*OFFICIAL* Alienware 17 R5 Owner's Lounge

@Papusan I ran it once on my SM961 factory drive, max temp was 75 degrees. (it takes longer to run than Crysal so temp should be right) Why four runs? I am out of time for now on running any more tests but at least I have the screenshot of this one. What's different from using Crystal? Just the way it does the different block sizes in one run?

 

Plus I am not sure why Crystal got it to the 90s maximum, but only the 75 max for ATTO. Maybe Crystal does a bigger write without stopping?

 

HWInfo *MISREPORTS* SSD temps, the second temp is JUNK and you should have it removed from your screen, do not take HWInfo as gospel ...

 

What ssd brand ? Run same settings and same version as posted below.

Different benchmarks tax the ssd's different. The numbers is all over the places. I would google what scores you should have.

This is older gen Samsung NVMe drive - Samsung 950 Pro and clean numbers

Can so be but remember high ssd temperature will always reflects in the Bench numbers (high enough ssd temp and the ssd will throttle down speed). (high enough ssd temp and the ssd will throttle down speed).

 

Those benching numbers look like magic.

 

Haha, they do right. Max temp 43?!

@Papusan Is this on your P870 or a desktop? Any special cooling accommodation like heat sink on the chip? Do you happen to know what your max temp is running the default CrystalDiskMark (recent version)? (since that one seems to run hotter on mine)

I don't see how any laptop can keep the SSD that cool, unless HWInfo is misreporting. (aren't they always going to be without much airflow no matter what the chassis?) My run was a lot cooler for ATTO than Crystal though, but still, 43 maximum is is no more than a hot day outside in the desert.

I did see how mine jump around on those large block sizes. I didn't realize SSDs could throttle, if that's what it is. (I mean maybe at 110 degrees or something.)

I know HWInfo is inaccurate on some of the other temp sensors on the MB but I thought it was reliable for CPU cores and other discrete devices like the drives, but that was just my personal assumption.

Hmm just learned something new, if it's accurate, ATTO uses compressible data to mimic more real-life usage stuff (documents), CDM is incompressible - but the results are sort of the opposite, if ATTO is choking for me but CDM isn't (CDM should be the more difficult test and have lower results):
https://linustechtips.com/main/topi...-and-atto-benchmarks-report-different-values/

 

I know that one of my m17 r4s had SSD over heating issues to the point that it would cut off. First it would cause the system to stutter severally, then it would either blue screen or just power off. I had them send me an SSD, then sent it in where they replaced everything but the frame/monitor and it did again as soon as I received it. This was only on their SSD, the one I installed never once got too hot. I remember one SSD they provided was Samsung, but don't remember them all.

Sent from my Pixel 2 using Tapatalk

 

g*ddamnit that's a chilly SSD just the way it should be. mine can reach up to 60C and i was happy with that (in the r4 it would become 88c frequently). you're really inspiring me to sell this laptop and to buy an LGA

 

Can you clarify what machine yours is in, where it reaches up to 60? Your footer says R4, but then you say "in the R4" it would be up to 88. Up to 60 in a laptop is great though, I think!

I am still not clear why an LGA book would be better for SSD temps. If we're not working the CPU during these tests of the SSD, what difference is there, all laptops are very small spaces for the SSD to operate in, with little airflow. Are we saying the LGA chassis has more airflow to the SSD? The layout looked the same to me when I compared the photos I've seen of them opened up (the 870 and 775). If we're talking while CPU and GPU in operation, and those cool better, then that makes sense, but if just SSD is being tested, what's different?

 

I have them both. An r5 and a r4 with a broken motherboard which Dell failed to collect. Now I have 2 systems but planning on getting a new mobo on my R4 and make some profit.

The clevo chassis (or barebone) have more copper, airflow and space between the components. Where as the AW line is a thin wannabe macbook pro chassis which obviously heats up the components.

 

Papusan's house has low ambient temps I mean really low.

 

Just a new CB benchmark after re-paste (Dell original pads) with Gelid extreme for CPU and MX-4 for GPU. CPU package power reached 105W with a thermal paste! Maximum temps of 93C. Frequencies stayed at X46 (minimum cache ratio was set to 32 in TS FIVR). This seems to be a really good thermal paste. Initially it may show some peaks but after running a few CB benchmarks (get it hot), it seems to stabilise.

1519 with x46, -195.3mV UV, 32 set at minimum cache ratio, CB run at high priority. No tweaking at all. Just run the CB.

https://imgur.com/a/m35xY6u

 

nice score ! but your undervolt did not apply. TS was not turned on (TS tray icon is green) and in the TS main windows it says turn on. Can you do another run with your undervolt applied ? (and TS turned on and active)

 

Nice - I decided on Gelid Extreme also. Didn't get to do it this past night but hopefully in the next 24 hours.

 

The downside is that it may not stay stable for long.I had to re-paste. I need your comments on this when you try.

 

This is a software thing. When you press turn on it goes red in taskbar (meaning switched off). When you press turn off, it goes green (meaning turned on). I think the developer has to change this at some point because it causes confusion. And TS was on during the test. Otherwise, it would never be possible to reach 46x on all cores and at those temps (at 105W).

By the way, the test was without any software or processes tweaking at all, as you can see. All the programs were running.

 

I think you meant the way it works now is, red is active, green is inactive (but confusing, I agree). Perhaps the developer thought "red means working because red means stop... ThrottleStop!".

 

well it can.. if you lock your clockspeed in the bios.. and you still have OC controls which sits in the command center... if i turn off Throttlestop it will downclock due Dellienware boosting VID and the powerlimit of 110 watts. How do you explain the text "Turn on" then if it was turned on... i mean i've used Throttlestop since 8.48 and i've never had this problem so i'm kinda curious how you did it. BTW i'm currently also using 8.70. But you have a nice score

@unclewebb is this true ? cause i've never seen this explenation/problem

 

For the lounge's viewing pleasure - I have the combined results of my attempt to cool my two SSDs using Thermal Grizzly Minus 8 pads.

I really didn't expect much or any difference, except maybe a slight delay in the heat ramp up if the pads acted as a very temporary heat absorber/buffer, especially when I saw that there is a plastic insert on the inside of the bottom/rear panel. However, I tried anyway, and, the results show a large improvement!

What I did was to use two layers of the 1.5mm pads, stacked, with all protective film removed. The resulting 3 mm seemed perfect to bridge the gap from the SSDs to the height of the cover. I can't verify if it's touching the cover, but it did seem a little more difficult to press the retention pins in around the area the SSDs are located. (I wish I had just bought a 3 mm pad, would have been almost half the cost, oh well - it used almost a whole 100x100 pad to do this! Might run out of it if I need 1.5mm for the main CPU/GPU heat sink overhaul, but I have 0.5 and 1.0 on hand still.)

I did not remove the labels from the SSDs. (not sure if the pads' adhesiveness/moisture will affect the labels themselves down the road, if I want to sell them, etc.) Not sure if removing would help more.

Both before and after were run on top of a cheap laptop cooler pad, fan on, lowest position to the couch (yes it was all done on the couch). Room air was within a 5 degree zone between 73 and 78 degrees. No other forced air in the vicinity of the machine.

The actual Crystal Disk benchmark results didn't change much - some were lower, some were higher.
​

The max temps were reached quickly during the write tests (write tests run much hotter than the read tests). The factory SM961 reached its max early on in the first write test and held it all the way through, so I don't feel it would have gotten hotter with longer testing. The 970EVO maintained 2-3 degrees below the max pictured for most of its write tests; it only reached that maximum for part of only one of the four write tests. So, the 970EVO is still running cooler than the SM961, but the temperatures are MUCH better on both.

I could feel a bit of warmth on the bottom panel, but not hot. So, I don't think I am in any possibility of melting plastic in the inside trim area of the bottom panel. I am also guessing it's not so much radiating the heat through the bottom panel, but perhaps instead (or mostly) acting as an air cooled heat sink, bringing the heat up and out of the cutout where the SSDs sit, to the air above the black plastic plate that covers the motherboard. This is all conjecture though. But I doubt the pads have enough thermal capacity to soak up heat for more than a second before they would stop removing heat and the SSD would continue increasing its temps (that is, if the pads were not cooling themselves off somehow and/or passing the heat along to the bottom panel). So I think this will continue working for however long the drives are working.

The controller and chips on the SSDs are much smaller than the CPU or GPU are, much lower wattage too of course, so maybe it's not that surprising it was able to do something for it after all.

Anyway, enjoy, maybe it will be of interest to others.

 

From what I can see from your picture... signs of dying ssd. Same as in the previous AW models. How long have you had it? This is clearly a RMA project. Boiling ssd's won't have a long lifespan.

Crystal Diskmark run shorter aka lower temp. Many NVMe ssd's run into throttle territory once the temp pass 70C. It's normal. But as you can see above... Not nice seeing ssd's being killed short time after you get them.

THIS. Thinner notebook chassis means less cooling around the components. You can't put in an 18 core oc'd i9-7980Xe + 1080Ti in a 5-10L desktop chassis either.

18-21C ain't coold. Even if I live here north we don't have ice or snow in our houses The house walls is well insulated due the climate. I don't live in an igloo

Works both ways. Green or Red doesn't matter.

 

I experienced that CDM produced higher temperatures the other day, even though shorter. And per the link I shared, I think it's because it uses incompressible data, whereas ATTO uses compressible data for its tests. The SSD can compress ATTO's test data, so they don't choke as much, from what I read about it.

I almost forgot to post this one - while my thermal reduction did NOT improve my CDM scores, they DID fix my ATTO results! Look at that, attached. Also interesting - now, the max temperature is the SAME in my ATTO run as in CDM, whereas it was definitely more before I added the pads.

 

The nand chips doesn't need so much cooling. The problem is the controller who will throttle down the ssd speed. BTW your 4k Crystal Diskmark scores is too low for NVMe drives. I would Google what results you should have. Look after reviews.

 

@FuturisticModz Check out my original ATTO test, similar to yours on the big block ones - those should all be at the drive's maximum like Papusan's, but they weren't:
http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-241#post-10807520

But now that I added the pads, my results look much better!

@Papusan - thanks for suggesting ATTO. I thought it was an old tool no one used anymore, but the way it runs and displays its tests is very useful! Basically one should expect to see a nice exponential curve with them all maxing out down the page, very easy to spot issues like I was having and Futuristic also. There's so many tech/monitoring tools out there, hard to venture out of the comfort zone.

 

Yeah that's the toughest test on the drive - I think mine are like half of what some online reviews posted? But I am not optimizing either - not sure if they do, but just saying, I don't shut down other programs/services, the drive is 60% full, etc. Doesn't that affect it? I figured it's just par for the course.

You mean on both my drives right, the C: (OEM Samsung SM961) and the E: (my own installed 970 EVO)?

Or are you suggesting both have issues and should be RMA'ed?

 

Old doesn't mean it's useless
You can find more tools if you look into my informasjon profile page.

I wold be sure have big enough OP (overprovisioning). Just follow up ssd drive remaining life in Hwinfo. And yeah, both have low 4K speed.

 

You're saying manually overprovision more than the drive does by default, is that right? What do you set yours to?

I will research the 4K speeds. Is there anything common to cause that? I'd love to improve but I thought it is what it is. Time for some sleep now though.

I totally agree on the NAND chips not needing so much cooling as the controller chip. I didn't feel like extending the experiment though and doing just the controller chips first (which would have used about only 30% the material), then the full coverage test.

 

Of corse. I put +20%. For those with very very heavy use the ssd guru @tilleroftheearth say +30%.

 

@Papusan - Actually, I looked this up when I first got the drive in and first tested it. Isn't it because CDM version 6.x defaults to the Q1T1 test, whereas 5.x's default test on the bottom slot was 4K but not Q1T1? Or are they the same tests, just different descriptions? I read about it once but I forget now - are they both non-sequential?

I guess I'd chalked it up to being that some reviewers use CDM version 5.x. I found a review using 6.x and the numbers on that bottom test were very close to mine.

My main concern when I first put it in was using RAID vs AHCI, since I can't use the Samsung tool in AHCI mode, but since the speeds were close to optimal for the large sequential blocks, I thought RAID mode was ok.

Thanks again. If I am missing something on these drives, I am eager to figure it out what it is, but haven't found it in ten minutes of looking around. Can you confirm it's definitely not just a difference in CDM versions like I assumed it was?

 

I mean it's too low.

 

Ok, got it. Very useful graphic! Well I don't know what I can do, I mean, there's nothing I can change beyond what I did on the temps, right. And I have seen reviewers get similar scores to me, so theirs are low also. Is it common for drives to be coming bad from Samsung like that? Or maybe RAID vs AHCI controller mode does matter (and not having Samsung Magician installed)? I'll check it out more later. Thanks for the diagram again, first useful one I've seen in all my searches I've ever done on CDM testing.

 

Try with MSFT's stornvme or Samsung NVMe driver.

 

You realise the largest source of heat inside the laptop is the GTX 1080 at 180W. Then it is the VRMs situated below the GPU at over 100C and 4 of them. Then the CPU at 40W-100W. And so on. You cannot test the SSDs, or anything really, without the GTX 1080 at its max - only then you will know how much heat is being taken away from the inside of the case and how much heat causes temp rises everywhere, including the SSDs.

 

Fast SSDs (e.g. samsung PM drives) need third party heatsink. I am using heatsink (EKWB) on both of them. And such fast ssds can get really hot without playing games, that is just by installing or opening big apps. So, heatsink all the way. The ones I am using made a huge difference. I also had to do a small mod on my second heatsink because of limited space

 

Can you tell me what heatsinks are you using? I want to buy them for my R5.

Enviado desde mi Pixel 2 XL mediante Tapatalk

 

Check my previous message. It is EKWB ( EK-M.2 NVMe) heatsinks.

 

Try lower Cashe ratio one or two bins so temp can stay below 90C. Could be the score will increase for you. Over 90c and the Cpu can be flaky but without clock drops and you may risk lower score.

1 bin lower clock speed - 45x@1525cb

 

Yes but I want my drives to run cooler all the times I am not using the 1080 at all, which is fairly often. The SSD doesn't get used much during gaming, I don't think, does it? Or if it does then oh well, only so much I can do... but I think it's more random/sporadic usage in-game. I wanted it to behave better during sustained read/write like for video and multitrack audio. So to me, testing just the SSD heat during high IOPS was useful. It may not be useful at all to someone who only plays games on their machine, you're right.

 

@Papusan - I am curious, I was having a discussion with somebody and they mentioned that using static instead of adaptive voltage might save up to 0.1 V in overhead, and thus static might actually run cooler (with the undervolt applied to that). To be honest, I only tried it once early on then assumed it wasn't doing what I wanted (i.e., undervolting). Does it make sense to use static instead in some situations, or am I incorrect on the adaptive causing overhead (and adaptive is always preferred)?

 

Do you know or can you check what max temp you get while running the write tests on CDM in a 75 degree room on the factory drive (assuming you also have Samsung SM961 by now with your replacement)? I looked at those SSD coolers/sinks but I didn't think they would fit in that tight cavity. Doesn't that one need to go on both sides of the SSD? It says thermal pad must be installed on underside... wouldn't that wick heat into the middle of the computer where we don't want it to go?

I looked at getting those but said to myself, nah, no way those will fit in the laptop. It doesn't give the height in the Amazon listings unfortunately, but from what people said in reviews, seemed to be only desktops using them.

Anyway yeah I am curious if it does better, the same, or worse as mine. It needs airflow to cool it so with it sticking against the rear panel I wonder if it does get cooled.

Also a few pages back, you asked about a cooler that attaches to the output vents on the machine - you mean something like this, right?
https://smile.amazon.com/Temperature-Auto-Temp-Detection-2600-5000RPM-Nintendo/dp/B01NACVLWM/

Have you used those before? I didn't think that would really work. But it is similar to what I saw Linus do in one video with a very powerful blower fan. I'd be afraid it would ruin the built-in fans if they are operating and have an external suction forcing them to move faster than intended, but perhaps it sucks airs from the rest of the laptop, not just from the ports that lead from those two internal fans directly.

Seems kind of kludgy to attach something like that too... I guess if it really works and isn't too loud, it could be useful. It sure gets great reviews!

 

Read...

"The major caveat of Adaptive Mode is that the minimum possible voltage for a given ratio is pre-programmed into the CPU. If you happen to have a very good CPU that can run at a lower voltage than the minimum adaptive voltage for a given ratio, there are only two ways to lower the value. The first method is to apply an offset. That’s why there is the option to apply an offset when in Adaptive mode. The offset value is added or subtracted from the Additional Turbo Mode CPU Core Voltage box, and the total is displayed in the Total Adaptive Mode CPU Core Voltage pane. The side effect of applying an offset is that it affects the entire voltage stack – from idle to Turbo ratios, which can limit the usable offset voltage range. The second method is to use the CPU Load-line Calibration setting in the External DIGI+ Power Control section. Using a lower value will lead to more sag under load, resulting in a lower voltage. Again, the issue with this is that it will affect how much voltage the CPU receives under all loading conditions, which can lead to instability when it is too low for a given load state, or when the CPU transitions from idle to load state."

Even with those caveats, we still recommend using Adaptive Mode for all normal overclocking, unless your processor can run at voltage levels that fall substantially below the minimum adaptive voltage for the applied CPU core ratio.

 

Thanks, good call... yeah I gotta try that. I was going to check firmware too, but since I am in raid mode from the factory, can't use Magician and will need to do a USB boot to get in via AHCI. I'll try changing the driver first though, I don't think the firmware can make a huge difference unless there was a known issue. I considered switching to AHCI originally too but doesn't seem from others' testing that it is different from RAID mode.

Oh, other thing I need to try is keeping the CPU clocks up during the SSD test, that could affect the benchmark too.

 

This temperature issue appears to be unique to the R4/R5. I have a PM in my R3 and the highest temp I could get is 70 degrees. Oddly enough that is the top end of its operating temp range. Each PCIe position came with a thermal pad the length of the drive.

 

Excellent info and refresher (on some of it, lol - a lot of this is new or stuff I haven't done in 6 years or more). Thank you sir!

 

Forget those exhaust fans - I want to see someone try out immersion cooling of an Alienware laptop. Submerge the system board in a mineral oil tank with a circulatory system to keep oil moving across the logic board.

 

Well, I got mixed results; luckily the process of switching to AHCI was simple - set reboot to be in safe mode, switch in BIOS, boot in SAFE mode, unset the next reboot for safe mode, and restart.

I did a test right then and there, the speeds were markedly worse (default driver installed), nothing unexpected there.

I installed the Samsung driver, and I got the best score ever on this drive for 4K - 50.1 read, 112 write.

Then (and I knew this might happen) I updated the firmware, and speed went down to 48, 105. I set up over-provisioning to 10% (I think drive has 7% already that you can't touch), did a TRIM. Speed was about the same. Then suddenly, few tests later, speeds on the 4K dropped back down to where they were on the Intel driver - 37 and 102, for read and write respectively.

Hmm, now I've closed Magician in case it was doing anything to slow it down, and I am getting 46 read and 95 write.

Considering the write is boosted by cache anyway, I don't think there's much else I can do. Plus when I look at this site, many people seem to be getting similar scores:

http://ssd.userbenchmark.com/SpeedTest/501128/Samsung-SSD-970-EVO-2TB

I did make sure to keep the CPU from downclocking (ThrottleStop open), and it helped a tiny bit, 1 or 2 MB/s. (could be a normal deviation too though).

Hmm, so in the end, I guess switching to AHCI was more painless than I thought, and at least now I can very easily manage the drive for firmware updates. It is kind of depressing though - of course it has the 5 year warranty and very long lifetime of data written, but for those small 4k random read/write events, it's not that much faster than a basic Sata SSD:

http://ssd.userbenchmark.com/Compare/Samsung-860-Evo-2TB-vs-Crucial-MX500-2TB/m430706vsm421719

(well it is significantly faster on the read, about 50% - just not twice as fast or anything, and similar on the write, although I am guessing the SATA drives will slow down after not too much data passes through the buffer)

 

Dell in a nutshell... ssd's, drivers and firmware is the second lottery... It's a huge mess. And why touch Magician? It's bloat.

 

With my Heatsinks, I rarely get over 60s even under gaming. Maybe 70s when benchmarking (which is what I used to check the heatsink performance). Thats on a 256GB PM981 and a 512MB PM961


Under normal use, they are at 40s even 30s

 

Never get Samsung EVO NVMe on AW/Dell they won't perform correctly or buggy most of the times. On Dell/AW use Pro series or WD Black or Micron based NVMe drives. Use PM or SM series for best results. 960/970 evo is a mixed bag on Dell/AW or on most BGAs. On desktops EVO work fine.

 

Not to mention the Pro is a superior drive anyways.

 

I use an 850 EVO 1TB only for storage space replacing the HDD, And it does not feel laggy at all. The opposite. For OS and games disk, I use 256GB PM981(came from Dell) and 512GB PM961 accordingly as stated before.

Here are all my SSDs benchmarks with their temps. In fact temps have been at 52C max when benchmarking. So cool! Much less than I could remember.

https://imgur.com/a/qn5WsH9