*** MSI 16L13 (Eurocom Tornado F5)/EVOC 16L-G-1080 15.6" Owner's Lounge ***

Did I install a faster fan or a Dyson? 5700RPM!

 

@alaskajoel What fans did you get? Did you need extra tools / help in order to install?

 

I forgot to mention it, but I did do the whole flashlight thing. All I could see the reflection of the flashlight.

I went ahead and removed the heatsinks so I could access the ribbon... plus, I wanted to check if I had used the right amount of paste (still too much - I'm getting there). In the process I popped off the wifi connectors - man, those make me nervous. I'm always afraid I'm going to jack them up. I'm posting here, though, so I'm guessing I didn't

The monitor ribbon didn't look right to me, so I pulled it out and inserted it carefully. The laptop screen is working again, so I'm hoping I've fixed the issue.

Thanks for your help - back to reading the last 110 pages!

 

I just purchased four of the Aavid PEAD1A238BH 120mm fans to use on my U3 modded cooler pad. I'll let you guys know when I receive them and test them out.

 

OK, I need two of those. One for CPU and one for GPU. If they are more powerful (and will fit) I may want five of them... three for the P870DM3.

 

Maybe in the next hour or two I will look into the 960 Pro firmware update.

See anything interesting here? (Don't ask for an ETA... just getting started... no idea yet.)

@Donald@HIDevolution

 

Where did you buy it?

 

Dyson.

OK. All done, bro.

 

200W of power draw for 4 fans is awesome! Did you order some bungee cords and hooks to hold everything on your table?

 

Just done another minor tweak to the U3 cooler and the results are quite impressive...



I used up some 15mm x 5mm draught excluder to form a seal around the bottom edge and the sides. This forces more air up through the bottom vents and allows air out the front and into the small vents. Idle temperatures dropped from 34C down to 30C but the most welcome decrease was to the M.2 drive, which dropped from 45C down to 40C (It's 22C in the kitchen at table level). Wacking the full fan button dropped temperatures down as low as 26C.

 

Nice job! So, that is a foam weatherstrip like what you would use for a door frame to seal it up?

 

That's the stuff!

 

Great idea...well done! Those are some sweet results.

 

Cheers! The only downside is it doesn't provide quite as much wrist cooling as before.

 

Hello guys,

I really appreciate if you could help me out with the following temperature issue:

Every time I play "The Witcher 3" or "Quake Champions", my GPU and CPU temperatures go way too high.

For instance, while playing "The Witcher 3" for only 5 minutes the GPU temperature reaches 88C (ultra settings / 60fps / fans @ max power ~4000rpm).
What I've noticed is the GPU Clock frequency has always been around 1,800mhz and 65-70% workload.
It doesn't look ok to me...





My idle temperatures are GPU~42C and CPU~45C (ambient 23C).
Both GPU and CPU are at stock speeds and CPU is also undervolted (1.075 static core voltage).

One workaround is lower game settings and downclock the GPU but I don't think this the right thing to do.

Please, any ideas of what is going on?

Thank you

 

Hey, thanks much for sharing this data. I'd call the change statistically insignificant, probably within the experimental variance.

 

Just ordered my BitsPower BP-OCCIHS-SL. Managed to get cheaper postage to the UK following a few emails to BitsPower, so the price is now quite reasonable. They are still on a 2-3 week back order though. That's something else to turn up in the future and surprise me as by the time it arrives I will have forgotten I had ordered it!

 

True, And I agree on on the result being unremarkable. I am glad it did not go down. I have never seen an SSD firmware update that enhanced performance and never would expect much unless it were fixing something that was broken and hindering performance.

 

Turn on hairworks and you should see your gpu usage go up to 99-100%. If you are playing in 1080p with it off, your gpu usage is normal. You should be able to peg the gpu by running it in 4k res as well to an external monitor.

Your idle temps are higher than mine, but sadly 87-89C is the norm for GPU temps in this laptop when you are taxing the gpu in a demanding game.

 

That was my thought, also. I'm not sure I'm going to risk the firmware update, since I'm not currently having any issues. If it ain't broke...

If your update had shown that they fixed a major performance flaw, that would have made a more compelling case for others to update.

 

Seeking some advice for a bit of further tuning...maybe:

I'm currently running very stably on 4.7 GHz CPU clock, static core voltage 1.150V. My cache voltage and clock are at defaults.

My only concern is that under really heavy load (such as I had last night when doing a compiler batch job on several hundred shaders), my CPU temps go up as high as 92C unless I manually max the fans.

I could probably bring my temp down a little by lowering voltage, but of course that's risking crashes, and I know it's an iterative process.

My questions are:

1. What specific CPU benchmark do you recommend as being the closest to definitive for "is my current profile stable?" I've heard people talk about AIDA64, Cinebench, and others, and there's also the "stress test" built into XTU.
2. Before I start monkeying with CPU voltage, should I consider changing cache or RAM settings instead?

I'm constraining the problem by saying the clock speed of the CPU is staying where it is -- if I have to remember to manually max the fans, so be it. I like the performance I'm getting now, and things are stable. But like most of you, I'm interested in tweaking that last little improvement. In this case, however, I'm looking for "slightly lower temps at same performance" rather than "slightly better performance at same temps".

 

Without getting into the tuning details, I'd just like to point out that batch job you ran sounds ideal for testing / checking the tuning you want to do, hang on to that.

That's what I do myself, have batch jobs for CPU / GPU work, and fire them off to test my tuning changes - it's better to use real world tests than synthetic benchmarks.

Using real work as the final test is what you want in the end, and it sounds like you just found it

 

That's a great suggestion, but the batch job in question is not easily repeatable. It's an automated job that triggers when I import a project into a new version of Unity game engine. I can get similar jobs to trigger, but not identical enough for repeatable testing.

That being said, I think you've got an interesting idea, and I'm going to investigate whether there's a way for me to tell Unity to do a "forced rebuild" of all its source code within an existing project. I do have a scripting interface, so maybe this is something I can exploit to force the rebuild.

 

OCCT run for a couple of hours has been the best (quickest) thing for rooting out unstable CPU overclocks on my desktop, but it does produce a LOT of heat, so not sure if that's gonna be viable in this laptop environment, you could try it.

 

@Mr. Fox I got my IC Diamond with the CMOS kit from HIDevolution, it was well packaged and arrived in less than 2 weeks via USPS.
Thanks @Ted@HIDevolution and @Donald@HIDevolution for your support.

Going to work on it sometime this week.

 

Please forgive my ignorance, but what's OCCT? That's a new one for me.

 

Have a nose here...
http://www.ocbase.com/index.php/software

 

An overly abusive component torture program that I never use for any reason whatsoever. Don't need it and it's extra silly. If you treated your children or pets the way this treats your systems you'd end up going to jail, LOL. If you want to check your overclock for stability, wPrime and Cinebench work as well as anything else. If you want to find out how hot your CPU can get with a workload that far exceeds just about anything you'd ever do on a normal daily basis, then you can run OCCT and let it fry. While you're at it, fire up Furmark and torch the GPU for good measure.

 

Hey gang, been a bit busy with traveling but just got an email back from Aavid regarding the stock fan.
---
Good afternoon,

I have received word back from the factory on these fans, as I said initially these are all custom fans that we make for the manufacturers of laptops and other devices that this type fan is used. These are not a standard item that we offer currently so the specifications are driven by the customers’ requirements not ours.

So in short, I cannot offer you a better fan as replacements, however I can offer to send a couple spares if that may help,

Thank you,
---
So the verdict is in, nothing they can do to help provide a more efficient fans. I know Mr. Fox is getting some dyson fans so I'll look forward to how that all works out.

In the meantime, please let me know if anyone would like some spare stock fans and I'll find out the price for everyone.

 

When I ordered mine (Skylake version of course) week ago, they actually offered to send me the Kaby Lake version as well, in exchange for doing a thermal tests on both and report back to them. I'd recommend asking them about it if you are interested and don't mind doing some comparative benchmarks.

 

Ha, true, I thought it might be too much for a laptop environment! I guess I just want to be doubly certain I've got a stable overclock on my desktop CPU by using that program - for reference a prior overclock of mine passed 8+ hours of Prime95v26.6 but failed after 18mins of OCCT. I definitely agree with the Furmark comment, NVidia artificially throttle their GPUs when Furmark.exe is detected anyway, it's an invalid test really, plus GPU stability is probably not as crucial as CPU stability - CPU instability could ultimately lead to OS corruption.

When I overclocked my i7-2920XM CPU in my Alienware laptop I didn't test it as tortuously as I have my desktop CPU - it just feels like excessive whipping when it's running at 90 odd degrees celsius on the brink of thermal shutdown! I managed 4Ghz all cores on that laptop CPU though, and put limiters on long & short term power so that it wouldn't temperature throttle under Prime95 loads (v26.6), from memory I used 72W short power draw and 65W long power draw limits in the BIOS.

 

My Aavid PEAD1A238BH 120mm fans came in and I'm going to be installing them soon. BTW, Aavid doesn't sell to the general public and I thought I would add that in case anyone else was interested in buying them.

 

Awesome, please let us know the results on the new fans! As for Aavid, they actually offered to send me spare stock fans (didnt even ask for them), so I'm not sure whats going on there. Sent them an email about the pricing, will let you know.

EDIT: Sorry Rage, unrelated post and got confused.

 

It's too much for an air cooled desktop as well. And, there is really no point in it. If your overclock is not stable, it's not going to pass Cinebench or wPrime 1024M without crashing. Anything longer than those tests is just like being mean and abusive for no valid reason. Prime95 is in the same category. These silly utilities are unnecessary and I think they are kind of like tossing a brick on the accelerator of your parked car to see if the engine throws a rod after it redlines for however long. And, pushing even a stock CPU to see how hot you can make it under sustained 100% load over a long period of time is also silly. If it is going to get too hot you will find out doing the things you do every day. If those tests are far more severe that what you do every day, then they are also misleading. If you can run benchmarks and play games, or render videos, without the system overheating, how hot it might get with OCCT or Prime95 doesn't matter.

 

Wprime 1024M a few rounds. And Notebookcheck.net have started with running Cinbench in a row. Should be more than enough. And @syscrusher want repeatable. I mean this should work for him. P95 - OCCT or Aida64 stress FPU only is not needed. If your OC is not stable in your ordinary workload... Give more volt or decrease clock speed !! Choose what suits.

 

These fans ARE definitely meant for servers, extremely powerful and yes they are loud. I buying a new fan controller with at least 50W per channel. If my DTR throttles with the use of these, I'm throwing it out the window...hahaha

 

Hey Rancid, thanks for replying!

When I run "The Witcher 3" using the laptop monitor, the max GPU temperature is 90C. Today I talked to Zoltan from HID and he said this temperature is normal for this notebook.
A few minutes ago I ran the game on my external monitor (also 1080p), which is the one I use for games, and the GPU reached 93C... then I closed the game because I was afraid of damaging the card...
Do you know if it is normal? What is the card max temp?

Thanks

 

Evening,

Your max temp is high. I had almost identical results with mine until I repasted the GPU. I used gelid xtreme gc thermal paste and my max temps dropped drematically. Witcher 3 in 1080p hairworks on and ultra preset your max temp will be 87-89c. Almost any modern game - Nier, Mass Effect, Battlefield 1, Overwatch will produce similar temps. I've read varying opinions on max safe temp, I personally wouldn't feel it is safe at 95C. Anything over 89C you will start to see throttling.

My advice - clean the gpu really well and do a repaste. Also look into undervolting the cpu which will help maintain overall temps over a longer period of time. Some games are both cpu and gpu intensive and as others have mentioned, the fans in this unit aren't strong enough.

 

This is my biggest concern, while I await my laptop to arrive. You mentioned you will regularly hit 87~89c while gaming, even after delid and re-pasting, which is too close to throttle temperature than I'd like.

Does anyone who plays games extensively on this laptop tell me what kind of operating temperature you are seeing and if you notice any throttling at all? I was playing some games on wife's 15" Asus laptop (7700HQ+GTX 1070) and that was hitting 90c+ and throttling left and right, so I'd really like to avoid similar situation with this laptop.

Am I getting ahead of myself? Will it require a cooling pad to ensure it will never reach 90c+ during gaming for more than an hour? I'm almost contemplating getting P870KM1 with single VC 1080 instead, if I cannot reliably play games without the risk of thermal throttling. I'd like to stick with 15.6" laptops for convenience while traveling, but I'm not sure if I'll end up regretting the decision down the road when it throttles due to high temperature.

 

I won't say I am the heaviest gamer on this thread, but I've tested at least 10 different games on the notebook thus far. Normal temps are within the range I mentioned on my unit, delid cpu/undervolt and the 87-89C range is with a 200/300 OC applied to the GPU. Surprisingly enough, overclocking doesn't add much to the temp and does increase performance by a noticeable amount. My unit no longer throttles after reapplying the thermal paste, but it can come close at times. If you look back at my threads when I got the unit I was fairly concerned about the temps as well, but they are what they are with this model and the weak internal fans. I have a 1080ti desktop watercooled with a 5ghz Kaby Lake processor. 4k monitor and I still use my laptop for almost all my gaming now because of how convenient this laptop is. If you do not restrict air flow, make sure you have a good GPU paste, undervolt the CPU, and keep the fans on max you will be fine.

I have tried 3 different notebook coolers and none seem to make any difference. I am going to try the notepal U3 with modified fans since they have a good reputation. The issue I think with notebook coolers and this model is there aren't a lot of intake holes on the bottom and they are mostly centered so turbulence probably plays into the reason why. Mr. Fox or others with the U3 should be able to give some before/after temps. My unit I see some slight CPU temp decreases, nothing on the GPU, and the additional airflow does help my M2 drives.

 

Hit up @Mr. Fox's bottom cover mod in this thread

 

Appreciate the reply. I'd like to avoid carrying around a cooling pad, especially something as beefy as modded U3, in order to keep the laptop from throttling. Perhaps I'm just being paranoid and modifications such as Mr. Fox's bottom cover mod and Bitpower IHS will be sufficient to keep the throttling at bay, but I must admit it worries me greatly. I would love to hear from others who plays games on it and if this is something I should even be concerned about.

 

Waiting on @Donald@HIDevolution to source some panels and I plan on modifying it. Thanks for the tip though, definitely good visibility for those new to the thread.

 

I don't completely agree, those tortuous tests are there to find times when the CPU can throw up errors - just because an overclocked CPU can pass something like Prime1024 on a 'borderline' stable overclock doesn't mean it won't throw errors (that may be undetectable at first) during normal usage. My understanding is that the idea of "tossing a brick on the accelerator" is to find those errors that would otherwise be undetectable until insidious issues start building up in the OS & data causing corruption & data loss. This approach to overclocking does mean that you obviously end up with a lower than otherwise overclock, but it's rock solid.

 

@Mr. Fox is correct, no longer than 15 minutes on the full on 100% load on CPU / GPU is enough to see if it's stable.

Any more than that is just abuse.

Will it crash after longer than 15 minutes, as the laptop builds up an unsustainable heat load? Maybe, but only in the 1/10 of 1% usage profiles is that going to matter.

So if you do run things at 100% GPU / CPU longer than 15 minutes as part of your normal usage, then sure do a real world test with that long running workload, but only with actual work - things that the laptop will actually be doing as part of it's day to day use.

Most people aren't going to get near that usage, and 15 minutes is enough to test an undervolt.

Now, you aren't done.

You need to test for stable usage at 0% usage, idle, where the balance of the stability testing is done.

You can watch video's if you like, or just let it sit idle.

If you come back and it's rebooted, reduce the undervolt by 5mV, and continue testing under idle.

That'll catch both ends of the spectrum without overburning your laptop for 8 frigging hours of 100% GPU / CPU testing... do it at idle 0% usage, it'll catch the same final instability

 

Still don't agree.

 

30 years of experience OC'ing says it works for me

 

Fine, not my viewpoint.

 

Burn'em up then, have fun fixing'em or explaining to people how it was the right idea to burn it till it's broke

 

That's why I said it might not be good to test in this manner and/or run such a high overclock in a notebook environment. I did say earlier I had that approach with 2920XM (in my laptop). You don't want to be stress testing for hours on end right on the thermal throttle point regardless.