*OFFICIAL* Alienware 17 R5 Owner's Lounge

Just curious, I was looking around for some info likd iunlock provided for the 15R3/17R4 - but how does the 17R5 perform with the 8950HK? Does the newer Alienware also need a repaste and change of pads + mods to make the 8950HK and 8750K usable without throttling?

 

I can say this - I believe 8750 max turbo (at stock clocks) is 3.7. My 8950 needed undervolting AND underclocking to right around 3.6-3.7 Ghz max clock (on six cores) to run without throttling. Now that I have re-pasted and re-padded (not sure pads helped, just changed while I was in there), I can get 4.2 stable on all six cores, if room temperature is 78 or below. These tests were WITHOUT GPU running.

I will also say - without a clue of temperatures, I did not experience throttling when I first used the unit. I played modern titles like GTA 5 and Shadow of the Tomb Raider all the way through with no mods for 10-20 hours on QHD graphics at Ultra everything, no issues. Now, when I started researching, I saw temps were in the 90s, and it may have been throttling somewhat, just not noticeable in those games. My max Cinebench score was only 980 at that time, which got me to research through this forum much more than I had previously. I think the paste then got worse, because at first, I was definitely able to hold 4.0 or 3.9 Ghz with -120 mV undervolt, but only 3.6-3.7 a few weeks later.

So, I think with the 8750 at a max of 3.7 Ghz, it will more likely be usable out of the box. Perhaps even with no throttling. But temperatures will be high. I didn't think I'd mind high temps, but if it degrades the paste, and makes the fans run longer/louder, you'll probably want to re-paste anwyay, eventually.

I know you were looking for a Yes/No answer but it is truly a conditional response based on how much you care about high temps, or if you just mean, will it work with no noticeable slowdowns. I think it will work, just not sure how long paste job lasts. (again, mine seemed to degrade after just a few weeks)

If I were buying again new, I'd strongly advise myself, if I am sticking with AW, to get it re-pasted from CUKUsa of HID or any of the other ones that sell AW. Just make sure you can still get the full Dell warranty if you go that route. The thing is, I don't think you can negotiate with them (being a reseller), so you will pay retail price, plus their add-on for re-pasting and any upgrades they do; with Dell, you can chat or call and it's guaranteed you're going to get 10%, maybe much more, off the price the checkout tells you! So again, it depends if you don't care so much about price, just want it to arrive ready to go, re-pasted, or if you don't mind doing that yourself. Considering how paste jobs may deteriorate in as fast as 6 months no matter what TIM (thermal interface material) is used, I think it's best to get to know one's laptop with these machines, unless you have someone close-by or don't mind mailing it in to have that service re-done if and when the time comes.

 

You can run other ssd benchmarks as well outside Crystal Diskmark. Run them 3 or 4 times in a row and put up the average scores. On both of your ssd's. You can also switch slots and see if you get same results. Put OS on the opposite ssd etc.
Anvil's Storage Utilities v1.1 2014 - ATTO Disk Benchmark v4.00.0f2 - AS SSD Benchmark v2.0.6821.41776

Maybe start here... http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-120#post-10755810

Core i7-8850H - Intel - WikiChip

 

@Aristotelhs2060 I think you will find this video useful; wish I had found this early on, summarizes a lot of stuff I read in many, many forums into one succinct video. To summarize, the graphite pads are very thin - could not see that in the product photos. I thought it was at least 0.5mm, looks like it's more like 0.1 mm, paper thin. He had to adjust it 5 or more times in one machine to get decent results. Many times he got 100 degrees at IDLE due to its poor fitment. He covers how the dual-purpose heat sinks don't line up well, hard to adjust properly with the tripod attachment the CPU has (stuff this forum has covered, but a nice summary).

A lot of commenters said they got great results though. Seems like each person has to try it on their own machine. Different heat sinks have different levels of fitting the CPU well. And it might take many attempts without visual aid since the tolerances are so small to get it to cool at all. I guess if I hook up my fan extensions, I can at least run the laptop without putting everything back on and maybe even be able to tighten the different heat sink mount screws while it runs (asking for trouble there, haha! ok, maybe shutdown in between).

 

I will have a look at the video later as I am new to graphite pads anyway. It would be really helpful if you could try the graphite. With LM I was getting about 92C max with CB while at 110W with about -150mV UV ( I was getting a different current limit throttle when trying to exceed to power limit, probably because of the low UV capabilities on that CPU) . However, as you said the biggest deal is to have a CPU that can UV well (this causes the biggest drops in temps). It was only my third system that I can get UV up to -0.211 mV. I believe this is causing the good benchmark scores rather than the thermal paste. I can only imagine what an LM could do on my current system with that high UV, especially when I could possibly eliminate the current limit throttle with the current high UV. But, I do not plan to use LM again for the reasons explained before.

Trying LM with graphite is a bit risky I would say especially because no one has tried this before but on the other hand if the CPU does not make a good contact with the heatsink, maybe this is the next thing to try. But start from single graphite pad first. Yes, the graphite pad is too thin. I had a graphite pad (I do not know where I left it) but I remember I was not sure how to apply it and it was too thin. Now I ordered a IC graphite (one for CPU and one for GPU) and the IC contact test kit. If the last one really works, this is going to be the starting point to finally be able to check the CPU areas that do not have good contact with the heatsink supposedly causing the temp spikes (on all the systems this has always been core 1 in addition to one other of the cores even 0 or 2 but this may just be because those cores are the primarily used by programs).

 

I appreciate the info very much. I completely understand what you mean , tldr is the 8950HK is a waste with the Alienware HSF haha.

I did search a bit, you didn't have to get petty. Maybe if I used a forum app I'd be easier to navigate. But anyway, hell, even a single sentence from someone reputable would've sufficed and often times I trust the info of someone reliable with their text findings over YouTube channels and questionable tech pages that are 'reviewing' for both their own benefit and for ours. However, some people do 'review' for their own benefit only, if you know what I mean.

 

Hi man! I have the i9 8950hk, it will not perform well if you don't take it apart for repaste and repad(just in case is needed), I did both (LM conductonaut and some stock pads and some Alphacool 14 W/mK), I was getting more than 100°C benchmarking and playing before (thermal throttling), now I'm getting on Overwatch on Ultra 70°C - 75°C stable at 4.3 on six cores with -160mv UV on Core and Cache, my Core is able to go more than that but not the cache, right now I'm using XTU for the UV (I don't remember why), but I know I can do better playing around with ThrottleStop, just need some time to start testing again. Also if I UV the GPU I will get better temps I'm sure!

So.... Are you ok doing the repaste and repad? Are you ok if your CPU get just to 4.3 because of the power limit? Yes? Then go for it! I'm more than happy with my system, my heatsink was really bad balanced, I had to bend it a little to acquire better CPU/GPU contact, it was terrible... but right now is fine, I think maybe I can put some more pads here and there, some heatsinks on the SSD (gets really hot on Benchmarks), I will change the SSD anyways, so.... Yes that's it, you pay more than 3k (euros) for doing a new product that you need to take apart and make some work....

Enviado desde mi Pixel 2 XL mediante Tapatalk

 

@Aristotelhs2060 I just finished my second re-paste. The second time around, I was able to make more observations. I'll just give a list of bullet points and maybe this will help others or garner some feedback:

1. I noticed this time that the seven attachment points/arms for the heat sink are numbered; while I did opposite tightening last time which is correct overall, this time I followed the exact numbers. It looks like they want you to tighten the CPU completely first, going towards the GPU, then from the outside of the GPU in, diagonally. Probably doesn't make a difference as long as you do it diagnonally, but heck, maybe it helps seat it better. Oh, also - this time around, after motherboard was back in the chassis, I did one last round of tightening on all those heat sink screws, in order; didn't turn, but just made sure I got some good torque on each one, one last time.

2. I didn't bother doing my fan extension mod; the extensions I bought were way too big on the connectors, as I knew they would be, but even the wire was such a thick gauge. Plus I might get the replacement, why waste the effort. (glad I took it all out anyway, see next point below)

3. I noticed again how there is that copper plate inlaid into the steel frame that is under the keyboard. I added some thermal pad last time to make the keyboard steadier; this time, I added another 1.5mm thick piece, right behind the CPU; I figured it might a) help wick heat away from the underside of the CPU, it might be a small amount of the topside, but I figured it might prevent a hotspot there, and b) help keep pressure on the CPU/motherboard against the heat sink - I mean, I doubt the motherboard and CPU is flexing away from the heat sink, and the pad isn't going to prevent that, but can't hurt since it's non-conductive and it's a small piece so airflow shouldn't be blocked too bad.

4. I got to see how my thermal pads worked out - all of them had nice deccent impressions, except for the one that runs between the two chips (the one right in the middle); I had used 1.0 mm for that per iUnlock's guide, this time I added another 0.5 to help it out. I didn't want to go to 2.0 since that might lift the sink away from the cpu/gpu.

5. I did try bending the one long CPU arm just a bit. Not sure how much I bent it, but am pretty sure it changed.

6. I got to see the pattern of the previous paste - still clay like since it was only 5 days old. I definitely used too much. But from what I can tell, too much cannot hurt, it gets squeezed out (online comparisons of different paste patterns and amounts seem to verify this, only using way too little is a difference). Weird thing is, I could see ripples on the CPU! Even though it looks flat when clean and shiny, I wonder if this proves there are depressions in the surface; or, are those ripples just from how the paste gets pulled apart when lifting the heat sink off? It wasn't random though, there must be a reason there was the pattern of ripples. GPU looked more random, for example (like putting tape on wet paint - you expect to see a random uniform effect).

7. Given the above, I decided to forego the graphite pad; I just couldn't see it helping. I was pretty sure the issue was the paste having to fill gaps and losing its effectiveness wherever it had to remain thick to fill the gap. Pads seem to only be good when you have a perfectly flat heat sink and die.

8. Instead, I used the Phobya liquid metal. Put too much on, hard to suck up. Used 20 q-tips to roll a tiny bit off with each end of each q-tip. Seeing as how hard it is to remove (it doesn't wick off at all), I am not as worried now about it moving about. I am still worried I used so much that it got pushed out of the contact areas, but I am not as worried about it rolling about just because the laptop may be vertical at times. I guess I really had to see it in person to get a feel for it. Even a video doesn't give you the 100% feeling for it. But I just felt like it was going to do the best job at filling gaps and remaining highly conductive even if the gap was larger in some areas.

9. Results - I could not believe it; I thought I didn't plug the fans in right, as they didn't even kick in more than minimal amount until the fourth time I ran CB (and even then, nowhere near maximum, maybe half at most the whole time). At my previous maximum setting of 4.2 Ghz, only one core reached 77 degrees! I still can't undervolt more than -125 mV reliably though. But now I can clock up to 4.4 Ghz reliably! 1340 in CB. Beyond that, and I am hitting power limit PL2 (even 4.4 hits it just once sometimes during the benchmark), like you. At 4.4, max CPU is 83 degrees, other cores range from 71 to 79 max. GPU also came down a couple of degrees while running Firestrike, 67 max, was 69-70 maybe 71 previously with my 150/300 core/memory overclock. I feel like I could get 4.7 at least if the power limit didn't kick in. I'm not that mad though, 4.4 AND running fairly cool at 83 max is pretty nice.

10. I know one time I reached 205 or so on single core (even before the last re-paste), now I can't get to 200. How do we set that up to do better? Does it need a completely different ThrottleStop profile to give more voltage for a single core boost? It doesn't make sense to me that my CB multipler from single to multi-core is 7.1x, shouldn't it be more like 5.1x since single core runs faster than 6 cores? Also in ThrottleStop, I see all cores getting activity, and nowhere near the 4.7 I set for one core, more like 4.45 Ghz. Not sure how CB runs one core, maybe it switches cores and that's why it's not getting the higher speed that it should for a one-core load? Any info anyone might have on that would be appreciated.

So at this point, I've done so much on this unit; was hoping for 4.5 and beyond especially when I saw the whole different range of temps the liquid metal brought (I am assuming it was all due to that, not my other tweaks, but I mentioned them anyway). Unless the replacement has a much better CPU, like -50 MORE undervolt capability, I don't think I'll even bother switching. This computer is so quiet now and cool-running, it makes even my last re-paste feel like a couple of years old machine already.

I don't think Dell's cooling setup can be faulted at this point; maybe their mating surface contact from heat sink to CPU isn't perfect, not sure how anyone could perfect that though. We can't expect the manufacturer to sand them down can we? (or can we... lol)

Also, I think what was getting lost to me is, in these desktop tests, I think they're comparing paste vs metal on the heat spreader-to-heatsink contact, NOT the die-to-heat spreader contact. At this point, since we don't have a heat spreader to make a nice even layer for the heat sink to mate with, I almost think liquid metal is a requirement. Maybe I am wrong in visualizing this way, but now that I look at it, I feel like the lid/spreader on a desktop CPU presents a nice flat surface for the heat sink, which is why paste or graphite pad works almost as well as metal; but between the die and the heat spreader, liquid metal is much better. I mean, I don't know how else to explain that I had a 15 degree drop in temperatures at the same clock speed. I certainly was not expecting it to do that!

Edit: worth noting, this is with room temp at 79.5-80.5 degrees F! At least 2 if not 4 degrees WARMER than when I started testing my last re-paste with Gelid (usually room not this warm, just turning heat on and need to adjust it).

 

Many things to read here. However, from a general look I had, it looks like you are limited by CPU UV now because LM temps seem good. Similar to my previous system. By the way, I have tried tightening the screws with different orders but always core 1 was always having spikes (in addition to one more core) and usually getting hotter (about 4 degrees) than the others. This was on all three systems I had. Whats your differential core temps now with the LM? By the way, I think you should reach 4.6 or at least 4.5 now with the LM but the CB score will be limited from current limit throttle (because of max -125mV UV). I can run stable 4.6 with about -195mV but this also depends.

 

Can someone provide some temp differences of the keyboard when you put thermal pads on the back of the motherboard? With a thermal gun pointing at the keyboard. The heat from the motherboard would go straight to the keyboard, there is no other way. I would not trade a warmer keyboard for a tad better temps. My keyboard can hit 44c and I would like it lower if possible.

 

You know, originally, it was core 0 (I assume this is what you mean, the first core) that was the hottest for me; after Gelid re-paste, it was both the second and third cores that were about equally hot, with 2 being in the lead. Now, with metal, it is just the first core that is the max temp again. I would say the temp spread is still about the same though - the lowest is 71, the rest are evenly spread between 73 and 79. So, the spread of the lowest and highest max temps is still 12-13 degrees. I know some people are getting better spreads. I am wondering if one of my cores is just that much more inefficient (and is also the core that triggers the crashing with the greater undervolts).

I do wish I could at least get 4.5, haha... isn't that how it always is. Just one more! I remember the last time I seriously overclocked a CPU was an AMD Phenom 965 BE I think it was - I was so happy when I finally tweaked enough settings to get that one to run at 4.0 Ghz on all cores (very hard to get it stable but I finally had gotten to where it was stable for 48 hours on Prime 95). I can see now why people are upset about the power limit on these machines, with this cooling working so much better, I really feel like 4.7 is doable. (and probably higher if supplemental airflow is added) On the other hand, 110W is a lot! With conversion loss and the power the GPU takes, I don't even know if the 330W (I think that's what it is) power brick would handle any more? Sure wish it would just give another 5 watts though.

 

My IR heat gun is in storage; I do think the keyboard is a bit warmer in the middle, but nowhere near uncomfortable. You say you get 44 C? 111 F... mine is definitely not that hot, I don't believe. Can't be sure though. If I get the chance to get the tools out, I will measure it sometime.

 

@Aristotelhs2060 I found this post, didn't realize there was a thread just for the graphite pads - he seems to find that the pads are not any better than basic paste/grease. There also seemed to be discussion in the posts before that about poor heat sink contact, whether paste plus the pad would help, yada yada. I think I will return my unopened graphite pad, no use for it now. Great for a desktop workbench though if someone is testing many CPUs/coolers and needs to keep changing parts out.

http://forum.notebookreview.com/thr...-test-and-review.815439/page-45#post-10782819

 

I have a 13 r3 so I wouldn't be surprised if the high end r5's hit 50c.

 

@propeldragon Did you mean with CPU and GPU both under max load, or just CPU? I still have to say CPU-only is nowhere near uncomfortably warm; with GPU on after playing for several hours, maxed out, yes, keyboard is very warm and toasty - perhaps even uncomfortable if you held the back side of your hand where it's more sensitive to it for a long time; but still perfectly usable for typing (but I agree, probably well above 44, perhaps even closer to 50 as you say). I am curious how much my backplane heat pad is helping though. Seems like some, for sure- and CPU is basically upside down, so I figure some heat is going to rise into the motherboard (and now, via heat pad into keyboard backplate), no matter how effectively the liquid metal and heat sink are working. Hard to tell how much airflow there actually is underneath motherboard when mounted, with that frame it's on - doesn't seem like there would be too much.

 

With the cpu and gpu in use, stress test or gaming. Hard to say how much airflow there is inbetween the keyboard and motherboard. On the other side however (bottom base and motherboard) there is some airflow. I have the 13 and my thermal pads on my ssd's was getting a little dusty (didn't have 100% contact throughout the 2 ssd's). So air does move around the ssd. With you r5 owners you have even more of a gap so you guys get even more airflow.

 

Are you sure these are the Panasonic *compressible* ones? Because the do make normal pads which are not and you definitely not want these.

 

The compressible pads are not forever "spongy" and they do deteriorate after a few fittings, and it is best to use them only once. For one, the CPU and GPU die make an "impression" which you then have to get perfect again if you remove the heatsink. As these pads are so light they move even if you breathe the wrong way, therefore I used a tiny bit of thermal paste (outside of the die of course) to hold them still while fitting the heatsink. That way the pad stays "glued" onto the CPU while removing / fitting the heatsink.

If you removed the multiple thermal pads around the CPU/GPU then the heatsink would have a perfect fit to accommodate event he very thin graphite pads. Otherwise you got to get the thermal pads perfectly right, and it is almost impossible, and every new open/close attempt brings different results. If you get it right once you do not dare open it again because it is unlikely your success will be repeated.

Here is what happens when you get the graphite pads right - very good score and in a hot room,
room temp: 27C 140mV undervolt, P95 10 threads, locked multipliers

3.5GHz P95 10t, 66-68W, 79C-81C
3.6GHz P95 10t, 72W-74W 84C-86C
3.7GHz P95 10t, 78W-81W 90C-94C
3.8GHz P95 10t, 85W-88W 92C-97C eventually power limit throttle

 

No, I was unable to find them for sale anywhere. I decided to do liquid metal anyway on the CPU/GPU, did not try the IC Graphite pad. You mentioned the Panasonic one for the CPU/GPU, or at least that's what I thought you'd meant it for. The pads I referred to were replacing the factory pads, and adding some behind the motherboard, all of which I used Thermal Grizzly Minus 8 pads for (which seem to allow a nice imprint for those areas, even if they don't "compress").

 

This is the "flexible" keyboard - it is flexing by ITS OWN weight !

In the background the metal plate sitting over the keyboard - it is not much sturdier .

 

What happens when the VRMs get hot. Observe that the CPU and GPU are fine, absolutely fine, yet the CPU has gone into a 800MHz throttle loop !!!

Edit: you will also see the PCH very hot, well it is a bogus reading, I consistently (many days, many times) measured the PCH to be 25C BELOW what HWinfo was reporting.

 

Farnell, Mouser, Digikey, and you do not pay super-inflated prices either. I have also put a link directly to Farnell in a few of my posts here, if you do a search.

 

Do you have a copper piece with plastic covering it that fits in the depression in that metal plate in the background? I can't seem to spot it in your picture anywhere... wondering if mine is different inside. It would be right underneath where the CPU and GPU backsides are located.

 

Yes that metal plate has an incorporated copper-looking piece which I believe is a heatshield. It is very thin, and does not provide extra rigidity, it's there to withstand the heat I presume. You know like you look inside a car's engine compartment and you see the heat shields surrounding the exhaust manifolds or the turbo pumps.

 

I don't understand what the issue is? All the notebooks keyboards are going to flex when it's apart like that... this keyboard is rigid as hell when assembled.

 

Its rigid as long if it is fully assembled. Thats the main reason why the keyboard decks of alienwares are have virtually no flex.

 

What paste do you use atm? Month after my first repaste the temps are still the same. Never had any 800mhz throttle...

 

Ermmmm, no.

Again, no.

I have posted photos and have mentioned it in the past, the keyboard flexes at the top rows and in specific locations because of the way it is put together. Typically the following keys flex downwards as you press on them or on their neighbour keys: ESC, F1, F2, 1, 2 and, maybe also F6, F7 area, and then again towards the NUMLOCK.

You can make it better/perfect by inserting gap fillers in between keyboard and metal plate, however this requires a complete disassemble of the heatsink + motherboard and battery, sub woofer and so on.

It is the VRMs (below the GPU) causing the throttle, while the CPU/GPU are < 80C.

 

I’m getting a replacement unit, and after going through customer service hell they finally agreed to upgrade my build to include Tobii Eye Tracking (this is my third replacement).
One thing I’m a little conflicted on is that that obviously means they’re changing the panel to a 120hz TN. I’d normally be really happy about this but colour reproduction is pretty important to me as I do some graphic design as well - and I also just really love pretty colours in games. I’m not sure if it’s even possible for me to ask if they could stick with the 1080p IPS and keep Tobii. Or is there a better panel I could buy and swap out myself?

H e l p .

 

You can also get the UHD screen, that is available with Tobii. I would demand they give you that. I have been through hell myself because I ordered with QHD and thought from these forums it was easy to switch to UHD later on; yeah it's easy to do, but it's damn hard to find a reliable source online. First one I got was Innolux, looked beautiful, but a few dead pixels and no G-Sync; just today I got a supposed AUO panel from China, but it smells weird, and the brightness maxes out at 170 nits. (no dead pixels though and G-Sync works, lol - but it has the "No Warranty" in place of "Made by AU Optronics", which according to a YouTube video I just saw, means it's a "fake" - I think it might mean it's a reject/"refurb" though) I'd keep this one but 170 nits is unusable even in a very dark room.

Now I just put my QHD back in in after two weeks of using the Innolux; the lack of color gamut in comparison is immediately noticeable. However, so is the bringing back of the 120 Hz (i.e., its best part). Oh how I missed that high refresh rate... mouse moves like it's on glass. And zero backlight bleed (although only appreciated during startup when it's all black).

There's lots of threads about the QHD TN 120Hz vs the UHD 60Hz. The UHD is spectacular, of course. No doubting that. But the QHD is a very solid screen. While I notice the color repro is less, it's not disgustingly so. More like "very nice" vs the UHD being "excellent". The reason I had to go for the UHD was I got sick of the color shift - it's not bad just off center, but moving in and out form display, I notice it. To combat it, I tilt the screen top towards me by 5 degrees, so that when I sit back a few inches, I don't notice it. Also, the blacks in-game were more like a dark green, which was annoying. (I have since bought a Spyder to see if I can correct it more, but I doubt it, black levels are inherent to the monitor.)

How much graphics design are you really going to do on it? Might you get another monitor for that? Keep in mind, color accuracy is usually for printing; I honestly think the QHD panel is sufficient for web/digital imaging. As long as you have references, I don't see why it would be completely unusable. There are some colors I noticed on the UHD that were more vibrant; not sure if that makes the QHD unusable though.

Also, are you gaming with the machine at all? If so, FPS or more RPG or adventure, etc.? 4K is great for more adventure style gaming or strategy, etc. After my last re-paste, I couldn't believe it - even after my first re-paste, I couldn't, but now I can play Assassins Creed Origins on all Ultra settings (except medium anti-aliasing) on 4K, and the lowest my FPS ever dipped was 38, usually 44-50 FPS. Before this last re-paste, it would only get 3-10 FPS. The GPU wasn't overheating before though, so I wonder if the CPU throttled it somehow (even though it ran fine at 85+ FPS in 1080p).

I still think the best option is to get UHD in the machine, and get a separate monitor for gaming. Cheaper and easier to do. Or even get the QHD panel later, it seems a bit easier to find online than the UHD. If you're FPS gaming on the laptop mostly, and maybe some design work, then perhaps QHD built-in and external high color gamut display is better.

I would say the QHD TN panel is as good as a low end IPS, it is really top notch for TN. I have a cheap TN panel in the house and now THAT one makes me gag. The viewing angles are pretty good, especially left to right. The up angle is good too (top shell pointed towards you), which is why I keep it angled towards me.

Also, no, you cannot get 1080p screen with Tobii. I highly recommend getting QHD or UHD because then you get the wider cable; if you get 1080, and want to upgrade later, you have to buy the proper cable, and then take the WHOLE machine apart (all of it) to get that cable threaded through to the top shell. Much easier to switch between QHD and UHD panels. I can change mine out now in less than 3 minutes, probably 2 if I really wanted to.

 

Fun fact I just discovered - Windows Hello still works even when Tobii eye tracker is disabled! I thought for sure that would kill disable it, but I guess it has two levels of operation (basic IR for Hello, and then the lights on for the eye tracking). I disabled it so it wouldn't auto dim while I run colorimeter calibrations, never would have guessed otherwise.

 

You change the screen panel without taking the machine apart? I have seen a youtube video doing this but there is a lot of stretch and very awkward removal of the panel connector whiuch is hard to remove even with the machine completely opened up ? Is this what you are doing too?

 

Of course. Takes about 5 minutes tops.

All repairs / replacements are documented in the service guide available on the dell support site.


Sent from my iPhone using Tapatalk

 

Did you look at http://www.panelook.com/modeldetail.php?id=25358

 

As long as you have the correct cable for the QHD or UHD, there is no need to take the motherboard out. The people doing that are the ones switching from the FHD to either QHD or UHD, because they need to switch the cable out from (hope I remember this correctly) 30-pin to 40-pin. Once you have the wider cable, you just take out the bezel, undo six screws, pull panel out, unclip that end of the cable, and reverse to put the new panel in. Literally 3 minutes or less once you're comfortable with how much pressure and where the retention points are on the bezel.

 

The question is at which CPU frequencies are you getting those drops to 800MHz on? The reason I am asking is because I see them on 4.6 Ghz and over with OC enabled so I am not sure if this is due to VRMs or just power throttle. I see on your screenshot you have those frequency drops at 3.5Ghz? I can test at those frequencies to see if I am getting the drops.

In addition, the drops can also be because of latest bios versions.

Try use TS and use lowest of 31 or higher and check if you still get those drops. According to my tests, this looks to bypass bios crappy settings

 

Run 4.7ghz in CPU demanding games-> no drops... Maybe the whole System is just to hot... maybe 95 degree cpu and 80 gpu heats up the rest of the mobo...

 

Correct but you are also using different bios version.

 

Maybe, dont know. Never touched my bios again

 

Thank you for the reply! I would LOVE to get the UHD screen but sadly I’ve only got a 1070 and seeing as how even the 1080 struggles with 4K gaming, I don’t think I’m prepared to sacrifice my frames like that. :’) I don’t play a lot of FPS games, mostly RPG and simulation, any game that really favours atmosphere. I was hoping there was some third party panel I could buy, like a 120hz IPS, where it would just be an easy swap out. Something like this panel?
https://www.ebay.ca/itm/120HZ-FHD-I...319873&hash=item4b1ef3eea0:g:ezgAAOSwJLJZids-
Not sure if that’s even a possibility though. Does FHD just not work with Tobii?
I’m going into animation as well as graphic design and video on the side and I just don’t know if I’ll be able to stand those washed out colours.. I guess I’ll see when it gets here!

 

Would you say that those retaining clips can only be pulled out a very few times before they lose their shape/retention capabilities?

At that time, or during similar times, the laptop was opened up, resting on its side, running most likely Subnautica, and I was using a probe digital thermometer to measure temperatures, eventually identifying the culprit as the VRMs positioned below the GPU. They reached 104C, the CPU would clock down to 800MHz, a few seconds, temps dropped to around 94C on those VRMs, then the CPU would go back to whatever it was set to, and the cycle would repeat. Obviously at the 800MHz periods you'd notice sever frame rate drops.

 

Seriously, the AUO UHD has to be seen to be believed. There are colours that I did not know existed. And it is 4K as well.

 

I really hope you have fully closed the door for Win updates

 

Ahh ok, 1070 - yeah you'd have to play with lower settings to get 4k playing acceptably. The 1080 does well with Assassins Creed Origins like I mentioned. Very rarely hits 38, usually 45+ FPS, which is suitable for that type of game. (the G-Sync panel was much better I noticed, even though too dim to keep - when the V-sync slows it down to 30 FPS, assuming it's not doing triple buffering, that's noticeable)

Don't forget you can play at 1080 or QHD resolution too! I read up a LOT on that. The trick is, you have to be sure to set the Nvidia Control Panel to scale to screen size not aspect ratio (hmm or was it vice versa). They should be equivalent, and maybe I was imagining it, but I felt sure that one of those selections made the game seem badly scaled, whereas the other was more of the 4:1 pixel ratio I was expecting. I think people think it looks worse than true 1080p pixels because the quadrupling of pixels sort of smooths it out. So it doesn't look as "sharp", but in a way, looks like a boosted/smoothed 1080p. QHD is also surprisingly good looking on a 4K screen, I would never have believed it. Finally, some games, like Origins, let you over or under-render the resolution - so you can add more detail to a 1080p output, although not sure that helps much, or more usefully, you can play at 4k with ultra settings, but set the engine to only process it at 70% of the resolution (50% would be 1080p, I believe, if it's going by vertical scan lines, not surface area, which I am pretty sure it is). Even before my last re-paste with liquid metal, when the machine would stutter playing 4k natively, setting it to 70% made it playable.

I keep hoping someone will make a high refresh AND IPS panel soon. Seems like 17.3" sizes just don't get much love being not as popular, I imagine. There are 120Hz FHD/1080p panels though like you said, I am not sure if they will fit though or you might need to modify the mounting tabs. But from everything I've read, FHD just can't be had with Tobii. No idea why, they don't share a cable or anything. Maybe the FHD top clamshell from factory comes without the space to mount the Tobii unit or something? And/or it's just a wiring thing? Which I think could be added later. The panel you linked to is 40 pins though so it seems like it would work with the way the QHD is setup as that uses 40-pin also.

I think they get worn a bit but I've done mine now 8-10 times in total and it's still secure when I put it back. If anything, after the first time, it's still highly retentive but much easier to get out. I don't feel like they're going to give out completely anytime soon.

It sounds like you have it pinpointed to the VRMs. You disabled the BDPROCHOT option in Throttlestop that links the GPU to be able to throttle the CPU, right? I didn't check the monitoring apps but it sounds like this was what happened before my last re-paste, stuttering when in 4K but not lower resolutions. I didn't adjust the VRM pads though, and they had good contact from my first re-paste, so I don't think that was the issue in my case.

Lol I said the exact same thing in another thread! It really puts even my MB Pro to shame in both sharpness/resolution and color. The Innolux Chi Mei is also the same, almost seemed like the colors popped even a bit more on that one, but I think it's because my AUO that I got is limited to the 179 nits of brightness; obviously, being so dim, it's at a disadvantage - Spyder5Pro did give it a 98% for NTSC vs 96% for the Innolux, Adobe and sRGB were the same coverage.

I literally didn't know what I was missing. That 100% vs 75% Adobe RGB coverage really covers some shades you don't see often.

 

Now that I've seen this, I'm really considering selling my AW... just a shame azor took the BGA road while Acer a second class brand can even source mobo's with socketed cpu's...

 

@Aristotelhs2060 I was wondering, when you said you can undervolt -200 mV - is that rock steady stable for you, like you can play whatever, run Prime 95, etc.? Or do you have to lower it to use the computer reliably?

I ask because I got my replacement. Was so happy with where my last one got, didn't really even want to do anything. But even with the factory setup, I could tell it had some promise - it was running 4.3Ghz with highs of 88 from the factory, and for some weird reason, even with only a -125mV undervolt, wasn't going past 85 watts (yet on my previous one, 4.3-4.4 is its max because that is pushing right about 108.x watts). So I guess better binned CPUs are not only capable of more undervolt, but also run at lower voltage automatically on their adaptive voltage setting? I actually asked this in the TS thread before I bothered re-pasting, but no one answered.

Anyway, I ended up doing a whole re-paste with liquid metal, mostly re-padded (left the ones around the CPU because they seemed to be making great contact and are a denser material than the Arctic-looking pads around the GPU) Since I hadn't run the computer much at all, the Arctic pads were also still in great shape. I think they don't hold up well though because in my last one they'd become very gummy and were barely holding together by the backing layer. I did my pad behind the CPU again.

So it turns out, now I too can undervolt to -200! Crashes right away though at -205. Weird thing is after crashing, I back it down to even -170, and it still crashes when opening TS upon reboot, but then after a few reboots, and manually turning off the custom voltage in the .ini file so I can get in, I can then start at -160ish and work all the way up to -200. That doesn't really make sense though, the CPU has no memory after I shut it down and restart, does it? So I am figuring anything above -160 isn't stable - although I just ran Prime95 for 10 minutes at -179 with no issues.

Anyway, thus my initial question to you about whether you run stable at -200 mV.

Highest stable clock I got was 4.7 with a CB score of 1470ish. Set to 4.8, clock rates fluctuate 47-48 but won't hold steady at 48, max CB score was 1498 after several attempts. (doh... just couldn't get to see 1500!) If I can back it off and know it will be reliable at 4.6 Ghz, which will run max temp of 82 from what I've seen, that will still be a CB score of 1425+. (and I don't optimize for CB at all, no task prioritization) I just hope it doesn't end up that -125 is my max stable undervolt and that the higher ones are just for fun.

@cruisin5268d - I have an update for you also sir! This replacement came with an SK Hynix, not a Samsung. I did test it once, scores were similar - a good bit lower for some but higher for others. Overall though I don't think it's a worse drive. Seems like Windows starts way faster, but that's probably because I am just using a local account for testing for now, not Microsoft login with all the other stuff that starts up. (but seriously, one time on boot I really think I saw the welcome screen in less than 4 seconds)

@doofus99 - and an observation for you also, sir! At first, I felt like the overall keyboard on this replacement was a bit looser. However, I noticed that none of the loose keys you or I have mentioned seem extra loose on this one - Esc, F5-F9, numbers 6-9, etc. Everything seems even. Esc key might sound a bit hollower, but does not give way. If I rest my other finger on the F1 key and the chassis above that, it feels like the F1 moves ever so slightly when I ding the Esc key; but nothing like my F7-F10 keys did on the other one (even after I propped them up). I doubt they changed anything, just luck of the draw... I wonder what part of it changes though to make it less or more supported? Especially since the top row was soooo loose on the previous one.

Ha, and now it just crashed doing nothing as I finish the post up on my other computer. But that was at -189mV. Weird how it seems to have a wide range where it doesn't crash right away, but does seem to crash eventually.

Edit: one other random observation - I have yet to run my colorimeter, but I swear the QHD panel on this replacement is just so slightly nicer than the QHD on the previous one (which I put back in after two UHD panels both had issues, waiting on a replacement for one of them). Seems like it has a bit richer color and a bit more contrast. I even swapped computers left and right, with same image, my eyes always picked the new one as looking nicer. I don't think that is even possible though for two of the same panel to look much different, but maybe it is. The newer one really looks like IPS to me (next to my IPS Lenovo, I actually prefer the QHD on the new one), whereas the QHD on the old one looked a bit bland next to my Lenovo (and is what irritated me enough to upgrade to UHD). I gotta be imagining this though. Will post if the colorimeter tests show anything conclusive.

 

I am confused, how many laptops have you got from Dell?

 

I have gotten one trillion, 293 million, 557 laptops. No, this is just the second one. Evaluating to see if it's worth switching over from the first. (and honestly, if the first one had performed as well as this one with factory paste, I might not even have re-pasted - the factory paste even looked good when I re-did it) I think the confusing part is that I did TWO re-pastes on the first machine, and now just one re-paste on this one.

 

The possible UV on my system depends on the frequency of course. At 4.3GHz, it allows me an UV of stable -211mV. AT 4.6GHz max possible UV is at about (-185mV to -175mV). On my first system, I could UV up to -150 at 4.3Ghz with LM pasting, gelid thermal pads and K5pro for CPU thermal pads. This is a huge difference. So, it looks like max stable UV depends on the CPU itself rather than the effectiveness of re-pasting etc. The temps depend on the re-paste though which has an effect on the CB score due to thermal throttle. But maximum UV has the most effect on the temps anyway rather than repasting. As Papusan said a -100mV UV causes about 10 degrees drop in temps by its own for a given thermal paste or LM. All frequency/UV combinations I talked about above were tested with CB so yes they are pretty much stable. And my max CB score (1519) was with a gelid thermal paste and default thermal pads.

So, finally, the max stable UV has the most effect on both temps and CB score (even though I doubt I would reach that score with default thermal paste-again thermal throttle could show up much easier at lower power consumptions). LM allows to reach higher power consumption (on my previous system, I could easily reach power limit unlike any thermal paste but probably due to much lower stable UV) because it reduces temps for a given UV.

Just to summarise, it is a combination of things. Generally, an LM (and possibly a good contact thermal pad) with good stable UV could give max CB scores. I still prefer to keep the thermal paste because my UV balances the loss in temps from not using LM and because thermal paste does not cause discoloration and cannot cause any unpredictable damage on a moving system (laptop).

The UV capabilities on my current system is the main reason I am happy with my current R5. As it the third R5, I am not usre if they have done any changes to latest machines or using better parts to allow higher UV. This looks like a lottery.

 

Each processor is programmed with a maximum valid voltage identification value (VID) that is set at manufacturing and
cannot be altered. Individual maximum VID values are calibrated during manufacturing such that two processors at the same
frequency may have different settings within the VID range. And the hotter the chips run (higher overclock) mean you have to put less undervolt vs. if the chips could run at much lower temp. Higher temp due higher clocks means you have to use higher voltage. The hotter it will run the more voltage has to be added.