*OFFICIAL* Alienware 17 R5 Owner's Lounge

I did say about NVMe EVO drives.

 

I did not notice any issues though. Maybe performance is slower than it should but not noticeable at all. I feel really fast (also check my benchmarks a post ago).

 

It was my understanding from reading many reviews that the latest Pro is actually only superior in some write tests, equal performance otherwise (this is comparing the biggest Pro which is now only 1 TB to the 2 TB 970 EVO - and ONLY the 970 EVO, the smaller ones are slower, yes). If writing a large quantity of data and using up all the write bugger, then it is faster at that point also for sustained writes (it does slow down though, but only once; the 970 EVO will have two slowdowns after using up the cache), with the second slowdown being half the Pro's extended write speed - IF I remember correctly).

The 970 EVO 2TB (again, only the 2 TB) also has the same lifetime as the Pro, and (all sizes) have the same 5 year warranty. I wouldn't buy the Pro anyway, just too expensive for technology that might advance rapidly now that Optane is putting some pressure on, but even if funds were a small consideration, are there other reasons to get the Pro in that case?

Not saying my summary above is right, but I did read many tests for a couple of hours, if that counts. [EDIT: I did double check, and the 970 Pro is a 5-year warranty; but the 860 Pro is a 10-year warranty, which I didn't know.]

 

Wow, this is some awesome feedback. I actually got a return extension from Amazon (woohoo!), as today was my last day. (Now I am even more glad I did!) I was thinking already, even though the 970 EVO 2 TB is cheaper than ever, maybe I will try something else - just because other reviews say they're equal or Samsung is a tiny bit faster, maybe in this computer, that's not true. (since my results prove otherwise) You seem to have the same sentiment/knowledge.

Heck, I was beginning to think, maybe I will just go for an m.2 SATA 2TB drive instead! But I'll give one of the other brand NVMe drives a shot. I'd get a Pro if I didn't need 2 TB (and if it didn't cost so much), unless I see a deal on an older Pro 2 TB or something.

I didn't know that the SM961 mine came with is MLC - so is this more based on their enterprise drives? (or similar to the consumer Pro line at least?) I don't get the 4K speeds seen in this review though either, on the 4k:
https://www.tweaktown.com/reviews/7762/samsung-sm961-1tb-2-nvme-pcie-ssd-review/index.html

I wonder what it is about these machines that doesn't get along. I guess there are still more ideal and less ideal combos of controllers and drives, just like always.

I wish there was a sticky "helpful hints" board for this model, I'd be glad to add my findings and discoveries, such as this... would sure save a lot of Alienware newbies frustration too.

 

In my opinion, the issues is 0 air flow for the nvme drives. Even just a little helps. I tested this on one of mine by operating it with the bottom cover off. With the cover off it functioned with out issue (still hotter than I liked but worked), put the cover back on and it stuttered and cut off etc.

Sent from my Pixel 2 using Tapatalk

 

I can't find a use case for Optane at all when it comes to our laptops. Optane is for folks that don't want to dish out the money for PCIe SSDs or even 2.5" SSD but want to improve performance with a SATA hard drive.

The only use for Optane in these Alienwares would be the base systems running without any sort of SSD.

 

Intel finally changed it so you can use Optane to boost any drive, not just the boot drive. But yeah, with only two full-size m.2 slots, wouldn't seem worth it to waste one on a cache like that. Even beyond that, not really sure how much positive result caching an SSD would even have. If it really makes them work like Pro NVMe drives for shorter transfers, could be cool. I totally agree though, main scenario is for boosting HDDs.

 

Uggh, well, now I know why I chose the 970 EVO 2 TB - it's the only non-Pro/expensive 2TB m.2 SSD available... (besides the 860 EVO) I was so sure WD Black came in 2 TB but it doesn't. I still don't know about the performance being abnormal though, even PC Mag only got 50 on the 4k read (granted, on a 500 GB model). I know their 4k write is 200+ but 4k write isn't that important and again, since it's cache boosting the write speeds anyway, I don't really trust any benchmark for those.

https://www.pcmag.com/review/362944/samsung-ssd-970-pro

Maybe I'll swap it out for another of the same just to be sure mine wasn't lackluster. Only other option is the 860 EVO. Might go for that actually, $200 less, same 1,200 TBW expected lifetime, and 5 year warranty. 4k read is 2/3rd that of the 970 EVO. I wonder if that means, I'll get 80% of that speed - on the other hand, if I actually get the full advertised speed for some reason, then it's not much worse than where i am at with this 970 EVO, so then it seems appealing. Heck, if it sucks too, I can swap again.

Was going to use a desktop to clone, but I didn't even realize they make USB 3.1 <-> m.2 conversion enclosures... makes sense, just never thought to look for one. Pretty cool.

EDIT: Didn't realize Intel had a 2TB in their 660p line-up as well as in their 760p series. Interesting, might have to go with that, very promising 4k read/write speeds.

EDIT2: I realized in time that those cheaper ($10-20) m.2 to USB converters are for SATA only; although there are m.2 NVMe adapters also, but more like $60-80.

 

Well if anyone cares, I decided to go for the Intel 760p. A bit pricey... would have really liked the cost-efficient 660p, with its new QLC, but, TBW/lifetime was a bit low (although still 5 year warranty), and also, Amazon only had it via third parties, and for much higher cost than NewEgg and eBay. I don't enjoy shopping at NewEgg anymore, they always delay orders for CC verification, bunch of BS, and I prefer Amazon for return if the drive sucks - luckily they do have the 760p.

Will be very interested to see if it does indeed perform more to par with its advertised rate vs. the Samsung EVO did. Very promising 4k random numbers - 67/194 on the read/write, with that read speed being a good bit faster than this place tested the 960 Pro at (basically similar to 970 Pro!):
https://hothardware.com/reviews/intel-ssd-760p-nvme-pcie-solid-state-drive-review?page=5

Sequential isn't at Samsung speeds but not bad either.

 

You could have looked at ADATA SX8200

 

Thank you for the suggestion - I don't see that it comes in 2TB though ( in m.2). Ok, never mind, just found some reviews of it - seems similar overall, maybe worse on some tests and better on others. The weird issue is, I can't seem to find it for sale anywhere! Only up to the 1 TB for sale wherever I look. So, I think I'll stick with the Intel. If it stinks, I'll probably downgrade to a SATA m.2 and call it a day.

 

If your AW baby has copper surface when you use dual sided NVMe SSD from Samsung then its good otherwise it'll thermal throttle under heavy loads. So you might have to add thermal pads to both sides.

 

EK NVME heatsinks come with thermal pads for both sides too.

 

Good.

 

Does anyone know the part number for the high capacity heat sink fitted to the 17" R5? I am getting a 8750H and a 1060 GPU but I want to replace the heat sink with the more capable one.

 

Did not know there were different heatsinks in terms of performance?

 

Just wanted to update everyone, I did my re-paste. Also changed all pads out to the grizzlies. The original pads were literally tattered! Some were just flaking off and leaving behind the woven fabric of the original tape/pad. Not at all like I thought it would feel like.

I tried disconnecting the fan shroud but the pipe is connected to the lower shroud, so I did indeed have to remove the motherboard as all tutorials/videos show.

Total time - about 3.5 hours. Using Gelid GC Extreme, and Thermal Grizzly pads.

I cut the pads, but not into small pieces like factory does - don't see the point of that, just use a strip, as far as I can see, why not. Very hard to judge whether to use 0.5 or 1.0. I ended up using thicker than iUnlock's guide, I figured it would be safer. I still used 0.5 in two places, but not all the places he says. It's different for each build anyway, right. Maybe it's not allowing as much pressure on my CPU though? Some ICs in those rows have the end one shorter, so for those I made them level by adding extra pad to the short one, obviously. I just hope using thicker, since some of those border the cpu/gpu on multiple sides, didn't keep the sink from getting maximum clamping pressure on the cpu/gpu.

Oh, also - instead of putting the pads on the sink, I put them on the motherboard/ICs first, then took the top layer of tape off when I was ready to put the sink on (and thermal compound applied). Much easier that way. Then you can actually see what chip each piece of tape is working on, gauge height differences, etc.

I wonder if we were to add fan extension cords for the two blowers, if we could just remove the heatsink WITHOUT removing the motherboard? I would definitely try that next time. Of course, don't want lots of fan wire to get pinched, shorted by screws, clogging the fan blades - but man, if they could just have put a disconnect in the wire, above the motherboard, I don't see why we'd need to take motherboard out at all!

Some of those cable attachments, especially the ones to the top shell/monitor on the back of the motherboard - I would NOT want to be redoing those connections too often. Got quite scared seeing the strands of wire under the tape, thought I'd ripped them out of the connector, but luckily not.

I'd say the R4/R5 vs. the R3 is like a Ford truck vs a Dodge - the Ford requires a 12 hour cab removal (whole body lifted off frame) to do any major engine work, Dodge does not. Would be so much easier if it was like the old one and just remove the back panel, and bam, it's all there. But overall, I didn't find the construction to be bad, just wish it wasn't so complicated to get the heat sink off.

I put the Gelid on both dies and on the copper, after cleaning very thoroughly; then used the scraper tool to trowel it as evenly as I could and thick enough so I couldn't see the mirror surface of either die showing through (and same with the copper). I figured I couldn't do too much. Should I have added more? I used a lot of the tube, more than I thought it was supposed to take. Not majority, but seemed like a third perhaps. Was expecting to use no more than a fifth.

Cleaning took 30-40 minutes, there was so much factory paste everywhere, even spreading onto motherboard (must have used almost half of my Arctic cleaner first step bottle!) - it looked like the factory paste on the CPU contact area was minimal though, and was LIQUID, even though I had not run up CPU in 12 hours (it was cold). This might explain why I've been getting worse CPU as it gets more use? I didn't know pastes could pump out, I thought that was only the liquid metal.

As for results - I still think I have a bad CPU, unless my re-paste wasn't perfect. Now, one or two cores get hotter than the rest by 5+ degrees (maybe that's just normal?). But they are DIFFERENT cores than the ONE that use to be the "hot" one! The best I can in Alienware Control Center with just a blip of a throttle is to run 4.3 Ghz with -100 mV, that gets me a 1250 score. Going down to -135 mV, I can get a stable 4.4 Ghz with Throttlestop. I almost got 4.5 stable but computer still crashes around -160 mV. It might still crash at -135 too for all I know. Highest Cinebench score I got using Throttlestop was 1378.

So, I was really hoping to get to 4.5 Ghz stable, but 4.4 is hard to get. Will probably just run at 4.3. If I do that at -135 mV, only one core touches 90, maybe 91 degrees. I will see if the replacement does any better stock. If so, then I will do another re-paste if needed on that one. Otherwise, I've done so much work on this one - don't really want to redo this and buy more pads, etc., unless it looks like I've got one that can do 4.5+ steadily.[/FONT]​

Good news, I saw that middle part that sits in the rest of the steel backing tray, that's exactly what makes the top of the keyboard feel loose - I put 1.5 MM square of pad on it, and it almost fixed it perfectly. Probably 2-3 mm would have been perfect, but it's good enough now where my brightness function keys do not rattle so much - a bit loose but I would never have noticed when new like this. I also added pad being the trackpad (forget how much - I think 3MM also? So now when I tap hard, the pad supports it and it doesn't click/sound jinky/loose. The battery has to come out to do that (the battery is what provides the backstop support to the pad I added). Anyway that fix is perfect.

Anyone else think I just have a worse than average chip? I am satisfied though if this is the best I can get. Not near what a Clevo can get, of course, but still smokes the MB Pro/Dell XPS with the i9, so I don't feel ridiculous anymore (when this was getting, stock, 980-1030 at first! lol). Still kind of crazy I had to go so deep and spend time and money on materials to get it to this point, but a stable 1350 score (this is with everything else running on computer) is a nice ways ahead of those other i9 laptops. Even 1250 makes me feel all right.

I am hoping the paste might set more as I use it like others have reported, and maybe give me better thermal transfer tomorrow.

Fans turn up and down much more quickly now. As soon as test is over, fans go down to minimum, temperatures drop to 45-50 range in no time. So I think my treatment was pretty good. The transfer of heat is working well, right, if the temp drops in a snap of a finger when the testing is over? So other than that it's the overall cooling system and the efficiency of my chip limiting it, I believe.

One other question - has anyone else tried using a pad instead on the CPU and GPU, to be sure it's getting great contact? I also keep seeing those conductive pads on Amazon. But from what I've read, paste will always be better on a die and heatsink because it fills the imperfections better and adapts if it's not a perfectly level plane. But isn't that true of any surface that needs cooling?

 

This is the proper way to do it especially if the heatsink does not have easily identifiable spaces to put your pads on.

Yes, I have thought the same, it would save a ton of trouble. Must buy the extension cords before opening it again. On refit, you would still need to undo the 3-4 motherboard screws to loosen it because it flexes and you will not get a proper heatsink fit if the motherboard is fixed onto something else which is not perfectly level.

You have to go very very slow on these, especially the monitor connector. Do not pull from the wires, use a magnifying lens and pull from the metal body of the connector, it is rather hard to do.

Dell use thermal pads ("stamps") on CPU and GPU - are you sure your laptop was new ?

I used pieces of electrical tape. Did your method fix the "clicking" sounding keys that we discussed a while ago?

It is only Panasonic that make the compressible conductive pads and you'd better buy from reputable suppliers. If fitted properly these pads beat the cr*p out of thermal pastes, almost as good as LM.
https://uk.farnell.com/webapp/wcs/s...estType=Base&partNumber=2723038&storeId=10151

 

Hey Doofus, thanks for reading my "essay" and for the feedback! Glad to hear someone else had similar thoughts/observations. I feel like I am playing a game working on the laptop in a way, like, gathering the parts, upping the knowledge, then completing the quest, lol.

This laptop was the first one I received from my order direct with Dell/AW. So I would assume it's new. It looked like pictures of the factor material - the GPU one still had a square shape too it (perfectly cut), but had pooled away from it, and the CPU one was just a mess, all spilled away from the actual CPU, as it were.

As for my keyboard/trackpad fixes - trackpad is 100%, no movement; keyboard, I'd say 75-90% fixed. Might be as good as it gets - honestly, the keys no longer mush/feel soft; it's just that the sound those Function keys make still sounds a bit more hollow, but now my mind associated that with them being on the top row/edge of keyboard, so it feels/seems ok. I don't think I really feel any movement anymore unless I press really hard. The few top row number keys seem perfectly fine now. I checked where that copper inset plate (not sure what that part is even) goes against the motherboard, and the motherboard has that metal frame/skeleton as it were, and one of the very stiff ridges of that rig goes right where the movement is in the copper plate that wiggles, so I think the pad I placed there is ideal for bracing it; i.e., it shouldn't push on the motherboard itself, just on that frame that is attached to the motherboard.

You used thermal paste on yours, right? Sorry, I forget. Have you considered those pads? Everyone seems to have different opinions on liquid metal - some say it's just 1-2 degrees cooler, others think it's a world of difference. Think I should go liquid metal next time? I didn't because I am too worried about pump out and also I read that one of the metals in it changes composition at freezing point, and can crack the chip it's applied to. Since a laptop could easily get stored in freezing temperatures, didn't sound very travel-worthy. I thought I saw actual tests showing it's not that much different though anyway.

Do you think I've reached the max of this CPU chip? Or if I'd done a better job somehow, I'd be getting more headroom?

I think my CPU is not great, because even though I can run Cinebench as many times as I want at up to -155 mV, a game will usually crash within minutes, even at -135 mV! I am not even sure if -120 mV is stable, to be honest; I usually just go back to the Dell OC tool at -100 when it crashes and skip ThrottleStop.

 

I was referring to the keyboard only. As seen on the photos I have posted on this thread, and as you may have discovered, the keyboard is held down by means of 4 tabs on the top row of keys. There are large gaps between the tabs and especially around the ESC / F1 keys - there is no support whatsoever. Similar issues around the F7/F8 keys and the DELETE key. These are areas most distant from the tabs and so they flex as you press down on a key. The area around the ESC key fares much worse because there is a disconnect with the vertical column of the macro keys. The whole assembly is held down with the motherboard and the back cover over it. If you remove the motherboard and try to type you will notice there is no rigidity at all. I added small cuts of electrical tape below the ESC key and also below the F8 and the DELETE keys (but less). That stopped the flex completely.

However there is a new keyboard problem that someone mentioned here, which is a click noise coming out of a few keys only. On my keyboard it's the ESC, F1, F2, `, and TAB, 1, 2, 3 keys mostly. They are quite loud especially in a very quiet room when you can hear only these keys. There are different sounds from large keys like SPACE, ENTER, BACKSPACE, those rattle a bit. But the new clocking noise is well, a new "defect" and has nothing to do with the flex.

 

I used Arctic MX-4 I believe, it was OK-ish, of course much better than the factory dried up stamp. I am not sure why yours was runny, the ones I have seen are usually dry. Runny may be better than almost completely dry powder.

However as you have changed all thermal pads, bear in mind that the worse area is the VRMs sitting under the GPU. There are photos showing them, I was getting 104C there. The test, I believe, is Subnautica. It's also a good game. What happens, maybe, is that once you "perfect" the CPU and GPU with graphite pads or LM, you then expose other areas of concern which may have not had a chance to cause problems because the CPU would cut out way before they'd get so hot. So you fix the CPU and GPU and then those VRMs get over 100C and the CPU throttles down to 800MHz until those ICs have cooled down to around 95C I think it was, and then the cycle repeats. I tested all this with the laptop open, on its side, running Subnautica, and I could take temp readings with a proper probe on those ICs. You will see how important and known to Dell those ICs are because of the shape of the heatsink which extends awkwardly far away from the GPU to reach over. They should have designed a better heatsink with more fixing points and a rigid base to sandwich the motherboard.

 

Hmm, I double checked, and I can't say I have issues with any of those other keys you mentioned. My only issues was with F8-F10 and 7-9, which is the center of that copper plate that's inset into the rest of the steel frame.

EDIT: I checked again to be sure - I did think the 1 key was a little hollow sounding, but doesn't feel different, have to really be checking for it to notice. Esc, tilde, and Tab (and all macro keys) are rock solid though, for sure. Just the 1 key, and to a tiny degree, the 2 key, but this is not something I would ever have an issue with, so I don't think it's the same level as what you're describing.

I hope my replacement doesn't have these new key issues!

I didn't really try to "type" when I had it all apart but I did press on some keys to see where they lined up with movement on the other side, and I didn't notice a complete lack of rigidity. But I was just reaching inside with the lid barely open to touch with one hand while observing the other.

Do you think they updated this along with the per-key RGB lighting, perhaps? I did notice I just had the small ribbon tape connector for my keyboard and trackpad to the motherboard, not the separate connectors like in iUnlock's guide. I figured that was an R4 vs R5 difference, but perhaps it's a more recent change.

For this unit in front of me, adding the backstop in the middle of the trackpad and the middle of that copper plate (I made sure that it would line up with the metal skeleton frame the motherboard is attached to) was all I needed and I must admit it feels pretty much fixed.

I didn't really look too closely at the tabs you carefully mentioned so unfortunately I cannot confirm if mine had the same layout of attachment tabs.

 

That's good info. Yeah I saw that row of VRMs with the awkward extension covering them. I kind of wish I'd screwed the heat sink in completely for a dry run thinking about it now (without paste), just to see if I could notice any impressions on the pads in all those spots. I think I made good contact though.

 

By the way. I am now using an app on my phone to check the temps remotely. This allows me to see the temp spikes, realtime power consumption, CPU frequencies and temp graphs etc at the same time (plus realtime). It is really helpful because I cannot really see all that in realtime while doing other things (eg. running benchmark or when loading heavy apps like adobe photoshop). Even hwinfo cannot display data so well (only if you do a log and open it later with a log viewer but this is not realtime). This is how it looks like (last two screenshots are during CB benchmark. I can also see the power watt and temps during CB Benchmark. However, when running on high priority, CB is using all resources (hence the higher score) and data are not transferred on the phone. So, I cannot see the real power consumption, cpu frequencies, temps etc but only when using CB run under higher priority.

https://imgur.com/a/P62nt4f
https://imgur.com/a/VKQu8oD
https://imgur.com/a/w5Js63M

 

I saw that one mentioned didn't get to try it - could be useful though in-game if it doesn't use up processing itself, rather than setting up overlays, which can use up some FPS (and don't really want that in my face all the time anyway).

 

Upgraded (?) to Intel 760p SSD

Here is my first CDM result after cloning my 970 EVO 2 TB to the 760p 2 TB. As expected, sequential speeds are slower than Samsung, but the random read got a 51 right away, so at least that is indeed improved with the Intel. Not sure if I'll notice the difference or not, didn't really get to use the old drive much with all this other debugging.

Interestingly, I ran CDM on my desktop testing setup with both drives installed, but I think the default driver; both did much worse, but with the default driver, both were about equal (with Intel still coming out far ahead on the 4K read). I really don't know about this stuff, but it makes me think, since Samsung is reliant on its own driver to get the high sequential speeds, maybe it's the same for 4K also, and not really a durable benchmark? That would explain why I see even pro-level reviews getting lower 4K speeds like I was getting, and others getting 50% higher (ranging from around 40 like me up to mid 50s).

I didn't install any Intel driver on this Alienware either, but it shows Intel NVME driver in Device Manager, will have to check to see if there's an update. Seems to be expected performance, maybe a bit lower than the review I saw on the 4K (a lot less on the write, but Samsung was the same).

 

I doubt it is using so much processing power to cause fps drops.

 

Awful results. I see Dells hardware/firmware bottleneck the ssd speed. Nothing new there. But what Intel NVMe driver version you use now? This is the last one Intel NVMe Driver v4.0.0.1007 WHQL

From Intels specs sheet...


Review of the 500GB version but the scores should be identical http://www.thessdreview.com/our-reviews/nvme/intel-ssd-760p-review-512gb/3/

 

Hmm, where do you always find these reviews with such good scores (painful, to me)... 78 MB/s 4K read, and well over 3,000 on the sequential? I did not see any score like that on reviews I looked at before buying. I mean, at those rates on their test, it's as fast as the Samsung ever gets on read pretty much, isn't it. I am pretty sure I only saw about 2,500 for the sequential read, and 58-60 tops on the 4K read. I did install latest Intel RST, but maybe I need to force a driver update to the one you mentioned? Intel's FAQ seems to say the Windows driver is the right one otherwise (it does say Intel NVME under storage controllers in device manager). I did update the firmware, also.

I don't really get how the scores differ so much in different systems - there actually isn't a controller on the motherboard involved (I was thinking SATA when I said that), right... are the slots not getting dedicated x4 bandwidth or something?

Weird thing is, my desktop testbed is an Asus Prime x270 (less than a year old), and even with Samsung's driver, I still got pitiful results on it, and that board does not share m.2 bandwidth with any slots, it cuts out SATA ports instead. Even with the driver, I was only getting 980 MB/s sequential, 32 MB/s random 4K read, for example. I didn't troubleshoot more than that, had to finish wiping the drive to return it today, but it's disturbing how these drives seem to get very different performance on different systems.

Oh, the other good thing is - with the same homemade heat sink with the 3mm of Grizzly pads, my max temp during benchmarking was only 48 degrees! I read that the Samsung SSDs are the highest watt/throughput figures, which probably explains why they run so darn hot. (the EVO, anyway, might not be that way for the PRO) So even if my speeds don't improve from what they are now, I feel like I got a win in that department.

 

Already stated what is the problem... Bandwidth issues due Dell's screwed up hardware/firmware/drivers. Maybe also due their proprietary software. See for example results from 4 generations Alienware 15 BGA models posted below. From first gen and up to the last and current from 2018. One thing as you can see goes over and over again... Problems with storage speed. This on top of the rest of problems as you probably know about.
First gen..


2nd gen...


3rd gen...


4th gen...

 

You can as well get the results from 4th gen AW17R5 which have different NVMe ssd model vs 4th gen AW15R4. But a you can see same problems. Whatever ssd model Dell put in, nothing really changes.

 

Thanks @Papusan for those snippets.

I did have to force the update for the driver to get it to the one you mentioned, kind of weird. The version number installed already was newer, but a totally different number, including screenshot for anyone who cares. Afterwards, it did show the Intel 760p (et al) in the Device Manager though, and performance went up another nice little bit. I am pretty happy at this point, the 4k performance is much better than the Samsung 970 EVO was, and the sequential read isn't that much slower than it was either.

Forget if I mentioned this but max temp was 48 degrees also (I think I did say that). Interesting how many times Samsung "wins" reviews, but when you really look at temperature/voltage used (efficiency) and the specific tests like the 4k random, they really are not on top, it's just their big sequential numbers that makes it seem so (and mainstream reviewers either don't care or get paid to go with that as the leading indicator). I found AnandTech's review to be more in-depth and useful, as usual. I didn't know they had their own tests until I read their review of it a few days ago.

 

Has anyone tried the IC Graphite pad on R5?

 

Nope - I am waiting on a test subject. Maybe I'll try it sometime. Did you see the post I mentioned it, Doofus recommended the Panasonic product for compressibility:
http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-247#post-10809514

https://uk.farnell.com/webapp/wcs/s...estType=Base&partNumber=2723038&storeId=10151

I cannot seem to find that product though. Only Panasonic pad on Amazon is actually the opposite, thermally insulating (lol, that would be funny if we installed that by accident!) - I saw this when I opened the specifications document under the image and the conductivity was only 0.020 W/m K:
https://smile.amazon.com/PANASONIC-...9830676&sr=1-2&keywords=panasonic+thermal+pad

https://images-na.ssl-images-amazon.com/images/I/61vn+4n55QL.pdf

But yeah, you mean this one right:
https://smile.amazon.com/dp/B07CK9S...7f-11e8-a0dd-5f2429a49a3d&smid=A2EPY4D0EP33QL

I might as well put an order in so I have it in case I decide to go crazy and spend 3+ hours again tearing it apart. Need to find those fan cables too so that I can avoid removing the motherboard again the NEXT time.

I do like how it says it remains solid and pumpout cannot occur. I am pretty sure this is what happened with my factory paste and is why it went from a stable 4.0 down to 3.6 over the few weeks.

Also of interest - I wonder if it's worth trying this:
https://smile.amazon.com/dp/B07CKHR...&pd_rd_r=213ad28f-d280-11e8-a0dd-5f2429a49a3d

That should really show us where the high vs low pressure contact points are. I guess it could work on the other ICs too where the pads go, if the pads are left on (I assume it needs the fit to be tight enough to touch though at least). Kind of expensive for a throwaway one time test, but I guess if one REALLY needs to know how the contact is at all the pad points and on the CPU/GPU dies, that would do it.

 

Hmm, not really sure on those pads still. Rating is 35 W/m-k, but I saw a test that showed the different rating don't really mean much - 30,000 is not much more effective than 6,000. Ordinary toothpaste is about 1,000. One reviewer theorized it was better and removing heat from hot spots - I wonder if that's true, like if it could tame the one or two cores that run hotter than the others, or if the chip isn't laid out like that. Plus I am hoping it will compress and provide a stable bridge. But it might not compress at all.

Saw this thread - person loved it, others bashed it:
https://forums.evga.com/IC-Graphite-Thermal-Pad-gets-thumbs-up-m2823920.aspx

Most seem to think liquid metal is still supreme.

I might get both the pads and LM to have on hand. The pads do sound good if they don't degrade though. Otherwise, not really seeing that reviews say they perform any better, just more convenient.

 

Does anyone know if the fan connectors on these laptops are a standard, or are they proprietary to Dell? I can't seem to find a clear answer anywhere. I might end up being best off just splicing in more wire to extend the fan cables, but an extension cable would be easier (and simpler to reverse if needed).

This is the image from iUnlock's guide that gives the best picture of the connector.
https://imgur.com/Yvip4aB

Interesting, this is a replacement heat sink and fan - looks like it comes with the tape/pads pre-applied! Hmm, so maybe Dell gets them like this too, with the paste packet, and the tape already on.
https://smile.amazon.com/HK-part-Re...ywords=alienware+laptop+replacement+fan&psc=1

Unfortunately, still doesn't say what the connector type is.

 

Hmm, this is interesting - the older Coolaboratory Liquid Pro, which is supposed to be lower quality than their Ultra, now as a rating of 80 W/mk:
https://www.coollaboratory.com/product/coollaboratory-liquid-pro/

Typo? Or they're reselling the same substance as Conductonaut and putting a slightly higher rating on it to look good?

I was ready to buy one or the other to try, but now I don't know - too many reports on this and other forums of pump out. I mean, it remains liquid at room temperature, right - so how the heck can we put enough to ensure the gap is filled, without it spilling out when we pick up the machine, or bursting/pumping out when heated? I can see why it's only recommended for flat and level application on desktops (or at least, something that will never move). Also, from comments on these forums, seems like the heat sink fit is paramount so that one doesn't need much of it. I can't feel good about using a heavy dosage of it if it's going to remain in liquid form...

 

I still don't really understand if the liquid metal ones are that much better, or not. On the one hand, you see reviews of how people got 10-20 degrees cooler than using a premium paste, but then you see graphs like this that show liquid metal is only a degree or two more than the better pastes:
https://hardforum.com/threads/best-thermal-paste-currently-and-liquid-metal.1924508/

So which is it? I've read other threads on here also - members in this thread have liked it, others had near-disaster with it (or at least, no longevity, with temperatures going higher than stock after a few months). I was all set to order it for this weekend (can only get Phobya liquid metal on short notice), but heck, if it's just going to waste time doing it and not even last, I'll just live with the GC Extreme re-paste I did already.

 

I don't see how anyone could get 10-20c difference in LM than a traditional high end paste. They'd have to be doing something wrong or spewing bull.

 

This is a good example - from experience posted there, seems like liquid metal with copper heat sink is a time bomb because it WILL solidify and basically not do anything at that point (AND be very hard to remove, perhaps needing grinding/sanding down of the heat sink to get it off completely?)

https://forums.anandtech.com/thread...ne-through-the-roof-after-few-months.2419707/

I have to assume a lot of the good reviews are from using it ONLY in between the heat spreader and the die on desktop chips, not so much between heat spreader and heat sink (or in the laptop's case, directly between die and heat sink). That would also contain it much better if it remains liquid.

Reminds me of model airplane engines - if you raced, sure, you could run 95 or 99% nitro - you'd sure get more power from your engine... but you'd also need to strap on a new engine every few races! So is this liquid metal, especially as applied to a laptop, kind of a daredevil overclocking thing, where sure, you will get awesome temp reduction for a couple of months, but then risk frying it all or at best case, needing to remove it and redo it?

 

Right, it just doesn't make sense. Maybe from a bad stock paste job or something. Hopefully Rinneh or the others who I saw in other threads will have some updated thoughts since the 1 to 2-year old threads I reviewed on here. I am holding off on trying it for now. I'll still order the graphite pads just in case I feel like trying it but I don't think they have any chance of working better than paste.

 

No need to be paid to say Samsung NVMe drives is one of the best. If you are worried about temperature/voltage used (efficiency)... Buy something else. I mean computer whitch can handlef the ssd.

You forgot that Liquid Pro was Coollaboratorys first Liquid metal paste.

Don't look on results done on test bench or on desktops better cooling.

Btw. Use the edit button. You can edit posts and add in more topic or info.

 

I like my Intel SSD very much - just wish mainstream reviews would stop using the sequential numbers only to put Samsung at top all the time (5-10 minutes of research, almost guaranteed anyone is going to walk away thinking "Samsung's the smart buy" - with no mention of the other stats). Clearly that's not what makes the drive fast for more everyday usage. And ok, maybe not paid - but, if they don't say nice things about the paid advertisers, no more advertising or less dollars spent, you know how the game works. (same as in pro photo/video/audio publications)

I don't know if those fans' connectors are proprietary - I looked at replacements for HP laptops and it looks pretty similar... just too small to see though. I agree, one would assume it would be proprietary, at least with some key slots or something. Anyway, I don't think anyone makes an extension that size anyway, so I will just splice in a regular 4-pin fan connector (hopefully it fits in there though, if not, I can try just making the cables longer - so then fan/heat shrink stays connected, but enough slack to work with it to remove and re-paste).

I know Liquid Pro is the first - that is why, it seems odd - in old postings it was rated at 32 or some such, now 80? Why wouldn't they market it and sell it for a premium then for a dollar more as the third in the line-up ("Ultimate" or what have you, with the line-up being being Pro, then Ultra, then "Ultimate"). I.e., how does one know if one's getting the old Pro vs. the new Pro that's rated at 80.

I saw some of your older posts on the topic - do you still use/recommend doing Liquid Metal, or just paste, nowadays?

You're absolutely right on the need to filter out different circumstances - desktop coolers have more capacity to cool down, plus with the older threads, those were all 4-core CPUs.

Regarding Editing posts - I thought it was bad etiquette to edit more than a few minutes after posting online? The idea being that any alerts that are sent out in email will contain the original unedited version only, same with search engines, so the edit might only be visible in the thread itself (and thus edits better for simple things like typos or wrong number entered). I will try to Edit more often though. I already do, I think this was the first time I multi-posted. You caught me!

 

Up to the buyers do their due dilligence. But You can't walk away from that Samsung drives is one of the best. Have been like this since the 850 Pro was lounced.

Don't expect those fans' connectors work in all other brands models. Same way as the opposite.

I still use Liquid metal. Conductonaut. I don't use Coollaboratory any more(replaced a long time ago). They have changed recipe / prescription long time ago. Even Silicon Lotttery changed/skipped Coollaboratory 1.5 years ago.

4 cores cpu or more cores doesn't matter.

Edit all the way. If not... Threads will be a mess.

 

If Samsung MLC are having a price difference of 20-30$ its worth every penny. Otherwise its not worth overpaying. Then again, Samsung drives are hard to kill based on endurance because they last really long. The endurance is simply atrocious.

 

On my previous system I used LM. Temps were really good and could reach the power limit (110W ) with ease. But I also replaced the CPU pads with K5pro. The amount of LM you use is more subjective. Generally the more you put the better but you do not want to put too much for obvious leaking reasons that can destroy everything. A mirror effect should be enough. But finally, in addiction to discoloration (which does not however have any effect on heating), I did not like the feeling that it could move to places that it could destroy the parts. Especially for a laptop that can move. So, I decided to try good thermal paste and try to get the best I can from them.

 

I buy the biggest Pro edition that is available for then re-use it in my newer computers (or just let it be in the old if I still have it). Over a long period (timespan) the additional expense over second-rate will be added over several months / years coming. See... Let’s say $100 over 36 months won’t be many $ over a month.

The more expensive is re-buy new ssd drives for every new computer you buy. Especially if you buy too small or add in bigger costly drives into the build directly from the notebook OEM.

 

But you've an advantage where your country actually listens to consumer rather than big companies. Then again, you always buy from trusted source.
I don't have any trusted online retailers that listens to you until you buy it and ignores you after you buy the product. I still remember 950 pro 256GB costing nearly 400$ here.

 

As you are discussing about NVME SSDs speeds, I have done a number of benchmarks using 1)default raid intel drivers, 2) default standard (Intel?) AHCI drivers and 3) samsung NVME (AHCI) V3 drivers. The results on screenshots below:

1) https://imgur.com/a/Eb5G0Gh (Default RAID mode and default Intel drivers)

2) https://imgur.com/a/bSK5tGx (Default standard Intel? AHCI NVME drivers)

3) https://imgur.com/a/meja3IT (Samsung NVME v3 drivers)

Waiting for comments. It looks like the only reason to have AHCI is the slighly better 4K speeds. I personally do not feel any noticeable difference despite the better 4K speeds under AHCI. My system came with PM981 from Dell and RAID enabled in BIOS.

 

I tried putting my Intel drive back under RAID mode; did seem a slight bit slower, like you saw, maybe a few % at most. No downside to AHCI mode either (I was just wondering if Intel's RST would somehow change performance once it had access to drive again over RAID, but it did not). So I went back to AHCI. The general consensus when I read forums/watched some video reviews is that RAID mode in BIOS should be no more than 1% slower than AHCI mode. Harder to update firmware for Samsung drives in RAID mode though too - have to boot from USB to do it (have a separate Windows/DOS/Linux boot USB to boot in AHCI mode for the update), or switch the main OS like you and I have done (which isn't that bad after all - no hiccups when I did it, am starting to wonder if I even really need to bother restarting in Safe Mode first).

Thanks for your history on the liquid metal too. I probably saw it when I took a deep dive in the history of this thread, but when you say "really good temps", what were your temps maxing out at when you hit the full power limit of 110W? Are we talking 90, 88, less? Thanks. I have both the graphite pads and liquid metal (Phobya brand - only one that I could get next day free shipping, Conductonaut would be a week out) coming in the mail tomorrow. Might try one or both this weekend. I feel like either my CPU just isn't a good one since I can't reliably undervolt past -125ish without eventual crash, or my heat sink just isn't mating up well even with my re-paste. My temps show it cools fast now though, so I am not sure. Maybe there is still a spot where the transfer is weak though due to gap.

I know it's a no-no, but I almost wonder if just getting a bit of liquid metal (or paste) on the graphite pads would work - get better thermal conduction from the surfaces into the pad, and the pad might keep the liquid in place while also ensuring there's no gap.