*OFFICIAL* Alienware 17 R5 Owner's Lounge

You could try a Chimei N173HHE-G32. I put one into my wife's 17R4 and it works great. The only real hard part about the swap was I had to install the 40-pin cable meant for QHD / UHD panels from Alienware. The Chimei is a FHD 120hz TN, with wide gamut (similar to the UHD panel) and good response time (5ms). The Antiglare coating is also less aggressive than what AUO uses so clarity is better especially on solid colors.

 

In a few words, not all CPUs are the same and the maximum stable UV is a lottery

 

As I said... worse it will be with worse cooling. Dell should take account on binning quality - variable (the chips is soldered and can't be replaced) and their stated OC profiles when they designed the cooling. As it is now... It's a lottery if you is able to get barely more than stock clocks on all cores. And the oc profiles (with too low power limits) is as always Dell's sales gimmick. Worth nothing.

 

As Papusan said, each chip has a pre-programmed default VID, starting from 800 mhz and going up to the highest OFFICIAL maximum one core multiplier (not overclocking bins, but stock speeds). The default VID stops scaling here, at least on K and HQ processors and 7850HK (stops scaling at x39), but I don't know where it stops scaling at on 8950HK. And without access to the internal IA AC DC loadline settings to set this to "1", its impossible to see the default VID as there is going to be VID boosting going on (1.80 mOhms). CPU's with a lower default VID will undervolt more than a higher default VID.

 

My undervolts have always found there instability when doing light casual use. Right when it hits that super low voltage, blue screen. I use my laptop for everything. Sometimes you need to switch power plans and think about battery life. I bet a lot of people only test with benching!

 

Thank you both for that info! Exactly what I was trying to ask in the TS thread before I bothered with the re-paste. So it all makes sense now. Even at the same undervolt, two CPUs might still operate at a different voltage. That was the missing piece to my understanding.

So I know I've seen this discussed before - the 8950HK is supposed to allow for setting the power limit, right? But the issue is Dell's BIOS has never supported that option? Or somehow they set the CPU to block it? (and it cannot be set from software tuning - or is Dell blocking that somehow also?)

I'm having a mixed experience on this new one - I ran 14 hours' test of Prime 95 small FFT with no errors or crashes in a warm room, 84 deg F, -169mV on cache and -179mV on core, at 4.7 Ghz. While it did get much warmer at times, at least it didn't crash. Then it seemed stable while I did some stuff. Now it crashes as soon as I open TS upon reboot over and over. ( @Papusan this is why I was wondering for more info in the TS thread - because it seems like the overclocking stops when I exit, but is there some setting carried over that keeps crashing until I do many clean reboots? or just coincidence...) Prime95 used to seem like proof positive that something was crash-proof, but like others say, it seems other stuff can still crash when Prime95 doesn't, and vice versa. I guess I will just drop the UV a good bit and just do general usage/benchmarking instead to test for stability.

 

You need to remove AW OC tools and Intel XTU which are hidden and marked as System component that will conflict with TS. You can use revo or BCUninstaller with system component checked off to see them and uninstall those SW tools and disable WU using wub so they don't get installed in background.

 

AW OCC does not conflict with TS. That is if you allow ACC to load and run TS manually( so not autostart). In fact, I noticed that AWCC has to start before you can actually set max frequency in TS. I have AWCC set to 4.8 and -100mV and then play with TS (max frequencies of 4.8).

Sent from my SM-G955F using Tapatalk

 

Try disabling the driver service for XTU xtuacpi or similar.

 

There are 'limited' overrides, but they are still locked down fully by the Bios. Even XTU can't override the maximum Bios limits. These limits could also potentially be enforced by the Embedded Controller as well. If there is a maximum 110W power limit and setting this to 150W in Intel XTU, and setting ICCMAX to 200 amps won't bypass this, then there is absolutely nothing you can do.

One user was able to find a way to tweak "IMON SLOPE" manually to "50" (50%) via booting into RU, but he was able to find the register which controls this, by dumping his Bios. This makes the CPU report it's using 50% less power than it really is, which spoofs the power limit. But this wasn't done on an Alienware. I don't know anyone who has been able to dump the Bios strings to find which hexadecimal offset controls (Core I/A-> IMON SLOPE). And without access to that, you have no way to spoof the power limit TDP reporting.

 

Actually, no. But since I measured those two VRMs at 104C, I would not be happy with such a system even if it did not throttle.

Hmm, max stable undervolt depends on frequency and temperature and actual CPU as aristotelhs2060 said. You should be very happy with -160 mV. My 3rd AW 17 R5 can only do -135 mV, and I may still discover a combination of parameters that prove even the -135 mV to be unstable. The previous one could do about - 155 mV. However, the number is not that important because the actual wattage (aka heat ) generated is also dependent on the programming of the CPU at the factory.

My 2nd R5 was by far the best at -155 mV but in terms of CB scores I am not sure. Because when you run CB you may get throttles and then it all depends on how many throttles you got while CB was running and how severe they were. I can say that my 3rd AW R5 gets 1275 cb at 23C with -130 mV (max). The previous one could do -170mV 1340cb at 28C-30c room temperature, but with a (badly fitted) graphite pad. Maybe if this one could also do -170 mV and had graphite pads then who knows?

http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-233#post-10803246

 

Vrms tend to get really hot. Also on well cooled desktop gpu boards. Lower is always better but not out the ordinary.

 

I forgot you tried graphite pad. How were the temps compared to a good thermal paste? Which thermal paste did you try? I have an ICC one now but have not tried yet so I need to have an idea how it performs on R5 so I know if it worths the time or not.

 

Oooh, hadn't thought of that. Intriguing! Overall, now that I've experienced two systems, I can see why reviews/reports are all over the place though. This new one had such a better CPU and an actually good paste job, that if it had been my first unit, I probably would never have opened it up, undervolting would have been enough to satisfy me. Plus I can see whenever I dig even further back in this forum how the different BIOS versions which did at one point allow some manipulation of power limit confuse the issue (I think at one point setting fan profile in BIOS changed max power limit also?). On the other hand, if Dell would just allow the higher limit, then even my first unit would get a nice overclock - it can't UV more than -125mV, but with max temp at 82ish, I had some thermal headroom. Like, if they would unlock it, less people would complain because even the lesser CPUs could exceed a good bit above the 4.3 default turbo.

Yeah, I finally see this thanks to Papusan and others' explanation, and my own experience. I wasn't understanding that two CPUs could differ so much on the voltage they're set/programmed to use until I experienced it for myself. Even at -155 I can get 4.6 with some headroom on the power. Just hoping 4.7 remains stable at -175ish, that would be really nice.

I know you directed this to doofus99, but I forgot to mention - I saw one post where the person actually used a sandwich of two IC Graphite pads and a copper shim in the middle on their laptop. I don't think it was an AW, but similar concerns, wanted to fill gap, etc. They actually had good results. I think it might have been a four-core so the temps weren't as immediately comparable to me, but I think the gist was that he did get the best that any thermal paste could have gotten, even with the copper shim. I probably meant to share the link here but I forgot to. It was another forum I believe, not this one.

But for me, after seeing liquid metal twice, I am in love with it. I mean, power limit is the ONLY thing on my mind now when I overclock. The only way my hottest core goes over 82 is if I either run Prime 95 on very small FFT sizes only, run it in a hot room (84+ F), and/or run the GPU hard at the same time (at which point yes I think it bounces into the 90s - will have to do more testing then, I might do a less extreme overclock for simultaneous GPU usage just to be sure thermal throttling never kicks in - but from gameplay experience I really don't think it has). It's so hard to believe that it can make such a difference, being that the two surfaces do already touch, but man, the metal really make the die and the sink act as one.

 

Could you + @Aristotelhs2060 run and post results equal @captn.ko? But with Cinebench R15 scores in the background. I mean he used -0.150mv undervolt http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-123#post-10757884

 

@Papusan You mean just show the total HWiNFO output, after running CB a few times to get it warmed up, and share the UV/clock used and the CB score? Sure I can. I am just working on migrating my drives over though to the new one. Thinking about doing a complete reinstall and using local account this time, taking lots of system restore points along the way - but maybe not and I'll just use my install from my old one.

Another interesting thing is, no WiFi problems yet on the new one. So something about the OEM image or one of the first updates it applied somehow caused that lack of ports error that was cutting out the Internet for some of us after ~24 hours!

 

@Papusan Here are two runs. The CB scores and stats are in the filenames, but laid out again:

Four runs at 4.7 Ghz for six core setting, -189.4 uv for cache/core, scores were:
1374 1443 1447 1459 (yes lowest was first and went up from there)

Five runs (but one was low and similar to the lowest one, so I ran one more time and got another high one - was entering scores in Cortana bar to try and copy paste them all and was afraid search activity might have affected that one) at 4.6 Ghz for six core setting, -149.6 uv for cache/core, scores were:
1424 1390 1423 1430 (with another low one after the 1390, I think it was 1374)

Now I know the results seem odd, max clock is at 4.7 even on the 4.6 test. I guess this is from when it shifts to 5-core or less mode before/after tests? Temps seem odd too, similar or even lower for the higher clock/uv. Thing is, watching TS, max temp is no more than 83 on hottest core (for both the 4.6Ghz and 4.7Ghz tests), and others are maximum 80, then throughout the 70s. So I think HWInfo is inaccurate on those. (this was on all four/five tests on each setting)

Room temp was 75 deg F.

Note: while I was testing before that, I forgot I had Edge open, and it was bringing score down by 50-70 points. I know I was getting higher scores on minimal Windows setup on this second machine, but now I have SSD from original machine in it, so there's more background tasks running. Otherwise, the 4.6 is so close to the 4.7, it almost doesn't even matter. Or if I run 4.6 at the -189 UV (what I used for 4.7), then temps get much cooler, max 79 if I remember correctly, and max of ~94 watts power.

Other weird thing - computer instantly crashes earlier when I tried -184. Yet at -189, ran the tests, did other stuff, Aida64 torture test (without GPU) for 5 minutes, was sure that would crash it, nope. One thing I did though after the crash was set minimum speedstep and cache clocks to 30 each. I am thinking maybe the dips to low clocks and (I assume?) lower voltage, is what's causing the crash, since that's when crashes occur - only once did it crash during the actual CB test and that was -200 or so.

 

I assume that running at only -155mV will definitely cause thermal throttle with my gelid thermal paste instead of LM. But I can try later.

 

Isn't it possible lock the clock speed you want in this notebook model? The clocks is everywhere on the table. And the scores as well. All too low for the clocks you try to put. Use fixed -140/150mv undervolt and try different clock speed from default and up to where it is able to run stable without thermal throttling/power limits.

Default i9-8950Hk clocks...
4800 MHz (1 core) with Thermal Veloicity boost
4,600 MHz (2 cores)
4,500 MHz (4 cores)
4,300 MHz (6 cores)
Just increase the 6 cores turbo boost for the testing (6 core active in ThrottleStop). No need to fiddle with the other clocks. Let default 48x for single core or whatever other clock bin be as it is. You can decrease the polling period in Hwinfo sensor settings for faster sensor tracking.

Depends on how bad the heatsink fits is. Try clock speeds who is able to run below the thermal throttling temp.

 

I've had a similar experience : I ordered an Alienware 17 R5 on August 20th, one day before the per-key RGB lighting was made available as a purchase option. Unfortunately for me, I only found out about this after my 30 day return window had closed :/ .

With respect to the part number the gave you, I'm not sure if it's correct as I have heard that the MPN was 44rc9, but, I've been in contact with Dell, and they have told me that things have been a little chaotic with the new Alienware laptop's and the per-key RGB keyboards and that they couldn't find the correct part number. I was supposed to be contacted by Dell the Friday before last, but they never contacted me, so, I guess I'll call them again Monday.

I'm also hesitant to order the part, as I've received conflicting information from Dell : One person told me that the new per-key RGB keyboard would need its ribbon cables replaced with those of the 4-zone lighting keyboard, but other people have told me that the per-key RGB keyboard's ribbon cables *are* compatible with my mainboard and that the macro keys are the only problem.

At this point, I could call up Dell tomorrow and go ahead and order the per-key RGB keyboard, but there's no guarantee that I'd get a compatible keyboard as there might be multiple internal parts with each of the different per-key RGB keyboards having different ribbon cables. Also, a rep told me that even if I could get the keyboard connected that there's no guarantee that it would work on the software side.

Hopefully I'll get some more information soon...

 

The clocks are locked, especially at 4.6 - they don't move off of it at all in HWiNFO while watching nor in TS. As I said, I believe the reason the clocks go higher in the "max" HWiNFO column is because in between runs, the system is moving into the state that allows 5-core clock, maybe sometimes 4-core clock, etc. I should have set all settings to 4.6 no matter how many cores, I guess. (or just start AND stop HWiNFO during a single run of CB.

The scores seemed reasonable to me - I don't recall exactly but aren't you running about 4.8Ghz and getting 1500?

Also I think some of the background processes may cause some of the fluctuations. One time, like I said, I was trying to copy and paste the last three CB scores while it ran; when I didn't touch it, it ran high again the next test.

I did 5 hours of non-stop Origins gaming last night at 4.6 so it at least that one seems stable.

Also, I calibrated the QHD screen on this replacement - it really makes it shine! I almost feel like I am beginning to see more colors on it than the old one. I suppose different screens can look better or worse in the same model. Anyone who had my old one would be saying "meh, it's ok, but it's washed out", whereas anyone with the one I have now would be saying "QHD TN is really not bad, if you see it you will be very surprised". Like, I can move my head back and forth several inches, and the shift is barely noticeable, more of a brightness shift and not a color shift. Left to right, it behaves just like an IPS, very small difference in brightness. Manufacture dates as reported in the EDID are similar though, early 2016 for both.

So unfortunately it seems a lot of the user experience is luck of the draw - CPU, SSD, screen, and general fit (keyboard support, etc.).

 

Oops, forgot to quote this one in my last response. I am pretty sure it will work, but it does look like it plugs in differently. From looking at iUnlock's guide on the older R4 and also the connectors on my MB - it appears the per-key just has one cable for trackpad and keyboard and macro. But I can see the unconnected ports that say KB" on one, which I think is where iUnlock's guide shows two cables going to, separate for trackpad and keyboard. I doubt they're shipping a different MB for per-key? It's just a USB device anyway to the motherboard, right, and the guts are in the keyboard?

I don't know about the software angle. One would think the single Control Center would do it all. I guess at least you can always return the part if it really doesn't work out, right? (little solace I know after the work to try it - although can't you just take the back panel off, plug it in, and test it on the bench before installing it completely?)

 

If you lock the clocks to a specifick clock speed (all 6 cores) they should stay there and not drop. And you should see higher clocks for some cores in light load if you don't tax all cores (if you have put higher clock speed for some cores). Background processes doesn't matter. 4.8GHz is more like +1610cb.

44x

 

Well I tried at 4.5GHz and gave me crazy spikes for certain cores. I need to use my lowest stable UV to avoid thermal throttle. Again the high UV of this machine ia the reason I do not use LM. Thermal paste does not allow temps of LM at certain UV. What do you need to check?

 

My first AW 17 had very, very bad fitted heatsink because temps would reach 100C with almost no provocation - I returned it within the 14-day window for a full refund (£3,400).

The second AW 17 had a much better heatsink and it was on offer at £2,900, much better than the first one. I opened it and messed with it. I discovered that the thermal "stamps" used as default by Dell, were quite dry, and I took photos to show the crumbly thermal grease (ie rubbish) on both CPU and GPU. I repasted, re-padded, used graphite pads, and at one stage I had it running really well, allowing the CPU to spend over 90W and got about 1380 cbs with 30C in the room. Unfortunately even though the GPU and CPU were happy, with the GPU at full whack (running Subnautica in the background), the VRMs below the GPU had their own mind and sabotaged the deal by reaching 104C and throttled the CPU down to 800 MHz, even though the GPU/CPU were around 80C.

One phone call to Dell, and *one* single screenshot later, and they ordered a new AW 17 for me. Whilst neglecting to collect the previous one...

Third AW 17 arrived and worked OK, cannot undervolt to over -135 mV and be safe, maybe #2 was a bit better in that area even though it's a curve so who knows, but nevermind, I am not intending to open this one up, or, am I? It ran OK with factory thermal grease and no alterations from me, until a couple of days ago where it also throttled down to 800MHz. Is it again the VRMs, I wondered. But looking at the charts now the CPU is really at 100C at just 40W, and no power throttle at 94C to save the day. It seems this variable power throttle which is designed to keep the CPU below 94C does not take effect for very low powers. I therefore suspect a badly fitted heatsink and/or crumbling thermal grease, AGAIN! How much will it cost them to use good quality thermal grease!

Shall I send this one back too? Or re-paste it. Re-pasting takes time, many hours and do I want to mess with it once more? At the same time I am not good in removing the WiFi card, I ended up breaking the connectors on the previous one, it is so small and fiddly I have no idea how to remove it without snapping it (it comes off the PCB!)… For the time being I can run my games OK with reducing CPU power down to 25W, I do not lose any FPS, so why bother you might ask.

 

I checked iUnlock's guide but where is the keyboard? Am I missing something simple? Sorry if it's a stupid question

Yeah, and that's what I think may be a problem because, according to my sources, the 4-zone lighting keyboard for the R5 has *two* ribbon cables that connect the keyboard to the MB. I'm not sure if the ribbon cables are both attached to the main KB or if the other one is meant to attach the macro keys to the main KB.

Yeah, I agree -- in cost terms it wouldn't make sense and from what I've seen over the years with electronics, for the most cost savings, many PCBs that can accept different chips or connectors just have contacts in place of a chip or, they have a connector that isn't plugged into anything.

I'm pretty sure that the Alienware Command / Control Center would probably recognize the new keyboard as it's just a USB device and the software probably has all the profiles for the different keyboards.

Thanks for the tip I've heard that the R5 needs a complete teardown to replace the KB, but your saying I can get the ribbon cable connectors to test the KB without having to go through all the mess of complete disassembly?

 

I've read a lot of people complaining about the high temps they're getting on the R5, especially with the Intel Core i9-8950HK. I have an R5 with that same processor and I was getting temps upwards of 100C even when I first powered on the laptop. I was able to overclock the processor to about 4.8GHz using the Alienware CC, but, it was a dynamic overclock with significant power and thermal frequency throttling.

I then bought three different notebook / laptop coolers and I finally settled on a Thermaltake Massive 20 RGB, which was one of the only coolers onto which my laptop would fit. I did not see any significant drop in the temps though. I then got a tip from someone on AlienwareArena to undervolt my CPU so it would have more thermal / power headroom in terms of Intel Turbo Boost. Well, using the Intel XTU and then Throttlestop, I was able to undervolt my, now non-overclocked CPU, with a -0.150 voltage offset. After that, I set my system cooling policy to "quite" and I've been getting idle temps at around the mid 50Cs and load temps from 60C to 80C with occasional 90C+ spikes.

I've heard that the problem is that the "Intel Core i9-8950HK" is actually a binned Intel 8700K that requires more voltage, hence the heat problem.

 

From my experience, re-pasting will be needed with any kind of paste. I do not want to replace the thermal pads anymore. Does not worth the hassle. UV is your best bet to keep temps low regardless of thermal paste or LM. That why I want to keep my #3 and try to see what I can. I am thinking of using the ICC graphite pad I got. What do you think from your experience? Was it keeping the CPU temps good below 90s during CB and at what frequencies. For the GPU you could use a thermal paste as the GPU temps are 99% not a problem on the R5.

 

@doofus99 I never realized the extent of your story. (we're like a support group right, we each have stories) I believe you're thinking aloud, but my opinions:

1) You know deep inside you're going to open it up and paste and maybe re-pad it. While my second unit would have been very acceptable without, it still works much cooler with the liquid metal. (as to whether re-paste with traditional paste would have made a difference - I don't think much; but I also didn't like the idea of leaving those Arctic pads in there which, from my first unit, seem to deteriorate within weeks of being new from the heat - I consider them garbage after the state I saw them in)

2) There is no difference in motherboards among different countries, or is there? It sounds like very bad luck that you keep having those throttles down to 800 Mhz due to the VRMs. I don't monitor clocks/load/temps as much as I should in-game, but I think I would notice the effect of the CPU down-clocking to 800, right... except for when I first upgraded my panel to UHD and had some jerkiness, it didn't persist at 1080p, nor after I changed from GC Extreme to LM. So I think I can say that I have never seen this strange 800 Mhz throttle. In game, all my cores do zig zag from 3600-3900, and heat goes up to low 90s (I think those are spikes though, when I view the graph, it seems to be in the 80s), but never any throttling from what I can see.

I guess I was never clear about it from what you said, but are you using those Panasonic pads on the VRMs that you linked to? Maybe they just aren't working out? You're definitely seeing nice imprints in the pads when you remove them to prove the pressure was on them? And did you let the pads extend over all the edges to soak up more heat if possible? I am just mentioning all this because ever since I had my first experience with the Grizzly Minus 8 pads, even though they're fairly solid, once I saw how much they cooled my SSD drives, I just had a good feeling using them throughout the heat sink areas as well. (VRMs and all the others)

3) I think even if you gave them good grease, it wouldn't matter. Toothpaste works ok as a thermal grease if applied properly. Their problem is lack of assembly line precision. My second unit had perfectly shaped paste marks, CPU and GPU separated beautifully from it, good consistency. First one had way too much paste, and what was under the sinks was watery and lacking. Maybe the higher heat of that CPU disintegrated the compound and the pads more, I don't know, but it looked awful. (I can only imagine if they applied LM at the factory, maybe 1 out of 5 laptops would boot upon delivery, the rest would be fried!)

4) You really shouldn't have to downclock to reach 25 watts, that's insane. (I mean, you shouldn't have to accept that as a solution, although I know you're frustrated and willing to accept a bit more rather than trying over and over - I felt the same way.) I tested my 4.6 Ghz OC last night and even though it crashed once while doing nothing, it did not crash at all over several hours of playing.

5) On the WiFi card - have you bought the proper tool for removing the antenna leads? It pinches the ball part I think and spreads or clamps them together perfectly. That said, I was worried about breaking them (iUnlock calls that part out and has a video), but honestly, if I grab the edges of the wifi card in one hand, pinch the lead top to bottom with my other hand, and smoothly but firmly push up, it just unclicks easily for me, and I've done three such disconnects with no issue. I just ensure I am pressing straight up and away from the WiFi card (as much as possible anyway, by pinching very close to the end of the lead). Getting them to pop on is more nerve-wracking for me. Then again, I did used to do a lot of model airplanes and such with tons of small part assembly, so I am sort of used to how much pressure to use on fragile parts when gluing them together without breaking them.

On the guide, I meant, you can see where the various keyboard connectors are at the bottom. At the right are the speaker and battery connectors, then in the next cutout are the SATA drive, mouse, keyboard, and new keyboard connectors. (the last few are not that order).

So if you just take the bottom panel off, I would think you could plug in your new keyboard, and power on with laptop sitting on its side, to see if you're good to go or not. The per-key RGB keyboard does a rainbow color shift while it POSTs, so you will know right away. But yeah to replace the keyboard in its final spot, EVERYTHING needs to come out. Basically a couple steps more than you would do following the iUnlock guide for re-pasting. (so be sure to do whatever else you want to do in there also! Such as re-paste, adding padding under trackpad/keyboard and ensuring they feel tight before you put motherboard back in for good, etc. (if that matters to you - it did in mine, on one keyboard was loose at top, and on both, trackpad had a tiny bit of movement, but putting padding underneath the battery and tightening the screws for it fixed it)

Regarding the two cables on the four zone - I cannot speak to this. Are you sure one is not keyboard, and the other trackpad? I don't remember now but I don't seem to recall a separate cable for the trackpad. (but I do remember seeing two more ports that were left empty, and I think one said KB for sure, and I thought the other had an M-something, but I could be imagining that - I guess the iUnlock is not that useful since it's an R4; M could stand for macro though too) Anyway, my point stands, since the R5 is sold with both configs (4-zone and per-key), I am pretty sure what I saw are three ports - two for the old style, and one for the new style, which I have. (and the new style also has a short and a tall connector - the cable fits in both, but as I found out, only the short one actually works for my keyboard; so I guess actually there are three empty ports; maybe that taller one is for the four-zone keyboard)

 

No no! I am trying to avoid having to do this. It's a process that takes hours but Dell ought to fix this, not me.

Oh yes, from 70FPS drop to 5-20 FPS in the game, and it's not even a demanding game.

No, I have not opened up this laptop. It's "as is" from factory.

In any case you cannot use graphite pads on the VRMs, the gaps are way too large, the pads would not make any contact and would simply fall off, and it's extremely dangerous as the graphite pads are electrically conductive.

My experience is the opposite, quite dry and brittle, "cooked", thermal paste, in the form of a "stamp" - it's like a thermal pad really. I believe Dell do not use thermal grease on the CPU and GPU, just those "stamps".

Today I had to constrict the power down to 18W to let me play the game without FPS drops.

I did not know there was a tool - do you have a link?
The problem is removing the leads not inserting back on. As I try to pull the leads/connectors off the WiFi card, the connector snaps right off the PCB. I do not try to pull at an angle, I really do try my best. On Laptop #1 the leads would come off easily no problem, on laptop #2 the connectors seemed way too stiff, even after half a dozen re-inserts, and eventually the Bluetooth one snapped right off. Took me ages trying to remove the broken piece off the plugs at the end of the leads, using watchmaking precision tweezers - they are absolutely minute.

I wrote to the Dell customer support explaining that laptop #3 is also suffering. Let's see what they come back to me with.

 

@

@doofus99 - I can't find the tool now, but I KNOW I saw one in a thread somewhere. $7USD or thereabouts. But since I'd already gotten mine off once, didn't feel the need for it. I tried for ten minutes, just couldn't find it again - it's probably a generic term for that connector style. Looks like a much smaller version of this ball joint popper for autos:
https://www.amazon.com/dp/B01LFG1LKK/ref=sspa_dk_detail_0?psc=1

iUnlock takes them off with a plastic pry bar while the card is still in the machine, it's about halfway through his video (it's a short video but he gets into it at 30 seconds) - he seems to think prying with a left to right twist is better based on where the solder points are:


I agree on the paste, I think it's a stamp of some sort. From the first one, I thought they rubber stamped a tray of cheap paste, but now I think it's just a thin slice like those wrapped cheese slices or a thermal pad shaped to their needs, and they just slap it on there.

I forgot those Panasonic pads were also graphite and conductive. Yeah that would be a bit risky, hehe.

Overall though, it definitely shouldn't be having those issues out of the box. Just get a new one until it works decently I say, then you'll know if and when you do re-paste or liquid metal it, it will be awesome, instead of just mediocre. I am sure they found a way to use some bad parts somewhere on the motherboard. I had an MSI or Gigabyte or maybe another cheaper brand motherboard smoke on me an hour after installation on a custom PC once, couldn't believe it (and I am 99.9% sure it wasn't shorted - it did work for an hour, replacement never had the issue mounted on the same posts, etc.). (have stuck with Asus after that and never had a similar issue)

If my second unit had been worse than my first, or if I'd had these other issues like what you're experiencing, I'd be singing a different tune, that's for sure. From your experience, I think you'd be happy even with the first unit I had (which was really just bad paste/thermals and maybe a "worse than 80% of the field" CPU since it won't UV as much and runs more watts just to get factory clock; but GPU runs cool, haven't gotten that over 75 yet in any game, remarkable). For you to have had multiple units with the VRMs overheating is staggering. I really wonder if there's something about the supply chain that affects the likelihood of such issues to different countries. (radio frequency/EM radiation limits or something that would necessitate a different motherboard? I mean, I highly doubt it...)

I also think it sucks that you guys only have 14 day return policy there. 30 days is hard enough! (I really like Costco for their returns, I think it's 60 or even 90 days! Which I feel is what one REALLY needs on a high dollar electronics item like cameras or computers - even without issues, it takes a lot of time to load stuff and really judge if it's going to work out. Cameras take a lot of effort to learn the features and controls, etc.) I know you're past that now. Hopefully they will keep being plentiful with the replacement units though!

@jdb2 - I found the picture I was thinking of; I don't think the ports are the same on the R5, but I know there's still about the same number as in the bottom of this picture:
http://i.imgur.com/5ZrXCkA.jpg

So I guess on the 4 zone it's keyboard, keyboard backlight, and then MK (macro keys? not mouse I suppose) But on mine, it's just one cable for those three, plugging into the one approx. where the one named "keyboard backlight" is in the image. (and there's two on mine in that one area, one behind the other, shorter one in front, which is the one I use - I can't tell if there's two there also or not)

 

I see, I think I did the same, there are not many ways you can prize them off except to pull upwards from the actual wire and that means at an angle - you can clearly see the wire bending as he is exerting force... But in my case, the base half snapped right off. His camera is excellent by the way, really good clarity and focus at so close a distance.

 

Laptop #1 had the CPU overheating at any provocation. Needed a repaste badly but I never opened it.

Laptop #2 had the CPU overheating much less, and I should have left it alone, but I repasted to very good results, but then got the GPU VRMs overheating.

Laptop #3 : the CPU is overheating (not the VRMs) with the GPU at full whack. I say this because the CPU hits 100C but no throttle fixes it and this is puzzling. You have temp throttles, power throttles, current throttles. The thermal throttle kicks in but in spikes and eventually it drops down to 800 MHz.

 

Yeah that was probably me saying that. Yuu indeed need to remove everything from the base plate to get to the keyboard. The keyboard is underneath the aluminum/copper base plate where the DC powerjack cable is adhered on. So you also need to slowly and carefully pry that off. The powerbutton PCB etc is also mounted on this plate. So in short, you need to remove everything. If you have never disassembled a laptop like this I would strongly advice to leave it as is. It's not an easy task and even a lot of Dell technicians screw this up and ruin a lot of plastic tabs.

From what I saw the RGB keyboard has 1 ribbon cable more for the RGB stuff.

This is a video of the whole process. as you can see suddenly a few of thos eplastic tabs are laying next to the laptop in the middle of the video. Even Alienware technicians have issues with this.

 

true. the rgb ribbon is underneath the keyboard ribbon. i suspect the keyboard has an extra ribbon that goes into that connector. Dying to find out what the right part number is so i can order one

 

I have a friend working for Dell who might be able to find out. Will post it here if i know more.

 

So Dell have go back to use thermal stamps for the Cpu in the Alienwares? Aka they haven't learned nothing from the past.

 

Does anyone know if LM is going to eat the heatsink material with time? I know it is causing discoloration but it only the outer colour or is it the heatsink material too?

 

Maybe it does join two into one somewhere... I know it's just one cable though to plug in at the motherboard, not two like before. (and I know that because I plugged into the taller one and the keyboard didn't work or light at all, nor did the macro keys, until I plugged it into the shorter jack - so it seems pretty sure thing that all keyboard comms and lights are on the one cable when it reaches the motherboard)

All I know is I did a lot of "research" (aka searching) and the only two things I saw that I didn't like were the potential to change state when in freezing temps, and then break when warmed up, and of course the potential to relocate and short stuff out.

I did just find this, which says it can pit copper:
https://www.osti.gov/servlets/purl/811932
"However, the discussion by Burton8 indicated that galinstan does react with copper, even at low temperatures. Liquid gallium at ~30°C also slowly reacts with copper to leave the copper surface pitted.9 This chemical reaction with copper is a concern if gallium or galinstan spills in the laboratory. Both gallium and galinstan are electrically conducting; thus the staff must verify that the spill is not in contact with any electricity before approaching to clean up the spill. Fortunately, it appears that neither gallium nor galinstan will react with electrical insulation."

Obviously though, anyone who uses it has it in contact with copper, I am sure if the pitting was that aggressive that reports of that would have come up after all these years it's been in use.

Here's the best summary I've seen of it with heat sinks so far, which seems to have some better reasoning behind what it might do with copper:
http://forum.notebookreview.com/thr...-compatibility-with-copper-heat-sinks.800890/

I also remember reading that someone thought it hardened due to dust or other debris having been left behind before applying, not sure if that's valid. The post above though seems to be saying it hardens because the actual galllium has transferred into the copper, which makes more sense. The one guy in that thread said he had to redo his after a year but even with the tarnished look, the heat sink still worked fine after he re-applied the liquid metal.

Also I saw that the heat sink for sale on Amazon isn't that much to replace, about $120, if for some reason Dell refused to replace it. (but again, I don't see it getting completely ruined by the gallium)

They could nickel plate the heat sink like the thread above states... but then it would lose a bit of thermal conductivity. But the heat spreaders are plated as such (and so LM in between the die and the heat spreader won't really react at all, on a desktop CPU; but then between the heat spreader and the heat sink, it's still the same issue, of course).

All I know is at this point I consider the liquid metal a necessity in a high performance laptop. I just did some serious work finally on my computer, nothing too taxing, but lots of browsing, video playback, loaded up a decent sized audio project - my fans are set on Performance mode and yet they NEVER went above minimum. At stock, especially on my first unit, the fans came on medium+ just to watch YouTube videos! This is really amazing that they don't need to come on above the minimum level unless the CPU is really being taxed (or the GPU, of course). The AW is quieter than my Lenovo, lol! (maybe I will re-past this Lenovo sometime too) Really feels like I am at a desktop when doing normal stuff, with the lack of fan noise like that. The GC Extreme was good, but this just feels like the heat gets out instantly. If it requires re-pasting every three months, that might change my opinion somewhat.

 

Well, I am thinking of re-pasting because the gelid extreme seems to need to be re-pasted again. Can you also please give the link for the heatsink on amazon?

 

that would be awesome mate

 

Aluminum...not a good idea!

 

Does anyone here have Desktop Mode Battery Charging as an option now for their laptop? Just got my new 17r5 today and am having issues enabling this option.

I’ve come across many recent threads on Reddit and elsewhere for the same issue - been telling people to install OSD but apparently that is no longer all that’s needed.

To clarify - this option should be visible by right clicking on the battery icon in the taskbar.


Sent from my iPhone using Tapatalk

 

That sucks the GC Extreme is failing you already. Temps are much higher, I gather?

I couldn't find the same part on your UK Amazon site - it had this one but looks different:
https://www.amazon.co.uk/RTDpart-He...1038666&sr=8-4&keywords=alienware+r5+heatsink

On the US site, I had seen these - they say R4 but look very similar (but perhaps not):
https://www.amazon.com/Replacement-...541038728&sr=8-4&keywords=alienware+heat+sink

https://www.amazon.com/HK-part-Repl...541038728&sr=8-3&keywords=alienware+heat+sink

If I REALLY had a lot of time I would even try getting one and seeing about planing it. I read on iUnlock's post on Reddit that his service can include up to that level of treatment (leveling/sanding the heat sink and the dies, plus the rest of the treatment with pads and paste). I sort of think the liquid metal gets you 90% there though since it seems to be filling in the nooks and crannies. Of course I don't know without experience.

 

I received the ICC contact test kit. When I have time I will try. I am yet to find if the temp spikes on certain cores are due to imbalanced heatsink or due to those cores being used most frequently causing instant and one sec lasting temp spikes

 

Cool, please let me know how that contact paper works. iUnlock also does that as part of his service, I read. (similar contact test paper) I almost went for it but I knew I didn't want to spend more than 3.5 hours on it.

 

Its not available since the 17r4 models.

 

Any idea why? That is a solid feature with real world benefit.


Sent from my iPhone using Tapatalk

 

In my experience, having done this a dozen times, when you fix the heatsink onto the mainboard with the 7 retaining screws, as you tighten these screws you can visibly observe that the mainboard flexes upwards. The heatsink is very rigid and spans a very wide area over the motherboard, and while you are bolting it down around the GPU, you will notice that you are indeed pulling the mainboard up ( screws 4 and 6 in particular if I recall, those below the GPU).

The finished assembly then consists of a rigid and, we hope, perfectly machined heatsink and a motherboard that has curled upwards around the GPU screws, depending amongst other things on the thickness of the thermal pads on the VRAMs or on the GPU itself.

You can always loosen those screws a bit so as to relieve the tension on the motherboard, but by doing this you are sacrificing your contact with the GPU and especially with the VRAM and the VRMs sitting below the GPU.

When you then place the complete assembly back into the laptop's "chassis" you will notice that as the mainboard has curled upwards, it is now rocking over its middle section from top to bottom, Ie from the rear of the laptop to the front, and while you are trying to tighten those screws in you need to flex it out by about 2 mm or so !

Initially I thought there was something stuck under the mainboard and that's why it was rocking over the middle section, but I took it all out, looked carefully and discovered the friction marks on the motherboard and the back plate that sits over the keyboard. There are a few raised tabs on that backplate and the motherboard was rocking over one of them in the middle area somewhere.

Conclusion: a totally crap design. The heatsink should be at least two separate pieces to allow it to follow the contours of the motherboard better. The heatsink extends over the VRMs and VRAMs but in a very haphazard manner with large gaps of varying widths, with areas left uncovered, lack of retaining screws etc

Conclusion: every single time you remove the heatsink for any reason, even to access the keyboard, you will get different thermal performance and you will not know it until you use it in earnest and run tests, which at one time forced me to run with the laptop opened up and on its side for hours so as to make sure it was cooling all parts sufficiently before committing to putting all the covers back on.