Are CPU temperatures of 93-97 degrees Celsius for this NP8157 configuration?

Model: Sager NP8157/Clevo 650HS-G
Central Processing Unit: Intel Core i7 7820HK Quad-core 2.90GHz
Video Card: NVIDIA GEFORCE GTX 1070 8GB GDDR5
Memory: 16GB 2133MHz DDR4
1TB 7200RPM SEAGATE HDD

With this configuration, I am getting sustained temperatures of 93-97 degrees Celsius when using the following programs (generally run at 1080p and at the highest possible settings for games) listed below for any period over 15 minutes:

PLAYERUNKNOWN'S BATTLEGROUNDS
TITANFALL 2
DARK SOULS III
TOM CLANCY'S RAINBOW SIX SIEGE
INSURGENCY
KILLING FLOOR
NO MAN'S SKY
Or any game in my library, actually.

Tasks that also cause the CPU to reach such temperatures include:

Exporting footage through Adobe Premiere Pro using lossless codecs
Exporting footage through Adobe After Effects using lossless codecs

Strangely enough, benchmarks do not cause these temperatures. Both the latest version of Furmark and 3DMARK Firestrike do not cause the CPU to exceed 86 degrees Celcius, with an exception for Prime95 which returns similar temperatures of 90-94 degrees Celcius at its peak.
A repaste of the CPU has returned no improvement and neither has cleaning the system. I have left voltage as is because undervolting has returned roughly the same result at the expense of semi-frequent crashes.

 

That's not unusual, Furmark and Firestrike are primarily GPU tests and so don't stress the CPU out much.

What have you re-pasted with? For ease of use I'd suggest Thermal Grizzly Kryonaut or GELID GC Extreme. Particularly on the CPU, make sure you don't use too much as that'll actually make it worse.
Since there's no IHS with mobile chips it's unlikely to be the CPU at fault so it'll likely come down to pasting, heatsink pressure or possibly a failed heatpipe.

You should be able to tell easily if it's a pressure or heatpipe issue by feeling the temperature of the air existing the exhaust. If the air is cold then the heat is not transferring properly.

 

I agree that you need to use something extraordinary for thermal paste. Today's CPUs and GPUs run a little warmer than what most would consider ideal when all conditions are perfect and there is no such thing as going overkill on thermal management.

Nothing is as good as liquid metal. If that freaks you out too much, get some Phobya NanoGrease Extreme or Kryonaut. Anything else besides liquid metal or these two thermal pastes is kind of a waste of time and money. GELID does work good, but I have not found it to be as durable as Kryonaut. And, you're going to have to undervolt. If you leave the voltage on BIOS defaults your temperatures are never going to be good. If you found it was unstable, you probably were either too aggressive with the undervolt or maybe another setting wasn't right. Each CPU is different, so some will require more voltage than others.

Another thing about this that is often not understood is higher temperatures requires more voltage. What I am saying is that if undervolting is not working, you may have attacked the wrong problem first. So, if you get some good thermal paste on it and make sure the heat sink is actually making good contact (lack of which can be a huge problem) that same undervolt that was previously unstable may actually end up working out fine if the CPU is running 10°C to 20°C cooler. If it runs cool and still crashes, then undervolt a little less, test again, and keep making small decreases to the undervolt until it becomes stable.

 

I suggested Gelid mostly because it's cheap and east to remove for multiple repastings. If the heatsink is at fault OP might have to do that several times to see where the fault really lies.

93-97C is still too hot (assuming a 20-25C ambient) for a properly functioning setup under a regular gaming load. Given the CPU and GPU heatsinks are joined together you'd have to repaste both each time which could waste a lot of TIM.

Still don't understand who's bright idea it was to add in that joining heat-pipe though. It literally makes the entire cooling solution worse in the P650HS

 

I've just repasted with IC diamond, as it was already using that to begin with. Admittedly, the air exiting the rear exhaust seems to be cooler than the air exiting the left side, but not necessarily cool enough to warrant a failed heatpipe. If it is however, how would this be fixed?

 

The fix would be to replace the heat sink if one of the pipes is bad, but that is the least likely issue. The most likely issue is the heat sink may not be fitting correctly, which could also involve replacing the heat sink.

If you're still seeing asinine temps like before with fresh IC Diamond, you've got a problem with heat sink fit or a defective heat sink. If it is under warranty, I would contact the place you bought it about making things right, which would likely mean returning it to them and having them replace the heat sink with one that works right.

You should at least try a modest undervolt (say -100) to see if the needle even moves in the right direction on the temps. If it does not show signs of improvement, then I'd send it back to get fixed or ask them to send you a replacement heat sink in the mail.

Nothing wrong with GELID GC Extreme. Good product overall. It is very effective, just has not been as durable for me. But, it would certainly work fine for testing and it is really easy to wipe off. And, yeah... those temps are insane.

 

I may have been using a faulty program, but testing around a bit allowed me to safely undervolt by -135. At this setting, the temperatures rarely rise above 93 degrees celcius, however, it still seems a bit on the hot side. How significant of gain can be achieved by applying a different paste? I have also opened up a request for a replacement heatsink, but it is unlikely it is covered by my warranty.

 

That's high even without an undervolt on this system.

That's relative to the pastes and the application. You might see a 2 - 4 degrees Celsius difference from, say, Arctic Silver 5 v.s IC Diamond, as an example.

 

Then it must be the heatsink, then?

 

That or the chip. If your GPU is not running excessively hot, the board may need swapped out for another CPU. Has your seller said anything about your heatsink replacement inquiry?

 

No, It's a bit too recent to have been responded to yet. The GPU is fine, stays below 80 degrees celcius mostly.

 

By chance have you looked at the PCH temps? HWiNFO64 might show them. It does not on my Clevo P870DM3, but it does on my Tornado F5 and it did on the P870DM-G. Just curious if your PCH temperature are abnormally high like the CPU, or if only the CPU is crazy high.

For the purpose of testing, if you use FN+1 and force the fans to run max speed does the CPU still get that hot? When not forcing the fans to run full blast and the CPU is slow-cooking itself, do you hear the fans ramp up? At 90°C sustained more than 5 or 10 seconds, the CPU fan should be howling full blast at that point.

 

I think that the fans on automatic are capped at 80% in every circumstance. nevertheless, the difference in CPU temperatures between running on automatic and max speed are either nonexistent or negligible. With HWiNFO64, how does one check PCH temperatures?

 

If running the fans full blast does not change the temps it almost has to be either the heat sink is no good or the CPU/mobo has some kind of unusual problem. Defective heat sink is the most likely culprit.

Open the sensors part of the app. It will show it in there. AIDA64 will also show it. Below are examples. But, this is only if there is a sensor for it. My P870DM3 motherboard apparently has no PCH temperature sensor. The examples below are from my Tornado F5/EVOC 16L-G-1080.

 

There is no PCH sensor, sorry.

 

As expected, they've refused, stating that these temperatures are in fact "normal under load for this system".

 

If you're experiencing throttling, you could bring that up. 97 Celsius is not far from the tjunction and is potentially dialing back on the longevity/lifespan.

 

Is Intel turbo supposed to have this big of a thermal difference? It seems like turbo is nearly always active, and turning it off results in an average of 84 degrees celcius.

 

If you turn off Intel Turbo Boost your CPU will run at its non-Turbo speed. Naturally, it will be a lower temperature because the clock speeds and voltage (and performance) are substantially less. Turbo speeds will begin to kick in whenever the CPU is not idle.

 

But turbo seems to be active nearly all the time, even while idle. Idle temperatures are noticeably higher as well.

 

If you have c-states disabled that will make Turbo Boost active all of the time. That's how I run my systems, but not everyone likes that.

 

How does one go about disabling c states?

 

If c-states is disabled, it was something you did (whether by accident or on purpose) because the default is enabled.

If you have an unlocked BIOS it is a menu item. If not and you are using ThrottleStop, having the C1E box unchecked will do it. If c-states is not disabled and your CPU is running in Turbo Boost mode, when you are not using it that could mean there are processes running in the background that are loading the CPU. With the system idle, start Task Manager and look on the Processes and/or Performance tab and see what percentage of CPU is being used. It should be at or near 0% most of the time with nothing else running. If it is more than 1-2% you need to see what is running (look on the Processes tab) and consuming CPU cycles. It could be Windows Maintenance tasks. Malware will also cause this.

 

Skimmed. Who is your vendor? These temps are so far from normal, you should be experiencing somewhere around 30c less with battlegrounds using max fans on your CPU. I believe if no repaste helped at all that you have a bad heatsink. Your system is not functioning CLOSE to normal.

Sent from my OnePlus 1 using a coconut

 

Eurocom. MX5 R3

 

So having C1E unchecked will result in turbo being on all the time?

 

I have to agree the moron that came up with sharing heat pipes should be whipped until he/she cries for mercy


Sent from my iPhone using Tapatalk

 

@Eurocom Support something is most definitely wrong with his system, possibly his heatsink. Most of these systems should be in the 70c range at most with the scenarios he's listing, especially using max fans. Can you take hold of his case?

 

Usually it's better to turbo up and process data and then go back to sleep again rather than take longer to do it at a lower clock so you may see bursts of turbo near idle but it should not be consistent just on the desktop.

 

looks like they aren't responding.

 

Well, nothing more I can do, sadly.

Sent from my OnePlus 1 using a coconut

 

Well, thanks for your time, I'm sure I'll find a solution.