*OFFICIAL* Alienware 17 R5 Owner's Lounge

1400+ should be average.

 

here is another run. average maximum frequencies are in fact at 4.290GHz

https://imgur.com/a/ArKDo7D

 

In CB you tripped the PL2 limit two times (this isn't normal since your PL2 limit is 110w), but it didn't impact the clock speed.

From here it is a mater of fine tuning and luck in the BGA silicon lottery! You need to test for your max stable UV, you have to reduce the nr. of running services, use Phoenix's guide for optimizing windows.

You can also increase the current limit (although, this wasn't what limited your performance in CB, it was only triggered in prime and considering how stressful prime is, I would consider it normal behavior).

You need to run CBr15 in a loop or just manually run CBr15 without pause between sessions to have a better picture of the avr. performance in longer runs (capture the scores for each consecutive run).

 

Ok just fixed the OCC. Regarding how easily it goes up and down, it looks like I am not far away

Back to business. Raz8020 what can I do to fix that?

 

Ok thank you for your time. I could spend hours on that. But, I am not trying to break the benchmarks. I am only trying to make sure my machine is generally working well. This is the only reason I started running the benchmarks.

 

It could still be a considerable amount of performance that you're leaving on the table (depending on the usage), but I understand if you don't want or have the time to further tweak you laptop.

 

captn.ko could you please play farcry 5 for 1 hour or so and post your temps? Thanks

 

already posted FC5 temps. @Aristotelhs2060



only 15 minutes but the temperatures had settled

 

Trust me they can go much higher even with LM. I would suggest play for 1 hour. Mine were at 83C maximum as yours but today the case is different (Even though average is at 80s I get spikes). I stressed it more? How could I? Temps under benchmarks are still as posted before without thermal throttle or temp spikes.

By the way, did you replace all the thermal pads including the white ones? Did you use the same thickness that iunlock stated? By the way, there are no 0.1mm thermal pads. What brand is your heatsink?

 

I trust myself. I played hours with this settings. No higher temps then that what you can see.

yes

CCI

 

Playing farcry 5 only for 15 minutes does not cause temp spikes.. You can only try playing for 1 hour or so to prove what you say. This is a way to help others too. To see if something is still wrong with mine or others. I understand that it may not be your favourite game but accoridng to my tests, it is the only one that can stress to the limits. As stated before all the benchmarks (prime95, cinebench, occt) cannot cause such high CPU temps. It may be because GPU temps are rising (not exceeding 71C after LM though) too

 

I have proven all my other statements.... I will not start this mess again...believe me or not. Your decision.

 

Ok. Did you replace the white pads too?

 

nope. I didn't touched the white and green on the cpu side.

 

I managed to fix the samsung 960 temps using a new heatsink! I had to do a heatsink mod for one of them because of limited space (this machine is made for mods).
Maximum of 70C during benchmarks! 65C maximum for the 256GB OS drive and hardly 70 for the slightly faster 512GB (use to install and run games from). This used to be 88C with my previous heatsink. That is impressive to say the least.

Here are my temps after benchmarking both:

https://imgur.com/a/wkgJ6Bl

 

This guy feels confident that copper shims are gonna get better temps... What do you guys think? He's using .4mm

 

if you can drink straight from the bottle why use a straw

 

This guy never posted his temps after doing that and before. It would be interesting to see what it does on i9 but I do not offer to try that...

By the way. If a shim of this thickness can fit between heatsink and CPU/GPU then this is a huge distance between heatsink and CPU/GPU right?

 

There are evwn 0,1mm shims. So in this case you would just create a little bit of extra heatsink pressure which can be beneficial.

 

I tested stock clocks. 5 rounds 1450 as lowest. 1459 as max. 0.996/1.0v load voltage.


Btw. 49x


As a rare curiosity from Papusan camp 40x all cores (No tweaked voltage - Same as I use for stock clocks).

 

I tried to give a comparison on another 7820HK but I have the same power limit throttle issue as @Aristotelhs2060 see:


Any idea how I fix that so I can run it again for you?

Check my signature

@Aristotelhs2060

 

I have been away repasting and life etc.
1) Dell have confirmed the secret power throttles are at the BIOS, 1.1.6 and 1.2.1 they said, and you cannot avoid it. This secret power throttle kicks in a few seconds after the CPU has thermally throttled, eg > 99C. After 2-3 seconds it will limit the power to bring the temps down to <= 93C. You will see the thermal throttle gone and the power throttle commencing. The system is not perfect, in the sense you can fool it by stretching only 1-2 cores at a time, and in that situation the power throttle is unable to stop random cores being hit suddenly and then you will see thermal throttles too.
The reason, I suspect, that captn.ko has not noticed it is because his CPU does not throttle thermally, which is the trigger for this system ?
2) My PCH reached 100C last evening while gaming, causing CPU = 800 MHz with no throttles (as reported by HW Info), and severe game stuttering. Therefore I took the plunge today and repasted. Twice. I also added an aluminium plate over the PCH, 1 inch square glued on with Arctic two part thermal glue, and an aluminium plate over the NVME SSD held down with kapton tape and MX-4.
I discovered a few things, not good.
As I had the laptop open I took the opportunity to repair the keyboard, my ESC key and the keys around it would sink visibly when a near by key was pressed. I thought we were dealing with a loose screw.
Took everything apart and discovered that the keyboard unit itself is soft and quite flexible and on its own cannot possibly provide support. Behind it there is a steel plate, unfortunately it is too thin, it wobbles and flexes so bad. To make matters much worse there are not enough tabs to hold the keyboard unit in place so when you press ESC/F1/1 etc the whole unit sinks in because the nearest tab is around the F4 key. Same around the '8' key.
Terrible design. So I added some tape as a filler in-between keyboard unit and back plate. That fixed it and my ESC is now solid, albeit there is still some imperceptible movement around the 8 key.
Second major issue is the whole construction, it is so fragile it defies belief. This keyboard back plate I just described also doubles as the motherboard base. We are trying therefore to screw in a motherboard with heatsinks over a flexible almost paper thin backplate, over a wobbly keyboard, the whole thing supported by a very flexible plastic/carbon fibre frame.
Therefore as I was trying to mount the complete motherboard onto the back plate, it rocked something terrible, it was like a seesaw. Trying to screw it down would bend the ends of the motherboard 3-4 mm, and the only thing holding it together was the heatsink. After some panic thinking the heatsink somehow had warped, even though I had treated it like a baby, I thought I had left a screw lying around and the whole motherboard was rocking over it. But no. After so much testing and further disassembly, I eyed the "steel" back plate and it was warped - actually it did not have a straight line in it. I tried to flex it back to some reasonable shape but it flexes back and of course we cannot go wild on it. What a terrible design, this back plate ought to be thicker and sturdier. Anyway after some flexing I put it all together and the motherboard was not rocking anymore, was stable.
So now we go to the actual heatsink and thermal paste, as others have already shown the GPU uses a "stamp" thermal pad and the CPU a thermal paste. Both were dried up even though the CPU looked "wet" it actually was quite dry. There were large areas over the GPU that were uncovered, the CPU was better. There are a lot of thermal pads of varying sizes and shapes, from whatg looks like terribly thin (0.1mm?) to 1mm thick. Most pads had solid impressions on them meanng they were making good contact. The pads looked to be soft, oily, left stains on the ICs, and you could not stretch them, they were more like paste than pads.
So far I am not sure where I stand. I think I have slightly better thermals but nothing too exciting. My best CB score was 1394 when the machine was cool, it then settles down to 1320-1340, and to achieve it I have a -140mv and 4.3GHz locked. The SSD seems to be at 72C with everything else maxed out, the GPU around 70C, the CPU at 4GHz and 73C and 24W. The PCH ius disappointing at over 90C, but with the undervolt and locking the CPU down to 35x we stay below below 88C and yesterday it was 100C. I think I need a better PCH heatsink...
As far as the CPU is concerned I may need to use LM or some other thermal grease. I visibly saw the dried up paste and thought it was an easy fix, but am not so sure. I will be running more tests.
Another thing I will try to do is get two small extension leads for the fans so that I do not have to remove the motherboard while repasting and that ought to make life easier.

 

The motherboard is screwed onto a metal frame.....not into a flexible palmrest. But the bottom of the motherboard is loose until you also assemble the mid frame. After that it is incredible sturdy. The keyboard plate sounds off. If it is the same as the 15R3/17R4 it should be rock solid with zero flex. With everything assembled it is a very sturdy package.

Dell thermal pads have often a fibreglass fabric layers between the silicone. Thats why they cannot be stretched. Also hard to remove. Needs through cleaning of the heatsink after removal.

 

I found some notes:

before repaste:

ambient 28C - P95 10 threads - temps 87C-95C, avg 91C ?, TDP around 64W.. power throttled

Today, after repaste:

ambient 28C - P95 10 threads - temps 87C-95C, avg 91C ?, TDP around 75W-80W invisible throttle

So I have gained 10-15W it seems



I have to correct my cut'n'paste! Today's temps strictly below 93 C, not 95C as I wrote above. ​

 

Doofus if you want to see real difference with temps you have to use LM. At least for the CPU. I have used both thermal grizzly and phobya nanogrease which are regarded the best but only LM can reduce the temps and keep them under 90s

 

How about the PCH? This is currently exceeding 93C with Far Cry, I have it running for a while now...

 

My PCH mod worked well. I can send you pics if you want. Used 0.1mm Gelid pad and a large copper shim. I can check the thickness and size of shim because I cannot remember. I used an instant glue on 4 corners of shim and sticked on the 3M tape covering the sides of the PCH chip. I used a light pressure to achieve good contact while sticking.

 

This throttling might be caused by CPU's max amperage limit and AFAIK it's not possible to set IccMax in BIOS.
That's why I use XTU to raise IccMax to 250.00A, and at the same time PLs.
Try to set all OC values to 0 in BIOS and set IccMax and PLs in XTU. Then you can close XTU (and its service, i.e. in cmd run: net stop XTU3SERVICE) and operate everything else in ThrottleStop.

 

What temps are you getting after hours of Far Cry ? I have reached 93C so far...

 

I checked again after playing farcry 5 for about 1 hour (3820x2160 resolution and ultra settings). 78C maximum instant temp for PCH.

Average CPU cores temps at high 70s (average maximum 79C for CPU cores and mid 70s for cpu package). Unfortunately, high temp spikes for two of the cores (always only core#0 and core#2 have instant temps over 90 only with farcry 5) and CPU package . I have to make a log to see how often those high temp spikes are happening. I may have to repaste. Temps drop by 30C in secs. How is that even possible?

 

What clock speed? What room temp?

 

PCH is GPU buddy, all GPU! Trust me. Underclock that mofo and it'll drop your PCH 5-10C

Sent from my SM-G960F using Tapatalk

 

Platform Controller Hub If you want more info. Just tag me.

 

Without Undervolt or overclock. Stock clocks. 28 degrees room temperature. Seems like at least this mod worked for me. LM worked too but I get those annoying temp spikes but only in farcry 5 (not under benchmarks) and only two of the cores (always the same cores).

I am going to repaste probably tomorrow and reuse stock white pads. High temp spikes only for two of the cores but really good average in farcry 5 (with the settings stated before) means bad CPU/heatsink contact? I cannot figure out what spikes on only two cores can be caused from. Does it mean LM is not equally spread on the CPU chip? This is strange because just after LM repasting, I did not have any spikes till now.

By the way if the thermal pads (not the CPU/heatsink contact) do not have good contact with the other motherboard chips, does it have any effect on the temps? I mean do we see any of those other chips' temps on Hwinfo?

 

Thanks will give it a go

Sent from my SM-G960F using Tapatalk

 

Nope. Good Cpu temp doesn't always mean the VRM temp is top notch.

 

I measured the temp on two VRMs (I think that's what they were) which were half-exposed by the GPU heatsink with the trapezium shape, they were at 56C and the GPU at 60C at the time. I would guess that CPU and GPU is much more important. The PCH is cooking however and typically when the GPU is very busy (lots of watts). It has a 1"x1" aluminium plate over it and it did nothing it seems. I measured the plate at 65C while the PCH reported 80C - I wonder if this reported temp is correct ...

 

Your spiking on 0 and 2 (and probably 4 is close behind) because your heatsync is not balanced properly in the north side of the die... Which also means you will want to be very careful moving your laptop (especially vertically in a bag) while it's still warm as there is a chance LM can slip out....

 

In my experience one or two cores can be utilised and they will spike, especially if the fans are quiet at the moment. I do not think it is anything to worry about. Since we have 6 different temp readings, the sensors must be very sensitive to instantaneous changes else they would all be showing the same temps all the time, more or less, as the silicon around them heats up even if the source is only one or two cores. If the temperature differences persist then there may be an issue.

 

the temperature is hotter on the inside of the chip ofcourse. But an aluminum sheet doesnt do much it makes the surface to cool slightly larger but you are also adding resistance by using paste. Use a proper heatsink while leaving some room for air to flow.

 

Hi, these spikes are a classic sign of the heatsink being imbalanced. Which cores are spiking and it may give you An indication as to how the heatsink is imbalanced. LM is amazing stuff but it only works best when the heatsink is perfectly balanced. For any such imbalance you'll want kyronaut instead.

Sent from my SM-G960F using Tapatalk

 

Phoyba nanogrease extreme*

 

I have used both phobya and kryonaut. Both ineffective on this machine. Repasted with LM. Still spikes on farcry 5. I have used both original (white) and gelid pads of same thickness at the CPU side. It is not imbalanced. This machine is incapable of sending heat away. And this is about hardware not improper (imbalanced) heatsink.

If anyone had spikes and managed to sort them out on R5 please say so. If not please dont just assume. And try farcry 5 on ultra settings, not benchmarks

Always core#0, core#2 and core#4. So where is it imbalanced? White pad side?

 

If your CPU cores have a large temperature difference between them it means the heatsink is not properly balanced. This can be because the heatsink itself is warped, the pads on one side are too low or too high and last but not least it can happen that some components are higher than others due to manufacturings tolerances.

 

Urgh I have such bad experiences with this paste. First weeks it is perfect but it degrades so quickly if the pressure isnt perfect. You wouldnt expect it of such thick paste but it just evaporates when temps are near 85c.

 

here no spikes at all. 90+ degree only with 4.5ghz+ allcore+ 99% GPU Load

FC temps @4.3ghz



there are 3 Drops in the graphs... there i died XD 1 core i usualy 3-4 degree warmer than the other 5. Thats ok for me.

RT 25 degree

 

Well, I give. I tried everything spending hours and hours. If this machine dies because of temps make it be. Captn.ko you even have a machine from outer space or you are a Dell rep. Your machine is the only one in the space. And if there is no one else this makes me wonder...

You have a perfectly balanced heatsink without doing anything special? right...

The only thing I can think about is that the 3M tape covering the CPU surface causes some pressure on the heatsink (I doubt it because I saw the high of it watching at the side) or the 1mm white pad used by Dell on R5 should be 0.1mm as on iunlocks guide (which is for R4). Or may have to use more LM

 

yes there is a large temperature difference. more than 15C between lowest and largest temp of cores. If it is core#0, core#2 and core#4 which side should it be. This sounds like the whole CPU. But temps drop by 30C for all of them in 1 or 2 secs.


Thus I am still getting low frequencies during idle (800MHz) and under high performance windows profile. I have unlocked the power settings and used the settings recommended in another thread. I also made a fresh format reinstalling everything and happened the same happened immediately..

 

Hi guys

iam from Germany and follow all your post since the begining I read of a lot of problems causing by the bad heatsink or paste from stock but I still ordered the same "****" right now
Here are my most important specs:
Alienware 17 R5
i9 8950HK
4K Screen
GTX1080
32gb Ram
Killer 1435
512gb ssd
1Tb Hdd
win10pro

i hope i have more luck with it. But i also keep you informed about my temps and scores.
and hope that you all will fix the problems you have. Good luck.

 

How quick?? Couple weeks?