*OFFICIAL* Alienware 17 R5 Owner's Lounge

Nothing special. Just repad, repaste and undervolt.

 

It's the top of the cpu that's probably got a slight gap... Some people have tried bending the top single arm on the heatsync, I had similar results and tried bending the arm, my temps got a tad better but not much, this suggests my board could have also been the culprit... Needless to say I also gave up and returned it, maybe shims could have helped me but i didn't want to try anymore, repadding did nothing for me.

 

after 2 weeks i already noticed a huge degradation. Was really disappointed by it. Mastergel maker nano works better for me.

 

Captn.ko as you can see you are the only one!

 

And?

You can see my results...so what do you try to say?

 

I have repasted multiples units with LM (15r4) with little to average succès. I also tried artic pads as captnko or rinneh suggested but without succes also. To reduce core differential i only found success when i put more LM so far. To give you an example of how much i would say as soon as i have a miror effect that covers all the cpu. I move the qtips very lightly at the end of the process to create something that looks like a bubble of water. If i put less core 2 or 0 will have a big differential. So yes in my opinion more LM will reduce core 2 but don't put to much.

Envoyé de mon SM-G935W8 en utilisant Tapatalk

 

My first question is how do you manage to run all cores with average of 4.3GHz or even more strange 4.5GHz while playing farcry 5 which is using 100% of the GPU? Where is this massive power coming from? Do you use a special edition of R5?

With undervolt it can push cores to higher frequencies because of lower CPU power consumption but undervolt with overclock (meaning 4.5GHz) crashes the CPU because simply there is not enough voltage for that kind of frequencies. Or you have a special CPU as well? By the way, after using LM undervolt does not reduce temps at all probably because it pushes higher frequencies (but still no overclock) which covers the temp difference with higher frequencies.

I have yet to see someone else with those results which makes me wonder again.. I believe my eyes and my eyes only.

 

I always use enough LM to create a mirror effect but this time I used less I think. I did it because I always think it may move when laptop is moved. But it also has those pad like things at the edges of the contact area which I think may prevent it to spread to dangerous areas. And yes accoridng to my results, less LM bigger differential. But, if there is too much gap, whats the largest amount of LM to use? You do not want to use too much. Do you create a mirror effect on the heatsink too? They say use one drop of LM but this drop can be bigger or smaller so still no clue of how much. What do you mean by qtips? Do you mean the ones coming with the LM?


The next and probably the last thing to try will be to install the heatsink without any pads on the cpu side. To see if it the pads keeping the heatsink away. Of course, this wont be a permanent solution but I will finally know it is the pads thickness keeping the heatsink away. To explain myself, there are pads only on one side of CPU. So, the only side that may not do a good contact can only be the side with the white pads if the pads are too thick.

If there are still spikes (regardless of the differential), then there is absolutely nothing I can do because it has to do with the heatsink itself. By the way when I received my machine (I had spikes in facry 5 from the beginning), the cpu came with a thermal paste and not dell pad! However there was so much of it to cover the area between the CPU and the heatsink which shows there was too much gap from the beginning.

 

Its magic...

Magic again...

Then you made something very wrong. Without LM and with undervolt i was able to play with 4.3ghz and temps in the high 80s to 90s... now its 20 degree less...

 

You are the only magician then. And magic did not help anyone else. You probably undervolted the GPU too but the CPU power limit remains at 110W (your magic probably increased that too?). Undervolt does not reduce temps because they already low with LM and again undervolt gives more room to reach higher frequencies without overclock (but not enough voltage for OC to run at 4.5GHz). Higher frequencies cause higher temps which cover the expected 10 degrees difference expected from a -100mv undervolt. So, I have not done anything wrong.

 

Then iam a magician

I do and never made a secret about this...

Ähhhm....wrong. it lower the temps... massively...

 

Well, this is something else. Underclocked/undervolted GPU causes less heat and GPU heat is known to cause higher CPU temps because of the heatsink design. But GPU temps have never been a problem. So, results with undervolted GPU is a new basis to work on. But I want to sort out the spikes first which is what I am mainly concerned on. With a non underclocked/undervolted GPU, I may never get rid of the CPU temp spikes in farcry 5 anyways because they may be due to heat transfered to CPU side instantly from the GPU itself.

 

Wow... thats new to me

And: undervoltet...Not underclocked

Come on... all these things had been said so many times in this thread...most of my screens show my GPU clock and voltage...

 

It is very simple and helps stop guessing: REMOVE ALL PADS just leave GPU and CPU and retry. See how that goes. You do not have to run 48h endurance tests, just power it up and have a quick look. If you see the CPU temps different then you know the culprit must be the pads!

Welcome !

That is true, undervolting allows you to run higher clocks at the same watts. Or allows you to run same clocks at lower watts, hence less heat. In general undervolt helps , if your CPU is stable of course.

 

Why should i run my GPU Stock @ 1860mhz if i can run it with 20-30w less and @ 1900 at the same time...

 

I will tell you this ,which I also posted a few pages up. Please read carefully. This laptop has no, I repeat NONE, cast alloy subframe to create a rigid and solid platform on which to mount the motherboard, heatsink and fans.

ActuallyI lie. There is a ~1.5" wide cast alloy frame from the screen hinges right up to top row of keys. The rest is all plastic, soft and bendy! Behind the keyboard you have a metal plate, which is thin and flimsy, and warps like a spring! On top of that rickety structure you mount heatsink and fans, but wait, some screws (on the fans) are bolted on the thin alloy strip subframe, which is sturdy, and most other screws are on this flimsy metal keyboard back plate. At that time you have the laptop is resting on its screen back cover which is not even flat !

Whilst re-assembling I noticed my motherboard would seesaw and rock over some central point which made me think there was a left over screw jammed under the motherboard but in fact was the warped keyboard back plate. Warped. And pushed out in places by a bulging keyboard which is also a soft structure.

If you ignore that rocking and screw it down anyway, you will be stretching and warping the motherboard and the heatsink. Warping the heatsink which only has 3 screws on leaf springs on the CPU side means a bad contact, which could be made worse each time you open and close the lid or move the laptop!

The ONLY solid structure in that assembly is the heatsink subframe which is steel (I hope) and incorporated into the motherboard.

So you need to

a) rest laptop on some thick towel to absorb the screen cover's convex shape.
b) handle the motherboard with extreme care and position and feel it to make sure all the retaining screws could be bolted down with no stretching of the motherboard. If you have to push even 2-3 mm it means there is a warp.
c) if you have a warp you must remove the keyboard back plate, and bend it to make sure it is as much level as is possible. You better tighten the screws like you do an car engine's block head.
d) reposition the motherboard making sure it rests on the mounting screw poles by its own weight and not by pushing it down. While you are doing this the DC wire connector is in the way and you musy pull it straight up so it is not causing a disruption. Same as the wifi lead. You want the motherboard to rest on the mounting screw sockets perfectly and without even the slightest push. When you lift the motherboard handle it from the heatsinks if you can, not from the PCB itself.

 

Some photos so you may see. Unfortunately I did not take a photo of the flexible keyboard and its flexible back plate. Another time.

Photo 1: the FLEXIBLE keyboard supported with a few (not nearly enough) tabs on one side, I have inserted a pair of tweezers SIDEWAYS with almost 0 pressure. This is the ESC key under there and yes it, and the keys around it, would sink down together when any key in the area was pressed. Other users have also reported this problem. No **** Sherlock!

Photo 2: the cast allow frame at the top from the hinges to the keyboard, what's that 1.5"? The rest is all softish plastic. Notice the sparsely placed keyboard tabs. I mean seriously?! Do tabs cost a lot ? The black material is the keyboard backing, it is like a soft cushion !!!! It bends, and flexes, and can be depressed like a sponge. notice the sockets onto which we mount the motherboard or the plastic flimsy cover on top of the motherboard.

Photo 3: the keyboard back plate in place. Notice the copper coloured heatshield. Not man enough, still 45C on the top of the keys, left 'Z' and right "SHIFT". That metal backplate looks the business on this photo but when you take it out it is thin and bendy. It should be much thicker and should span from the hinges down past the battery... But as it is it provides very bad support to the motherboard because it is impossible to ensure it is completely level. It is like building on sand.

 

When i apply LM i start with one drop and spread it over the CPU. At this point there are no miror effect. Then i add half a drop and srpead it over the cpu very gently so it looks like a water drop and i obtain a miror effect. If i put to much i use the other side of the Q-tip to remove a little. I do not create a miror effect on the heatsink side. Just enough to cover it. This tehcnique give me almost no core differential after an undervolt of 160. If i put less i get a core differential of 10 to 15C on core 2 always. But even this little succes give me temp of 95C playing farcry 5 at a constant 3,9GHZ. I never was abble to achieve anything close to captnko. My next attemp will be to try with the coper plate between the heatsink on cpu and gpu. And maybe, then, i will retry a repad with fujipoly. Here's a list of my failed attemp:
1- Bending arm 1 (no improvement on the new 15r4 but works on 15r3)
2- Artic pads (no improvement in temps, very hard to install on the heatsink since they do not stick enough, maybe it's my fault)
3- Putting not enought LM on CPU (core differantial is 10 to 15 C, then open it up and add half a drop and problem solved)

 

It is not so simple. Temps during idle (in the 40s) or when during normal stuff are good even now for all cores. LM makes an instant difference compared to thermal paste. Problems start with farcry 5 with temp spikes for certain cores.

Simply because heat is not an issue for GPU and you have to run those apps that reduce the volts on the startup along with throttlestop. The only good point I could think is that GPU may cause those 1sec random temps spikes to CPU during heavy gaming because of heat transfer to the CPU side.

As you are repasting R4, can you please tell me if the white pads stated as 0.1mm on iunlocks guide are really 0.1mm or thicker on R4? On my R5 they were all at least 1.0mm. I have never seen such a thin pad and I doubt such a thin pad even exists.

 

and? I put those apps into automatic Startup and thats it... with undervolt and 1900mhz (and VMem OC) my Gtx 1080 is faster than stock, cooler than stock and it helps my cpu to keep its head cooler... its a win win win

 

Ermm, can we possibly compare the AW 17 with the AW15 ? Don't they use different heatsinks and fans? I would expect the AW17 to have larger fans and heatsinks etc?iiiiiiooooooooo

Yesterday I undervolted and OC'ed the GPU trying to get more FPS in Far Cry 5. From what I see, by default it runs core at 1860MHz, memory at 1251.5MHz and 1.050V-1.062V. I OC'ed it and ran FC5 for a couple of hours at 2000MHz, memory 1377MHz, 1.000V. Maybe I can OC it more because I am still getting slowness in FC5. Or maybe turn the graphics settings down in FC5. Will run the FC5 video benchmarks now to get an idea.

 

I noticed if I play FC5 with -140mV undervolt maximum temps reach 89C maximum (on one of the problematic cores) without spikes. I used 3820x2160 and ultra settings. How do you undervolt GPU? MSI afterburner or nvidia inspector? I checked them and they both do not give undervolt option...

 

Afterburner. You can find a guide in this forum.

 

15R3 (7700hq and 1070GTX) same heatsink as 17R4 and white pads are 0.1mm (looks like a peace of paper) and 15R4 same heatsink as 17R5 and pads are light green at 1mm for the 8750hq and 1070GTX. They are thicker on the latest model. I can assure you that the 0.1mm really exist on previous generation.

 

Do you have a link to buy one of those 0.1mm pads? I have seen 0.5mm minimum on ebay etc. Those pads are thicker on R5 (1mm or more on R5, measured by myself) but using thinner ones may resolve CPU/heatsink gap.

R4 has same heatsink but one of the CPU screws is placed in a different position. This may cause huge differences between R4 and R5 regardin CPU/heatsink gap or contact.

 

Do you have an appropriate one? I have seen guide videos from the forum but I do not see how to move the graph to the left (meaning lower volts). They only say how to move it up (meaning higher voltage in lower frequencies) which is unlocked anyways.

By the way, the temps you report for FC5 are with CPU undervolt?

 

just play with the curve, Its not very difficult.

 

I believe they are the same since they use the same motherboard. That's why i always mention that this is for my system wich is 1070GTX and 8750HQ. Most of the people here have the 1080GTX and I9 chips and it might be little different.

 

Thanks for not answering my questions..

 

MSI aftrerburner: you need to unlock the voltage settings first. Type "youtube msi afterburner mr fox" and pay very close attention. On the youtube page, check the description and copy the settings.

 

no Problem...

come on... a bit own initiative must be possible. I have to seach the guides too...

as i said a thousand times: yes. just look at the voltages. Why do i post all my screens with curves and HWinfo if no one is looking at them?

 

VGA tests in FarCry5
default core 1860MHz, RAM 1251.5MHz, Voltage 1.0620V = FPS 38/43/53 - 38/44/54
boost : 2055MHz (+200), 1377MHz (+500), 1.062V = FPS 41/47/59

Not really a great improvement, maybe not even worth starting the MSI Afterburner at all

Now I will try the undervolting stability.

 

OK, I see, I can see a lot of empty space on the left of the motherboard, the GPU side, so you could fit this motherboard inside the 15" chassis, but how do you fit the heatpipes, heatsink and fan which sit well outside the 15" dimensions?

 

Further tests with MSI Afterburner: undervolted to 950mV at 1885MHz and it seems stable. Maybe that setting will yield a bit less heat.

 

This video shows how to take the graphs points up and down. To undervolt you need to move the graph to the left?

Anyone knows how to bypass the current limit throttling reported in Intel XTU? It starts throttling the CPU even at TDP of 73W while it is supposed to limit at 110W? This current limit throttle is obviously the reason that some CPU benchmarks (OCCT and prime95) run tests with core frequencies much lower than 4.3GHz

 

Be fair, it is completely unintuitive, weird, and must have been written by equally weird people. Everything on this GUI is crap - what does what and where do you click a little bit like TS. The piece de la resistance must be the voltage curves! Voltage is shown on the X axis, and in Western society it is almost given that the variable x is shown on the X-axis (hence its name), and its function, f(x) otherwise also known as y, is shown on the aptly named, Y axis. Unless I have missed a turn or two, in which case I apologise, the purpose of the curve is to select the appropriate voltage for any given frequency, and therefore frequency F is the variable and should be shown on the X axis.

 

not realy ...if you are able to press CTRL+F rest is self explanatory

 

No, what you do is you select a voltage that you would want your GPU to run at its max frequency, which you must know, or you concurrently run the stress test from GPU (click on that self explanatory '?' on the front page of GPU-z) so say your max F is 1880MHz (default settings) you then choose, say, 950mV to be the voltage at that frequency so you click on the "dot" corresponding to 950mV and you drag it up (you can also use arrow keys) to 1880MHz. Then you press 'L', then you go back to the main window and click on apply. Easy huh?

When you are stuck at 73W, what are the CPU temps? If they are 88C-93C then this is the "invisible power throttle" we have been talking about lately.

 

It could be as easy as this mess is due you upgraded to latest bios version (Dell released and throw out Cancer code)... Aka even if you reverted back to older firmware version, you're stuck with what the latest firmware from Dell implemented of code (Not reversible). I expect you are on latest Microcode now (0x96). I have said this everywhere. OS patches, software, drivers or firmware... Never be others GP.


A working system!! Stock clocks.

Spoiler: Stock clocks.

 

If you read why I said that, voltage is shown on the X axis. That is weird for anyone who takes x to be the variable and y to be its function, you know f(x)... We do not plot the variable on the y axis, not unless we are drunk maybe

To be fair, it throttles at 94C and results in the CPU hovering around 88C-93C during stress tests. Would you want your CPU to be at 94C and over? I think it is a sensible value and sensible system.

 

If you get 94C with stock or medium Overclock... Fix the problem. All today's Processors have own safety (100C). The OEM's shouldn't screw up with cancer code.

As you can see.... Dell walk in the blind. From previous firmware. All this mess will in the end screw up how it will works.

 

Doofus, the current limit throttling has nothing to do with the thermal throttling (or power throttle) you describe about 93C. At least not anymore. Temps while this throttle occurs, are in the 80s the most (so not even touching 90s). You rarely get temps close to 90s with LM (so not even close to 93C) which I am testing with. I suspected this is a Dell bios implementation. I reverted back to even 1.0.6 bios (before they starting touching the CPU performance) but the current limit throttling remains.

If what Papusan said about not reversible is true, then we are stucked. And if some geniuses in here knew that this new irreversible firmware causes CPU lower frequencies (indepedently of temps), they never said anything and are proud of having higher scores, thats a shame.

I suspected this new current limit throttling is a bios implementation in firmware but I thought it would be reversible when installing older bios?

I tried old 1.0.6 bios and Intel XTU settings for turbo boost power mode, even tried to disable it. No matter what. The current limit throttle happens as low as 63W (approximately) with temps in 70s or 80s the max. Only CB manages to slightly push higher frequencies for all cores probably because it uses less Watt with undervolt.

The only solution now would probably be if someone hacked the EC firmware but I dont think anyone will bother anyways. And the new EC firmware is obviously the reason frequencies of CPU drop to 800MHz even with high performance profile in windows. Simply because EC firmware bypasses all those user defined settings.

 

1.1.6 was necessary to get AW CC working correctly. And with 1.1.6 i have the full 110w available. So the mess started after this update.

 

I know. But read the context in my statement.
And what will come when the RMA number being too high? I'ts quite normal that the OEM's will put in more cancer code after a while to reduce the RMA numbers. And people will be forced to upgrade to the cancer code due needed (new) features, safety problems +++.

 

How could they know without beeing able to test this bios? so please no accusations. maybe they are just careful and do not rush on every update?

 

So you knew this happened with new bios but never said not to install. And thats why you always asking arrogantly on what CPU frequencies. Thats a shame.

So, you do not actually expect anyone to reach your score anymore. So, I understand why you do not even bother helping. And I bet all new machines will come with the new EC firmware installed.

And it is not accusation. You knew it.

Doofus do you have the latest bios 1.2.1 installed too?

You probably knew it from inside Dell?

 

Haha no.
Papusan posted a Screenshot about 1.2.1 and 1.1.6 and its,,thermal,, improvements... i only said that i have 1.1.6 and no problems and concluded that the mess must be started after this Update... where do you read that i knew that before?

But i see... you are looking for a guilty one. You can not accept that its your damn fault that you rush on every new update...
At this point i will not try to help you anymore. Have fun

 

I am not going to say where and why. Lets see what Dell is going to release next.

 

Hahahahahah stop making a fool of yourself

Let the witch-hunt begin!

 

Same Bios version (1.2.1) used for the reviewed AW15R4 with the unlocked i9-8950Hk.


Test Alienware 15 R4 (i9-8950HK - GTX 1070)Notebookcheck.com