*OFFICIAL* Alienware 17 R5 Owner's Lounge

Shameful. Absolutely shameful.

My new R5 i9/1080 is hitting 100 degrees on core 0, 2, and 4 under light load - just downloading 30GB on WiFi. Obviously with that core pattern the heat sink is poorly mounted.

Crazy to see thermal throttling from just downloading over WiFi. I DO have a gigabit fiber connection but that is just a drop in the hat for a system like this. In no way should that tax the processor like this.

Cinebench score 1171, 131 FPS.


This was an exchange unit I was sent. Not sure how to proceed - if I should have Dell send a new heatsink, if I should repaste, or if I should reject this and wait another month for another replacement.


Sent from my iPhone using Tapatalk

 

This is not a bad score at all.

Are you within a replacement/refund window? If yes, take it, and the replacement may have been put better together at the factory. If not you may need to open it up and repaste it.

 

So, if I am not wrong you suggest not to fully tighten the screws? After having done repasting a number of times, on different R5s, I am yet to see what you are describing above. Can you post a pic to show what you mean? Maybe it is there and I cannot see.

And the new machine will have exactly the same issues. The only thing you can get more is a better UV if you are lucky.

 

I concur, I have not had my motherboard get bent to the degree that it "rocks" when placed back on the chassis for re-attachment, not in either of the two machines I've worked on. @doofus99 - I really don't know what sort of bad luck you're having on yours but it sounds like you're getting the absolute worst builds? I even paid attention on my last re-paste, and I really didn't notice the motherboard bending to accommodate the heatsink tightening, nor it bending back or rocking when screwing it back to the chassis.

@cruisin5268d - welcome to the club! I would first do some basic undervolting in ThrottleStop if you haven't already. See how much UV you can get to before it just crashes outright. If it crashes at 150 or below, probably not going to be a great CPU to OC past 4.3Ghz on six cores. But can't judge too much by that; just see what is the max clock you can get with undervolting to get an idea.

1171 CB out of the box sounds VERY promising though (as for how much undervolting and/or re-paste will add/help out). My first unit only got 800 out of the box! (then up to 1100ish with undervolt, then 1280-1320ish with liquid metal) My second one got 1100 (and now is always 1350+ with my lm and UV). But yeah, unless you re-paste and (I think necessary) switch to liquid metal, those fans will sure spool up no matter what you do. (mine would for just opening the browser, let alone watching a video; accessing SSD for copies would also get those fans going, but no longer, just did a 4 GB unzip from ones SSD to the other, no fans) With my re-paste with Phobya LM and the pads I replaced, I really need to be hammering the CPU or GPU to get them to come up. (I have to say again, as it stands now, my system is pretty damned delightful... 4.6 Ghz at -145.4mV has been stable for two days now of general, varied usage, from low CPU to gaming, no restarts... I just hope the liquid metal lasts for at least 6 months!)

Basically though I think you need to see the spread from your hottest core to the next one, how stable it is with undervolting in general, etc., to decide if this is the keeper and to go in on it for the re-paste, possible re-pad, etc.

 

Hmm hard to say, maybe they deem their battery control intelligent enough to deal with this. Also it was a bit buggy on the older Alienwares. Sometimes it worked, sometimes it didnt. Sometimes it even killed the battery calibration and it was hard to get it back charging to 100% again.In the end I wish they had the same system as the old Vaio's had. Options to restrict it to 50%, 70% and or 100%. Samsung also had this and it worked flawlessly in those machines. Hell the Battery of the Ativ Book 8 was incredible and constructed in such a way that it was rated foor 2000 discharges. Gave my old Ativ Book 8 too my mum after using it for 2 years. Its now 6 years old and only 10% battery wear and she uses it daily.

 

When assembled, the chassis straightens everything inside.

 

Well, it is really simple, around screws number 4 and 6 I think, the ones below the GPU. The fastening arms extend diagonally a distance from the GPU - as you press down on them to screw them in, the opposite happens, the mainboard pulls up. You can see this by simply letting the heatsink rest on the motherboard, what is the distance between the arms and the socket underneath? On my #2 mainboard it was quite the distance and required a good pressing down to meet the socket, and as it did it simply bent the motherboard up. In the beginning I was happy because I thought that the stiffer the arm is the more the pressure down on the GPU, but no, the heatsink is much tougher than the mainboard and it flexes it.

 

What?

Running 3.9GHz all cores aka equal clock speed as the $600 usd cheaper i7-8750H.

 

My heat sink arms are not that far away. I only bent the single CPU arm just a tiny bit (maybe half a mm actual end result at most?), but left the other stock. If you didn't bend your GPU arms on purpose, I wonder if someone else did (i.e., a refurb heat sink, or the entire unit is used/refurbished?). On mine, tightening the screws in order was very easy, it was never fighting to get the opposite one in or anything.

Hmm, you are running on the 8700K though. I wonder if the higher cache (it does have more cache I think right?) affects it, not just speed alone? Also I see your C0% are pretty low across the board. I am not at my computer right now but with all the background stuff mine has going on I am pretty sure all of my cores are 1-2%. I know you said you don't prep for the test in any other way (such as prioritization), but still, it's too bad there isn't a tool that takes into account background stuff. I know we have the Dell bloat too but I don't see that taking up any CPU in the background, it's mostly Tobii eye tracker (three or four processes, one of which is always at 2.5-4% in task manager, and the others are 0.8-2%), then I have SpyderPro running, three game clients, OneDrive, DropBox, etc. I know if I leave Edge open with just 6 tabs, it brings my score down 50 points, so I am sure background services are also affecting it for us AW users who aren't killing tasks to get a lower idle threshold. It is surprising to see 1300+ at 3.9Ghz, that is for sure.

 

If you have 1-2% load on cores while idle you need to clean up or find what eat clock cycles. Even with software, antivirus etc running in the background + surf on the web simultaneous you shouldn't see that much. Get rid of the bloat. Wasted clock cycles. Even worse if you run a lot on the battery.

 

Here is why I am NOT willing to open up this laptop to repaste it. As you can see in this chart there is a 800 MHz throttle period, and at the time the CPU was only at around 85C! Which means something else is causing the 800 MHz drop and in my experience it is the GPU VRMs. I have no way to improve on the VRMs cooling, so tomorrow I am going to call it in. It is very hard to reproduce though, you have to be in the game for a long time before it happens. For certain this did not happen when I first got the laptop - maybe some thermal paste / pad has already deteriorated ?

 

I think that is probably it. Like I said, when I re-pasted my first one, which ran hot, the thermal pads around the GPU were completely shot after the few weeks. They were a gooey, crumbly mess, like when you have packing tape that you pull off a box and just the thread in the tape remains. Now that I saw the second one with new thermal pads (having not run it much and it running better from the factory anyway), I can see how much those pads are supposed to look, and how much they can deteriorate. Not sure if they're using Arctic pads (same color) or a knock-off, but uggh, I wouldn't want to keep those pads in. Probably ok for a dual or quad core ultrabook that's rarely going to hit 85 or something. I know others say the pads can stay stock, but I don't like them.

I can't really think of any other reason why your VRMs are running so hot and causing such an issue.

 

Repad and undervolt the gpu. Pascal is ideally made for undervolting anyways imo. You can drop the clocks a little bit and lose a lot of watts with Pascal.

 

@propeldragon You mean underclock, not undervolt, right? I saw a way to overvolt, but not undervolt (and unfortunately not just enable the power limit/temp limit like in desktop nVidia cards). Underclocking will help with heat but undervolting would be more ideal... (hoping my searching was just bad and there is a way to do this, would be great!)

 

No, I mean undervolt. I use msi afterburner and adjust the curve.

 

Those 800mhz drops can happen even during idle when GPU is minimally used. I insist it is related to latest bios versions changes and not temps. I would recommend comparing your bios versions to others before looking at pads or something else. Dell has made many bad thermal behaviour changes in latest bios versions to avoid recall of R5s. And yes I believe those changes are actually causing more problems (e.g. frequency drops) rather than sorting out things

 

I did a test disabling all tray services - at my 4.6 settings I got 1450-1457 over thee runs, see screenshot. You can see my C0% is still high even at idle though. wsappx was using a few percent of CPU I noticed (windows update or something?). When I restarted again with Tobii re-enabled and all other tray stuff, score was around 1355. So disabling all tray tasks gives me a +100 CB score.

I also did a 3.9 Ghz max clock test when I had disabled stuff, and I got a 1240. So I am pretty sure whatever other services are keeping my CPU busy are costing me the last +100 score that you are getting (comparing 3.9 to 3.9). I can see why it didn't make sense from your side then. I need some stuff to run though, I can't just disable it all.

One other thing - I doubt this has anything to do with it, but you're on BIOS v1.5, right? Did you check and pick an OC from the CPU Performance mode in the BIOS? I still think if that's not chosen, it might use a different power limit for the system, including the VRMs. TS will overwrite the exact settings, but I wonder if having that enabled in BIOS during the boot does anything.​

Ok thanks for clarifying. I meant using the actual voltage slider. I've seen a mod to turn it on, but only to go up (on laptop GPUs), not down. I don't really feel like adjusting all those points on the curve and risking instability. Did you just bump them all down and leave it, or did you spend a while tweaking that?

 

I discovered yesterday that the bios on the R5 can be downgraded without issues. I went from 1.5.0 back to 1.0.6 and back to 1.5.0 Again via 1.1.6 and 1.2.0.

1.5.0 stil seems like the best bios up till now, in 1.0.6 and 1.1.6 i was not able to change my powerlimits, so i suspect something else is going on.

 

I have never tried undervolting the GPU but MSI afterburner has a strange/confusing way to UV. But looks like the only option

 

Of course you can downgrade and it is going to be successful. However, some firmware (I am not sure it is called like that) implemented changes are irreversible. Those seem to be the most important ones anyway. If you have used one of the last 3 bios versions you are stuck with those changes.