DivideOverflow's P170SM-AS / P177SM-AS Thread (CPU Temperature Tests and more!)

Edit: I've got the 173, not the 177. It is the same as the 170, I think Prostar just uses that designation to differentiate the "S" loadouts.

I've recently purchased a P173SM-AS from Pro-Star, which was a great experience. They were very responsive, helpful, and fast in shipping and delivery! I had my computer in about 9 days from when the order went in. So here are my specs:

17.3" 1920 x 1080 Full HD (16:9) LED Backlit Matte Type Display
i7-4810MQ Processor, 2.8 GHz (Max Turbo Frequency 3.8 GHz), 6MB Smart Cache
NVIDIA GeForce GTX 870M w/6GB GDDR5
16GB, PC3-12800/1600Mhz DDR3 - 2 x 8GB ($65 in Value)

Thermal Compound FREE! IC Diamond Thermal Compound on CPU + GPU ($35 in value)
Enhanced Cooling Upgrade Additional Copper Heatsink Apply to Stock Heatsink 1

mSATA SSD Drive 3.6mm Crucial MSATA M500 series SSD 120GB
(1TB) 7200RPM (Serial-ATA III 6GB/s) 1 $20.00 $20.00
6x Max. Blu-Ray Reader / 8x Max. DVD±R/2.4X Max. +DL Super-Multi Drive
Intel Dual Band Wireless-AC 7260 AC + Bluetooth Combo

So as you can see, I have the IC Diamond thermal compound, PLUS, I opted to add copper heatsinks to the stock system. As I have mentioned in other threads, I'm expecting to run software that will max out the CPU for long periods, and I wanted to ensure that I have the maximum clock frequency on the processor during that time.

With that said, I'm going to hit pretty hard on temperature results, because that it what I actually care about. I may go into more details on the overall system later, but for now, just know:

1.) The screen is awesome. I bought a ColorMunki Smile hardware calibrator for $75 on amazon, and it is really impressive now. The screen did have a blueish hue out of the box, but I didn't notice it until I saw the "before and after" comparison that the ColorMunki provided. I highly suggest picking up one of those calibration tools. The ColorMunki is nice because you can load it on as many machines as you want, and you can set up multiple monitors on the same machine. The entry level Spyder product doesn't do multiple monitors, and I think only lets you calibrate one machine (it registers the software).

2.) The keyboard is comfortable, quiet, and they keys feel good. The right shift key takes a little getting used to for me, as it is truncated due to the arrow keys. Other than that, I really don't have any complaints there.

3.) I like the trackpad on this laptop. I know others found faults with it, but honestly, it is no worse than any other laptop I have owned. I was even playing some Just Cause 2, and Skyrim with it, and I thought it did just fine. My only comment is that because it is flush with the case, sometimes you lose the edge of it if you are typing then go to reach back down to the pad. But after about an hour, I was already over it.

4.) Fan noise: I'll hit this up front because I know it is a common question. What fan noise? When not running programs that are really using the CPU, I don't think the fans are even on. And when they do come on, they ramp on nicely. Not only that, but even running Prime 95 and Intel XTU stress test, it REALLY isn't bad. I mean it. There is an option in the hotkey software to turn on the fans full blast for gaming, and when you do that, it is very noticeable, but still acceptable for gaming in my opinion. On its own, even when running Prime 95 or Intel XTU, the fans NEVER hit that maximum level. I was playing Skyrim on full graphics settings with that "uber fan" mode turned on, and using the built in speakers, and it was perfectly acceptable. Which I guess brings me to point #5.

5.) Speakers: Plenty loud and clear. I had heard complaints on these before too, so I made sure to test them out. They sound great to me. I don't have to blast them at maximum volume, even with the fan mode forced on high. Music sounded great, games sounded great. Do note, I installed the Creative Labs Sound Blaster X-Fi3 software that came with the machine. It provides nice customization options: You can kick up the bass, "surround" effects, use voice modulation/modification through the microphone, mess with the whole EAX Effects and EQ settings, and simulate positional audio for headphones. I have a logitech G430, which also comes with software that simulates Dolby 7.1 surround sound on headphones, and they both work very similarly. However, I think the headphone jack puts out a little better sound than going through the USB (which you have to do to use the Logitech software), so I'm more than happy with the X-Fi3 software.


On to the main event, CPU Heat!
Do remember: I have both the IC Diamond thermal compound and the additional heatsinks. I also performed all tests with the rear of the laptop raised off the desk. My room was 23.3 deg C (74 deg F) at the time of the tests.

Test #1 - Stock settings Prime95
With stock fan settings, the CPU temp got to a max of 80 C (averaged around 77-78 C). Processor frequency fluctuated between 2.89 - 2.99 GHz.
Max fan: Processor Frequency 2.99 GHz solid. Average CPU temp 70-72 C.


Test #2 - Stock settings Intel XTU. This was interesting, apparently when running Intel XTU, the processor seemed to stay at a higher average frequency.
With stock fan settings, CPU temp got to a max of 82 C (and averaged around 80 C), but fluctuated between 3.19 - 3.29 GHz average.
Max fan: Average CPU Temp dropped to around 75-75 C, and frequency fluctuated back and forth between 3.19 - 3.39 GHz continuously.. so I'll call that an average of 3.29 GHz.


Test #3 - Undervolt -50mV - Test under Intel XTU
3.39 GHz solid. No fluctuation. 82 C max CPU temperatures, with an average around 80 C.
Max fan: brought temps down to an average around 75 C, no change in Processor Frequency

Test #4 - Undervolt -100mV Intel XTU - failed. Kicked it back to -50mV
Was running solid at 3.5 GHz, until it crashed.

Test #5 Undervolt -50mV - Prime 95
2.99 - 3.09 GHz (mostly 2.99), 78 C Average. Max temp 80 C.
Max fan: 2.99 - 3.09 GHz (3.09 GHz). Avg temps down to about 73 C (fluctuated between 70-75 C).

Long story short: I'm sticking with a -50mV undervolt. I'm super happy with the temps. All of these tests were utilizing all 8 logical cores. I didn't have a chance to run single core tests, but maybe I'll circle back around to them. I'm just happy that my temps really don't hit past 80 C. I'll run some game tests later, but the quick runs I did in Skyrim and Just Cause 2 showed no real heat issues. I think the GPU was under 80 C the whole time.

 

Did you do any tests with a laptop cooler to see if it makes any difference?

 

Not yet. Was running out of time this morning.

Edit:

Stock fan settings:
Arma 3 with full graphics settings cranked up, CPU 71 C, GPU 78 C

Metro 2033, ultra high graphics settings (highest they go): CPU 64 C, GPU 84 C.

Seems to be running great! Metro 2033 was extremely fluid and ran perfect with the settings cranked all the way up. I'll check later and see if the notebook cooler lowers the GPU temp in Metro.

Just Cause 2 gets the CPU to 69 C and GPU to 80 C

 

I am interested to see if a laptop cooler does anything but so far all those temps seem pretty good on their own.

 

Tests with Coolermaster Notepal U3.
There are 3 fans on this cooler, and it lifts the laptop off the desk significantly to provide air. However, I had the laptop propped up about 3" in my initial tests, so this really will be a "do the fans help?" comparison. I positioned one fan directly under the CPU inlet, another under the vents for the heat pipe, and the third under the GPU.

Test #6 - Undervolt -50mV - Test under Intel XTU with laptop cooler
Frequency held at 3.39 GHz for the most part, average temperature was 78 C, with a peak of 81 C.
Looks like a 2 C improvement on the average.

Max fan: Average 72 C, max 76 C. Also looks like a 2-3C temperature reduction compared to the previous run without the laptop cooler.

Test #7 Undervolt -50mV - Metro 2033 with laptop cooler
CPU temp of 65 C, and GPU max of 76 C! It looks like the fan made a huge difference on the GPU.

I figured the CPU wouldn't have that much of a difference, since it just increases the turbo frequency to take advantage of the increased cooling. It does appear as though it is running about 2 deg C cooler on average during the stress tests. It only peaked at 81 C for a split second, and then jumped right back down.

 

So it looks like an 8C difference on the GPU pretty much nothing on the CPU which seems pretty normal.

I have been looking at the CM Storm SF-19 and the Notepal U3 that you have. Trying to decide if it is worth it to get one and which one to get. From what I have read the fans on the U3 have a fairly high failure rate and the ones on the SF-19 are standard fans that I can just replace. Given where the Sager laptops input vents are either one should be pretty much blowing straight into them.

 

I'm pretty sure the failure on the fans actually is a failure of the fan controller. The little capacitor or resistor or something is what goes bad. Other folks have hooked the fans up to another little controller and they have been fine.

 

How's the build quality on this? I'm leaning towards this vs the MSI gt70 but I really need it to hold up for 4+ years

 

So far so good on build quality. Everything looks and feels really nice. I can't see anything that will be an obvious trouble spot.

My brother owns a gt70, which seems decent except it already needs to go in for warranty work to fix the charging system. It won't charge after 1.5 years. The AC adapter burned out and took the charging circuit with it.

 

My NP7378 should arrived on Tuesday and I will probably be able to post temperature information for that on Friday.

 

I do wish that the NP8278 had an option for the 860M maxwell chip. It looks like the cooling would be better with that one and a bit more upgrade-ability. However, there is nothing any of us can do about that one.

 

I find that for testing stability, there's no substitute other than to run the system at the intended overclock for what you want it to do. Prime95 will only tell you so much. For example I was able to run -100mV undervolt on my 4900MQ and have it pass Prime95 for 30 minutes. When I proceeded to play Sleeping Dogs, I noticed some artifacts during particularly intense scenes. The system never crashed or gave any blue screens, just that there were artifacts in the game.

 

Good advice! That is why I've been testing games too.

It seems funny you were getting artifacts from a CPU under clock... I would think that would be more GPU related. Did you under clock the cache frequency too, or just the clock?

 

It was a -100mV undervolt, didn't underclock the CPU or the cache. I think what happened was the physics calculations got messed up during really intense fight scenes.

 

I meant undervolt for both, not under clock. Sorry, I misspoke.

I'll keep an eye out on mine, but haven't noticed anything on -50mV.

I don't know why I was saying under clock.. Lol. I think I was reading about a thousand threads on Intel XTU and people's results.

 

-50mV should be fine, the default voltage is too high for Haswell anyway. The reason some people can do -100mV and others -50mV is due to inherent varability between chips. My rule of thumb is to leave 20% tolerance, so if -100mV is right on the cusp of stability, then -80mV should be fine for 24/7 use.

 

Yeah, my test failed after about 5 min. -50mV has been 100% stable, even running intense games and overnight downloads.

 

What do you do to undervolt the CPU on these systems?

 

Intel XTU. It is super easy, and I found its stress test to work just as well as Prime95.

When running prime95, the CPU temps were actually a hair lower, but that was because the frequency would drop lower than it would in the XTU stress test. Not sure why. Maybe it was a more consistent load, and XTU varies more?

The dynamic CPU voltage offset is what you want to change. I'd start at -50mV stress test it, and then drop it by -10mV increments (stress test for each) until it fails. Then go back in and go back to the last stable voltage, and then add 10mV for a buffer. So if you got a blue screen at -90mV, I'd set it for -70mV to leave room for stressful programs.

 

Once I make the changes, verify everything etc how do I make them permanent?

Can I do something similar for the GPU?

In addition to running the stress tests I will also run a MATLAB simulation to verify the CPU is still giving acc

 

1. Download HWMonitor
2. Download and install Intel XTU
3. Run HWMonitor to watch your temperatures (it records the max values)
4. Open Intel XTU, Choose "Stress Test", check box for "CPU Stress Test", and run for at least 10 minutes. See what your temperatures are at in HWMonitor.
5. In Intel XTU, choose "Manual Tuning", "Core", and set your "Dynamic CPU Voltage Offset" to -50mV. Hit "Apply" which should now be yellow.
5. Go back to Stress Test and run for at least 15 minutes. If no problems, hit Save, and then it should be saved every time you restart your computer (even if you don't open the program).
6. Go test some games that are hard on your system, and make sure they don't have any problems.

If you want to keep going, follow the same steps, but increment the voltage offset by -5 or -10mV until you find the unstable range, and then once it crashes, restart your computer (it will default to the previous setting), and then bump it back up by +10mV.

This is obviously all at your own risk, but from what I have read, there is no risk of hardware damage by undervolting.

 

Fastest way is to go directly to -100mV and stress test. If you see instability, up to -80mV then test. If still unstable, up to -50mV and test. Then try -60mV and -70mV and see where the edge of stability is.

 

Thank you for the information. I have the software downloaded already and ready to be installed on the system once it arrives tomorrow although I won't do any tuning with it until this weekend.

 

Yes since every single chip is different, each one has to be iteratively tested.

 

Yeah, I'm getting some weird lighting artifacts in Just Cause 2. The red lights from police or military sirens look solid, and aren't drawing correctly. Tonight I'll check and see if the undervolt has anything to do with it, or if it is a driver issue. Skyrim and Arma 3 seem fine... So I'm guessing it is specific to just cause.

 

Have you tried other graphics drivers?

 

Not yet. I think they are updated to the latest through the Geforce Experience app. Also, it may just be that game... I was reading up on it after I posted, and lots of people have graphical bugs with NVIDIA cards in JC2.

 

Keep an eye out for beta or hot fix for that game then.

 

In case anyone is having the same problem, it is as bug in just cause 2 with Kepler chips. Just disable decals in graphics options in the game:
https://forums.geforce.com/default/topic/543925/just-cause-2-graphics-glitch-with-320-14-beta/

I'm probably the only person still playing this...lol. I picked it up at the holiday steam sale this year.

 

Cool, thanks for sharing.

 

I just ordered a NotePal U3 Plus for cooling. Overall the system is staying cool so far without any undervolting (have not had time to do that yet) while playing games and during normal usage the fan is normally off entirely.

However while running MATLAB simulations that use CPU and GPGPU the system temperature goes up to about 85C or so and starts to throttle on the CPU and GPU. I think a large part of the problem is the notebook is not getting enough air through the bottom and so just having it on a platform should help a lot.

 

Raise the back feet up a little, it helps the fans breathe.

 

Getting the cooler seemed to help. It raises the laptop up and lets it breathe easier. The temperatures did not change but instead of staying at 2.95 Ghz it now spends a lot more time at about 3.3 Ghz on a 4810MQ and this is with running 8 threads and the CPU usage at 100%.

I am not sure how much the fans are helping vs just raising it but overall the system is working better now.

 

NP8279 order placed myself just now, once in hand i will see if i can contribute any info, i went with the 860M on mine to keep temps down more, too bad it is Keplar, but what can you do!

 

The NP8279 has better cooling and the only reason I did not go with one of those was kepler vs maxwell. I know I will be doing a lot more GPGPU stuff and maxwell is more compatible with the new tesla cards and that makes it easier to do dev on the laptop. So far though the NP7378 is nearly silent and even when playing games the fans don't run much. It is only when doing GPUGPU + CPU stuff that the fans run at full power.

 

My temperature tests with and without the cooler are basically the same laptop position, the only difference is the cooler tests have fans running, and the "without cooler" tests the laptop is on the U3 stand without fans attached.

So I got a few degrees cooler on CPU with the cooler fans, and about 4-6 C cooler on the GPU.

It sounds like yours throttled boost more than mine though, so you are likely getting more of a CPU temp benefit than I am. My main benefit is keeping that Kepler cooler...haha

Edit: Oh, and an update. I took my laptop out of town with me, so it was the first time gaming on a flat desk without lifting the rear. The fans were definitely louder! I popped a notepad under each rear corner, and it was much better.

I highly recommend a nice laptop stand that lets it breathe. My notepal U3 does the trick even without the fans. I'm going to pack it in my backpack next time I go out of town.

With that said, it was gaming 100% fine flat on the desk. Skyrim was blazing along beautifully. I just noticed the fan was at a higher setting than it uses when on the notepal.

 

My comparison was the desktop flat on the desk and then putting it on a cooler so the comparison was pretty different. Even with the fans off the system was running better.

I was also pushing the GPU and CPU pretty hard. So far no game has come even close to some of the GPGPU tests I have ran.

 

If you're using the U3 for cooling, make sure to place the cooler fans directly underneath the laptop fans. I know this seems counter-intuitive, but the cooler fans help direct more air into the laptop fans, and my P370SM runs about 2C cooler with the cooler fans on full.

 

Why would it be counter intuitive to place them directly underneath? I made sure to line them up pretty precisely. Then again I graduate in 14 days as a chemical engineer and I did very well in heat transfer so maybe different things are obvious to me.

 

Counter intuitive because it apparently blocks the laptop fans even more. (but the blockage isn't a solid mass and can actually push air through actively)

 

I found it made little to difference having fans or not. The gain from a stand was the angle and the raising of the back.

 

Uh, it would only impede airflow if the fans are off. Obviously more cfms at the fan intake vents would be where the air needs to go. Where else would you put them? The rest of the locations are solid plastic.

 

If the machine is raised fans under the grill make no difference since the difference to air resistance would be minimal. Air is drawn in more from the sides of the machine.

 

I have found in all normal operations that only the raised part matters. Fans on or off do pretty much nothing. The fans on only actually mattered when I was doing MATLAB simulations. Even with CPU and GPU pretty much running 100% for a few hours at a time without the fans the system worked and the CPU clockspeed stayed around 2.9 Ghz the chip is rated at but with the fans on it spent most of its time around 3.3 Ghz instead and when running 8 threads that is a pretty huge gain.

 

The vents over the heatsinks. I actually found it more helpful to have the cooler fans blowing into heatsinks than the fans. My personal experience is that with this setup and cooler fans on full. the max temp reached after several hours of gaming is cut by about 2C. Not much of a difference I agree, but 2C is better than nothing.