
Dunno why but it seems fine now:
http://imgur.com/a/WLZKE
max of nearly 3.5ghz, temps around 60-70.
I'm running on battery too btw.. but I have it on performance mode.
And I am mostly clueless at this sort of thing
I suspect I am happier than I would have been had my aw15 r3 with gtx1070 arrived. -
[email protected] GHz all 4 cores. + fully locked
Spartan@HIDevolution likes this.
-
Guys guys guysss! some news, now i undervolted to -150! here are the result, the only downside still is the cores not being even but at these temps i don't think it's a big deal, i hope tightening the screws will get that better, but keep in mind : THIS IS WITHOUT REPASTING, or touching screws or the heatsink! I didn't update the Bios yet, some said it takes temps down -5 or so, others says it messes things up.. anyway:
![[IMG]](images/storyImages/CyEHQItXEAAJkiQ.jpg)
Question is, how low could it get after a repaste? and adjusting pressure on the screws to get cores to be more even? i think i don't have the best, but it's not a unit to return IMO -
The Windows Power Options Control Panel, for the Balanced Power Plan, set the Advanced Options for the Processor Power Management - 0% minimum, 99% maximum - anything under 100% in maximum disables Turbo.
In Windows Power Options Control panel, for the High Performance power plan, set the Advanced Options for the Processor Power Management - 99% minimum, 99% maximum - anything under 100% in maximum disables Turbo.
Then there are the other notes I gave you
The longer you test, the higher speed the fans should run and the more they will cool things off. Initially the fans won't respond quick enough, they ramp up slowly so heat builds up quicker and doesn't cool down as quickly.
That's why I run 100% fans until I get to the lowest ambient temperature as read in hwinfo64, and then run the test. Then after the test I let the fans cool down back to the coolest ambient temperature, and then run another test. Always start the test from the lowest ambient temperature to get consistent comparable readings between runs.
You do have high temperature core differential, which means your heatplate isn't applied with even pressure across the screw points, putting higher pressure on the cool cores and less pressure or thermal contact on the hot cores.
You can tweak it yourself, either just adjusting the screw pressure - releaving the pressure from the cool side and allowing slack to be used to add pressure to the hot side - even out the pressure to get the same thermal contact across the CPU die and even out the cooling.
Some completely disassemble to re-paste with something better - like ICD or Gelid Extreme, I don't recommend Liquid metal pastes for this as they are a very thin coating that needs excellent alignment of the CPU die and heatplate - which is hard to do on the AW mount. Thicker pastes will work better, like ICD.
AW was more cooperative earlier with fixing this problem, but many report more problems after AW sends an in home repair person, and won't RMA the laptop unless the temps are over 100c - which is ridiculous.
For now I'd recommend returning it and getting something else until AW follows through on this:Last edited by a moderator: Dec 5, 2016Spartan@HIDevolution likes this. -
-
Going back to his original question, trying to stop coil whine in the power section, he is looking for a way to stop the load related coil whine. He gets this even in sleep mode, which is odd, but I thought trying some ways of reducing the power draw under various power plans - disabling Turbo - might work.
Dropping out of Turbo mode stops the high power draw, reducing the CPU speed to base speed does this, 99% in the maximum setting position for CPU Power Management is a shorthand for "disable Turbo".
At 0%/100% the CPU will Turbo to maximum in the Balanced Power plan, at 100%/100% the CPU will *stay* at maximum clock rate even after the load has gone away, even in idle it will stay at high clock speed - unless it power throttles or thermal throttles then the clock speed will drop.
I made other suggestions, including charging while off, to reduce the power draw of charging combined with running.
So your question is unconnected from the original problem, read his posts again.
With those temperature differentials, high CPU load will cause thermal throttling on those high temp cores first.
He needs to run something that actually loads the CPU long enough, like a CPU intensive game, Prime95, Heaven / Valley, etc.
The core differential at low loads with low temps aren't the issue, it's the differential causing thermal throttling at high loads and high temps, which you get while gaming that is the problem.
You can make it look good by keeping the ambient temperatures and load low, but it's gonna be a problem in real use - there isn't enough play room in the thermal range for anything but even temperatures to work correctly - high core temperature differentials ruin it's potential.Papusan, Spartan@HIDevolution and ex-28 like this. -
-
I find it hard to accept that the overall temperature is low enough to allow cores +10c to not hit thermal throttling in usage.
If you can understand what I am trying to say, the CPU can't possibly run cool enough to allow 10c+ on some cores and not hit temperatures that are too high on those cores under normal usage.
Run more tests using other benchmarks or games, don't fix your estimation to one test or benchmark.
Try prime95.
Don't argue that OCCT is higher usage, actually run Prime95 and other tests.
I have seen situations where someone has mis-configured a test such that it wasn't really hitting the hardware that hard as preferences were set, saying it was ok, but when a game was run it crashed - so they wondered what's wrong?
The test was set up wrong, giving a false sense of stability, and real world usage showed it would fail.
Run more different tests. A 10c+ core differential can't be good under load.
-
-
I still don't think you are loading it enough. I have never run across a situation where such core differential's were usable.
I don't recommend "living with it".
When I run stuff at load, it's bouncing off of the limits, +10c or more additional temperature wouldn't be tolerated.
But, if you have convinced yourself that your usage works in that realm, then good luck
When you finally come across a situation where you do thermal throttle on the hot cores, please come back and let us know what usage that is
-
-
It sounds more like you are trying to figure out how to convince yourself how to live with a "broken" laptop
A 10c+ core differential is the definition of a broken laptop. It's actually Intel's definition of a broken CPU, if it was the CPU that was at fault.
It needs to be fixed, not to justify it's broken existence, that's Alienware's Job
Papusan likes this. -
-
So it was broken enough for you to return it, but you want to figure out how someone else should justify it and keep it in it's broken state? WTH??
Yes, I do have proof, I tried to live with one of the CPU's that had a 10+ core differential, and I went through the same self-justification path, testing and finding in most things that it was ok. At stock speeds in an air-conditioned room.
When I turned off the air-conditioning and let room temperature hit the PC - then ambient temps went up 10-15F+ - all of sudden my previously non-throttling runs started to thermal throttle.
In my environment it was "ok", when in the environment like it would eventually be running in it wasn't ok.
And, of course I was already undervolted to the maximum, and running at stock speeds.
Then I tried to OC to 5ghz, and the hot cores limited my OC to well under that.
After swapping the CPU with Intel, a 2700K, I was able to OC to 5ghz, and it's still running at that speed years later, without a single reported crash
That wouldn't have been possible with that 10C+ core differential CPU.
That's why I am saying that any CPU that has a couple of cores with 10C+ higher temperatures isn't right, it's broken, and it's gonna show it's failure at some point - it's better to get it replaced now than wait for it to show it's failure point(s).
This is a well established manufacturing fault that AW is fixing, to be fixed in a "month or two", and the symptoms could get worse over time.
It's best to return it and wait to reorder when AW has fixed it, and new owners at that time confirm they are getting good results - no more uneven core temperatures.Last edited: Nov 24, 2016Papusan likes this. -
-
-
And, the symptoms could get worse over time for existing owners - thermals usually degrade over time. Winter is cold, summer is hot, not all portable locations will have optimal ambient temperatures, etc.
AW thinks it's broken and is fixing it in manufacturing, so it's broken, whether you can "live with it" or not
It's best to return it and wait to reorder when AW has fixed it, and new owners at that time confirm they are getting good results - no more uneven core temperatures.
There's nothing to "agree about", it's a broken design / implementation that shouldn't have ever left the factory and made it into end user hands, there's no question of that now.
What did you get after returning your broken AW?
Last edited: Dec 5, 2016Papusan likes this. -
-
No, it's some other component interaction - or hardware problem, if you can't alter it with software, swap the hardware
Their comments are more worrisome than reassuring
100c CPU temps are "ok", AW releases a BIOS that changes the shutdown temperature from 100c to something higher, "magical" instantaneous changes were made to all inventory - and there wasn't anything wrong in the first place, but now it's all fixed?
I'd sooner believe the word of someone working inside AW like that Chat tech, his comments ring true from my own experience in such things.
AW has pre-built chassis stock, boards, and assemblies already in process that they would have had to pull out and re-work / verify in QA - not possible in the time from news to their reactions.
AW has pre-built complete laptops in stores, in the shipping channel, in their outgoing inventory that would have to be pulled back and re-worked, and there is no way that stock would even be in house yet for re-work, let alone "already fixed".
It takes time to redo the processes such that the assembly is done correctly - training and instructions would need to be done, and then run through a few cycles of manufacturing with feedback from QA as to how successful the changes have been.
Again, there hasn't been near enough time to hold enough meetings about about the problem toward a fix, let alone be done implementing the fixes successfully.
The "month" or "two" are reasonable estimates for the implementation, testing, and QA feedback on the success of the process changes, that he must have heard about in a meeting.
Much more believable.
And, clearly we are still seeing people with uneven core temperature differentials, some much worse than others, but all are broken - in different levels of failure.
I'd rather error on the side of people not getting stuck with a lemon.
And, if you really believe AW has fixed it now, swapping for another unit should end up with a correctly built laptop with zero core differential issues, right?
I'd still recommend returning it now and wait to reorder until the problems stop getting reported in forums like this, and instead of complaints we start seeing lots of happy owners reporting no problems, just lots of great benchmark results.
Hopefully that's within the "month" or "two" estimates
Last edited by a moderator: Dec 5, 2016hasnain.hyder, Papusan and Cass-Olé like this. -
It's a relatively good system in terms of temps after I repasted it. GPU doesn't go beyond 66 degrees in COD max settings at 4k resolution. Cpu at 3.5 ghz is 88 max (less than 78 at 3.2 ghz). It's already a replacement system and I'm afraid i will get a bad one after that
hmscott likes this.
*OFFICIAL* Alienware 15 R3 Owner's Lounge
Discussion in '2015+ Alienware 13 / 15 / 17' started by katalin_2003, Oct 24, 2016.
