Alienware R4 2017 - Are these normal temps?

Hello Guys, whats up?
Recently I´ve bought a R4 2017 with the 7820hk and the GTX 1070.

I am experiencing some weird CPU temps for example:

3D Mark - Fire Extreme / Time Spy
Max temp of the GTX 1070 is 67 celsius
Max temp of the 7820HK is 92-93 celsius

Dell Stress Test 15 min
Max temp of the GTX 1070 is 67 celsius
Max temp of the 7820HK is 94 celsius

Prime95 (only CPU)
Max temp of 7820HK is 73 celsius

and when I am playing (Overwatch, PUBG, BF1 and R6 Siege) both using GTX 1070 and the CPU, the temp of the CPU reaches 94 celsius.

So when the GPU and CPU are being stressed at the same tim the CPU is almost reaching its max temperature.

I want to know if this is normal or if I should call Dell support due some heatsink issues or design flaws.
I am using the latest version of the BIOS 1.19 I believe.

the CPU should be getting lower temps and the GPU should be getting highter temps... LOL

Thanks

Stefano

 

CPu temps are too high consider doing a repaste.

 

It's because Dell thought it was a brilliant idea to combine 2 heatsinks with the GPU and CPU. If you use both you considerably warm up both.



You can try to repaste, I highly doubt it will help to much tho.

 

Yeah, I believe it will not do a lot of difference, but in this cases for this design these temps are normal?
I mean... CPU so much hotter than GPU.

I am worries with thermal throttling since I use a FHD screen and its rare to see the 1070 on99% usage. So the effort goes to the CPU.

Is there a chance of assembling issues (motherboard X chassi X heatsinks ? Is it worth the try? Or these temps and behavior and under the specification?

Stefano


Enviado do meu iPhone usando Tapatalk Pro

 

Hey but if you see the VGA the max temp is 67 celsius, only CPU is suffering.

If I ran a CPU stress test it does got far than 73 celsius


Enviado do meu iPhone usando Tapatalk Pro

 

Again, you can try and repaste, no harm in doing that.

When looking at notebookcheck it seems the temps are pretty expected. There are massive throttling going on on stresstest, even on gaming like witcher 3 the notebook would throttle and reach extremely high temps. The ingenious developers made a cooling system that would transfer the heat from the graphicscard on to the CPU to heat it up instead of cooling it down.

I rate a solid 10/10.
Here is a screen of furmark only

This is the GTX 1080 version, but it tells the same story, GPU heating the CPU up, good job dell. First you put the mosfets next to the GPU an CPU on your XPS series, then you decide to warm your CPU up in your Alienware models.

I can't help but feel that Dell thought it was a good idea to hire monkeys around of the year of 2015 and fire their emploies because they wanted to save money somewhere, meanwhile they completely destroy their own reputation. Sure the old Alienware models werent perfect either *cough* harddisk next to GPU *cough*, but this disaster is getting worse and worse.

 

Hey Danish...yeah..its shame, I´ve waited like 3-4 years to buy an Alienware (before I used to buy Asus) to have problems like these.
The thing is, in my case is worse because I have a FHD screen and the CPU gets very hot while the GPU reaches at most 67 celsius.

My CPU can't go more than 3.4Ghz, since at least without any OC it should be reaching 3.9.
Again, Its been worse than a 7700K, I cant have the full potential of the VGA because of the resolution and also the CPU because of the overheating.
:/

I will see if I can open it up or just call Dell.

Thanks guys!

 

You think you have it worse because you got a FHD screen? It has absolutely nothing to do with that, if anything, the people wiht a 4K screen got it even worse since they use GPU even harder. As you can see on your own benchmarks on Prime95, this is completely GPU related, even tho 1080p requires more CPU power, the fact that the GPU heats the CPU up is downright concerning. I don't even know how they are allowed to sell these things. You have like mid 95s on overwatch etc, if u had 2k or 4k screen u'd be at full on throttle 99c.

Please tell us how the whole support thing went, since I've heard a lot of bad stories about dell support and am legit curious if u're going through the same.

 

Guys, I´ve ran several tests including Furmark, Unigine, 3D Mark (Fire Strike, Time Spy and Sky Diver), Dell Support Assist Tests and Prime 95.

Below is what I have concluded so far:

- Dell Support Assist 15 min stress tests raises the CPU temp until 94 degrees, but no Thermal Throttling yet. :|

- Furmark (same tests as our colleague Danish did), CPU never passes 90 Celsius :|

- Prime95, CPU never passes 73 Celsius

- 3D Mark (All tests put the CPU around 92 Celsius) :|

- Unigine (Several tests with different resolutions put the CPU around 88 Celsius)

1. The VGA temps never passes 67 Celsius in ANY tests / games.
2. The clock never passes 3.7 GHz (Never reached 3.9 GHz in any circumstance) - You guys know the cause?
3. For sure one of the issues its the heat-sink design sharing between CPU and GPU the same duct.

So before I raise a ticket at Dell I will try to play as much games I can during the weekend and let you guys know.

Thanks

Stefano

 

You do realize that Furrmark is a GPU only test right? It shouldn't even raise the temp even close to 90c in the first place.

 

Guys regarding the clock I got it, I just had to change the performance to L1, L2 or L3.

Now I can reach these clocks, but now its worse.

Thermal Throttling is happening only on Core 0 and I got some severe core temperature differential.

The laptop can't sustain 4.2GHz.



From 74 Celsius it spikes to 97 Celsius on Core 0, sometimes 30 degrees of difference.

According the article below, if the AVERAGE temp is higher 93 we should contact Dell.

https://www.dell.com/support/articl...-or-throttling-during-high-cpu-stress?lang=en

 

the funny thing is, GPU never goes more than 67 Celcius which is very good.
It might a partial heatsink issue or just a design flaw?

 

Even on the benchmarks on notebookcheck, it's the same thing, as far as I can tell the heat gets taken away from the GPU and somehow gets tranfered to the CPU instead of the Heatsink.

 

Ok! So I checked my heatsink, its CCI.

Laptop was made in July/17.

I bought at Microcenter in Chicago but I live in Brazil, I already transferred my warranty to My country.

Should I open a ticket to Dell an ask for heatsink and repasting?

I have some thermaltake paste here TG2 I believe.

Please advise.

Stefano


Enviado do meu iPhone usando Tapatalk Pro

 

I don't know what kind of warranty they have, you should check if you're allowed to repaste before doing it. If you're allowed to and the temps don't get any better i'd just give it back and buy something else instead.

 

Liquid metal thermal application and your golden.

@iunlock has FAR more experience than I with this platform but hopefully he can spare some time to look over this thread before you start going into RMA hell

 

I have on site support from 1 year here im Brazil but I can expand to three years, add complete
Care, premium.. but they cost a lot of
Money.

I will check with them.
Thks


Enviado do meu iPhone usando Tapatalk Pro

 

Yes, I want to avoid RMA


Enviado do meu iPhone usando Tapatalk Pro

 

Guys, at home did run some another round of tests.

1. Returned from 4.2 to the std profile.
2. Did an undervolt of 0.100
3. I ran Dell GPU stress test, max CPU temp was 88, GPU always on 67.
4. I ran Furmark for 3 hours, Max temp was 90 in the CPU and 73 on GPU.
5. No thermal throttling, all 4 cores running at 3.6GHz, no drops.

I have concluded that If I ran my 7820k with the 7700 core clock I can have a more efficient performance with no issues at all.

After my tests on weekend with games I will see if it worth call Dell to check about repasting or heatsink replacement, but now it seems better.

Just for a comparison purpose, I let furmark on my desktop ([email protected], Strix 1080 OC) and I got 82 CPU and 88 on GPU in a mid tower case with water cooler on the CPU).

Alienware R4 did 76 on CPU and 90 on GPU so its not so bad.

My temperature at home was 28-29 Celsius because here in Brazil its spring and its vey hot.

The undervolt did a major difference.


Enviado do meu iPhone usando Tapatalk Pro

 

You have watercooling and your GPU and CPU temps are 82 and 88?! There are people who have watercooling and run games on 5ghz with mid 70s c.

Also I'm going to be honest with you, running your CPU on 3.6ghz is no solution at all, especiall since the CPU is rated to run at 3.9ghz on stock, If your notebook cannot run stock speeds, then just give it back. Why spend so much money on a barely working system?

 

Hello Danish
On my desktop I have a NZXT Kraken X61 on CPU, idle I get like 47-42, full with Furmark I got this temperature, but for gaming is less.
The VGA is not in watercooling solution, I use the regular fan of the board.

 

That explains the GPU temp, but idle 47c is way to high, you need to check your cooling.

 

Guys,

Dell came here and replaced the motherboard (we also did a repaste with a thermaltake tg-4 paste) but unfortunately the thermal throttling still occurs, only in Core 0, the other cores are fine.

If I run with a undervolt of 0.120 I can stay under 93º in the CPU, GPU never passes 67-70 Celsius.

I will keep the ticket open and ask for a heatsink replacement.

Any other ideas?

 

Did you clean the heatsink prior to applying new paste?
Use Noctua or Thermal Grizzly Kryonaut or Phobya nano grease or Cooler Master Maker Gel nano

 

Again, just get rid of it, it's way more trouble than it's worth man.... You're only going to make yourself sad and you're supporting pure lazyness by buying Dells products.

They won't learn until it hits them where it hurts, on their bank accounts.

 

The TRIPOD mess continues although we now are soon 1 year after the release of newest Alienware models. It’s a lottery what you get.

 

well I didnt have these things in hands when the technician arrived.

I am informing Dell that the problem became worse.

 

well, I paid like 2300USD with taxes in this unit in US.
I live in Brazil and I dont if will be possible to do this here, but let´s see.

 

All the more reason to get rid of it, you paid a premium of 2.3K USD + taxes and your system doens't even work stock.

 

Guys, since I notified Dell regarding the overheating with the CPU I realized if I disable turbo boost (keep only 2.9Ghz) I dont have thermal throttle and my max temp running benchmarks was 85.

Anyway, with this setup I get low performance for a 7820HK.

So I will keep running tests.

 

Don't disable turbo unless you're on battery mode. Otherwise you paid it for nothing, you could have got a Macbook in that case.

 

Hello Guys, I´ve made contact with Dell Brazil and now they are dispatching a heatsink kit I believe


1 FRPY8 ASSY,HTSNK,W/FAN,G3,K,NEW,AW17
1 W6CFD PAD,CLG,PSX,33X12X0.2,CPU,PORT
1 06335 PAD,THRM INTFC CLNG,SINGLE

Let´s see now.

Opening the laptop for the third time.

Stefano

 

bga paired with inferior heatsink design, it is gonna happen.

 

sorry what you mean?

below are the parts that they are dispatching, are they wrong?

1 FRPY8 ASSY,HTSNK,W/FAN,G3,K,NEW,AW17
1 W6CFD PAD,CLG,PSX,33X12X0.2,CPU,PORT
1 06335 PAD,THRM INTFC CLNG,SINGLE

 

wasnt referring to the parts. was just saying bga parts in general sucked, paired up with a shared heatsink. i dont particular mind shared heatsink because i dont use GPU much but thats just me.

 

Of course it will help, anything is better than the stock toothpaste Dell use. There are 100's of pages of evidence on this forum showing that a proper repaste will significantly reduce temperatures and the core differential can even be fixed by rebalancing the heatsink if you know how.

Sorry but that's absolutely rubbish and not what is going on, there's plenty of evidence to prove this. Whether you have repasted or are stock, the GPU never get's hot enough to heat up the CPU die directly like you are suggesting, that's crazy lol.
I.e. There is no way on earth that the heatsink temperature is peaking above 98C. If anything, it may only reduce the ability for the CPU to cool itself a little bit but no where near to the level of 98C at ~1.25V running 4.2GHz. The real reason you both have such high temps is probably two fold but at least the former in that:

• Your heatsink is poorly balanced and not flat against your CPU die
• You have insufficient / poor quality thermal paste applied.

Of course, Dell should have had these issues fixed out of the box, but they're just like any other manufacturer of high end gaming laptops I'm afraid...

So fix those two issues and you'll have a fully capable beast running at 4.2GHz. I run mine daily at 4.3GHz with CPU temps always 85C or less no matter what game / what I'm doing

 

You'll have a really hard time disproving, when you clearly see how the CPU gets higher temps on Furmark than Prime95:

Spoiler

Prime95, which is a CPU only stresstest: CPU 86c on max

Spoiler

Furmark which is a GPU Only stresstest: CPU 99c on max.

Basicially what happens is that the heatpipe takes the heat from the GPU and puts it on the CPU, evidence is right in front of your nose. If you say anything else without any kind of evidence supporting your claim, then you're being dishonest.

How about you read the thread before posting? OP clearly has done the same tests as notebookcheck and has observed the exact same thing, the moment he stresses his GPU, is the same moment the CPU gets insanely hot. OP gets 70s temps in prime, once he does anything that requires the GPU to work, is the moment his CPU temps are way to high, so the one who's writing rubish is none other than you.

Let me quote for you:

As I can see you're having the same notebook, go ahead do the test yourself, lets see what you get.


Basicially what happens here is that the heatsink takes the GPU as the vaporator and the CPU copper plate as the condensor instead of puling the heat completely through, if you know how a heatpipe works, then you know that the condensor and vaporator points can be on different places, it's not always condensor right and vaporator left of the pipe, or wise versa, it's all depending on how the temperatures are that those given points in the heatpipe itself.

In dells poor attempt they basicially created a heatsink that uses the CPU copper plate as the condensor, which again showcases how terrible Dell is at constructing something. The only way to fix this is either to create ur own cooling solution or Liquid metal might help, but I refuse to tell an inexperienced person who paid 2.3K USD for a broken system to fix it with tools which he can break the system with while throwing his warranty away. If a customer paid 2.3k USD and it doesn't work, then get rid of it.

 

Danish, you are absolutely right!

For some reason the GPU temp is good while the CPU even for not stressful applications for the CPU, it gets hotter and hotter.

I´ve ran some tests in my desktop and when I run prime95, GPU keeps cool, Furmark, CPU keeps cool, but on Alienware R4 the CPU gets hot with every single thing when the GPU is running together.

If I use intel graphics and run prime95 I get 73-75 at most, but if I am on performance mode, prime95 goes under 88-92 on the CPU because the GPU is generating heat.

--------------

Just to update, now Dell have dispatched a heatsink, I am waiting Unisys(Dell provider for field services) to schedule the visit to come here where I work to do this procedure.

I bought a Thermal Grizzly Hydronaut and I will be asking the technician or even myself to use during the heatsink, thermal pad replacement. I won't use LM because it can damage even more due the elec conductivity.

 

If done correctly, LM is outstanding. I’ve yet to read a bad moment except when someone did a squirt past the tape or on the actual mobo not covered by tape. But all accidents.

 

I can't even be arsed to reply. @iunlock , can you?

 

So there is a really simple way to prove your suggestions are right and wrong.
You are correct that the heat is shared because its common sense that if two thermal sources are being sinked by the same sink that the heat would be shared.
However, that is not the cause of your overheating. There are plenty of more variables to consider that you are not considering but I am going to keep it simple.
If the sink was reacting as you are suggesting then people would not be running Firestrike Combo loops with CPU temps staying in the 70C(s) paired with GPU temps staying in the 60C(s).
Wouldn't the CPU always overheat? Meaning, wouldn't it be impossible to get good temps ever on a heat sink that cools both the CPU and GPU?
Not only does the AW have a one piece heat sink but so do many other systems.

So the bottom line is...
Like others have said prior, you for one need a repaste and also need the heat sink to make better contact with the CPU die.
I also would suggest raising the bottom of the laptop to let more breathing room. There is only so much clearance under the laptop.
Most laptops have little clearance under the laptop unless they have huge spaced feet or come with kickstands (which some do).
If this was not the solution you would not have people like myself, unlock, etc etc etc running Firestrike combo loops for an hour 7820HK(s) at 4.6Ghz still in the high 70C(s) / low 80C(s).
While I would definitely change the heat sink to where it has better contact out of the box. Once the proper contact is made with the CPU, the heatsink cools quite well.
There are tons of people that have gotten fixed better temps, so I guess that right there busts your myth.

 

Exactly what he said lol...

Sent from my SM-G935F using Tapatalk

 

Don't use p95 and Furmark because they are Power viruses and even @Prema recommends not to run them.

 

II know that Prema and many ohters don't recommend using Furmark, since it can kill your GPU (especially when Overclocked) but I've never heard about P95 being an issue, could you quote on that somewhere?

Wrong, you have to read the last part of my post. If you understand how heatpipes work, then you know why this is happening and why Liquid metal and/or other mods can actually make the condensor point from the CPU to the heatsink again. On stock, this is a problem that caused the heatpipes to make the CPU the condensor point and basicially putting the heat from the GPU on the CPU instead of the heatsinks, which again, is proven by notebookcheck and op.

You have yet to prove me wrong at all. So far your argument is:
"Some people fixed = Can't be true"

Pretty weak. I'd be interested to know what your thought is after seeing the screens and hearing OP's experience. Give me your explanation as to why the CPU doesn't go beyond 70c on a CPU stresstest but goes far over 90c on GPU tests.

To be perfectly honest with you, you have even proven me right by writing:

By writing this and the fact that the CPU actually is hotter on GPU only tests, you have basicially proven that my suspicion was on the mark. By improving the conductivity of the CPU to the copper plate (probably liquid metal only, if you know some who didnt use lm and "fixed" this, please inform me and which tim he used!), the heatpipes basicially make the condensor point where the heatsink is and not on the copperplate of the CPU.

Again to understand fully why it makes sense you need to read up on how heatpipes work and why heatpipes condensor and vaporator points can be on different points on the heatpipe rather than being 1 end vaporator and other end condensor.

 

Ok, so would you recommend OCCT and 3D mark?

 

I'd say use unigine valley and XTU Benchmarking tool mb?

 

Guys, got an email from Dell, they dont have the new heatsink in stock here in Brazil, it will arrive at the beginning of november, according them a new version (revised) will come, because they are registering several cases of overheating and if its passing 93 Celsius they need to do a repaste/change the heatsink and also the thermal pads.

In Brazil we have been officially commercialized the R4 17' and the R3 15', we dont have officialy the R13.

I bought mine in US which is legal if you pay taxes here in Brazil also, but the support is global and they have to have the parts here.

For safety I will increase my std warranty from 1 year to 3 years.

 

Yep, OCCT is a nice realworld CPU only stress test. I have lots of reference data to compare with you on this.
I view 3dmark really just as a bench marking tool but if you have the paid version it's useful for assessing game like stress data if enabled on "loop" mode and ran for at least 40 - 60mins. However it is a little dated now so I'm not sure how "real world" that will be if you run the latest games which is why I use BF1 on spectator mode (first person view only) to assess high performance CPU and GPU gaming stats over prolonged 2hrs+ periods.

 

Ummm I can't resist the temptation to actually reply to you to prove you wrong so:


Firestrike loop running at 4.4GHz with GPU overclocked to just under 2GHz also...
Note that unlike the Op all my cores are unlocked and running fully at 4.4GHz and never dipping, not even for a second. Also, unlike the Op, my system is drawing well over the CPU power limits as well. Note this is an OC'd GTX1080 which draws a lot more power than the 1070 and generates more heat therefore as well...

And also for a more intensive game like BF1 for a much longer duration (as I'm sure would be your next point)? Well here you go:



Are you still going to say "this proves nothing" despite asking for this exact proof?

 

It pulls more current than necessary thereby shortening its lifespan using p95. Ask @Mobius 1

Yes, OCCT on Linpack. Alternatively, I have used GPU Caps viewer with OpenCL CPUs and Vulkan/OpenGL based rendering to simulate real world gaming test with 8x MSAA.

Yes, Valley is a good test, also look into Superposition Benchmark running at 4k ultra or 8k low settings.