AW 17R4 LM Repaste Results + Heaven Benchmarks (FHD & UW-QHD)

Unfortunately I don't have any saved results from out of the box as I did a fresh windows install whenever I did the repaste. Pushing OCCT on stock system, I would hit throttling within a few minutes of starting the test. I also had a spread of about 12 degrees between cores.

I found iunlock's guide on repasting and repadding and was able to follow it with success. I had never used LM before, but after watching a lot of youtube tutorials I decided to tackle it and I'm glad I did.

Here is my current setup:
i7-7820HK (4.2GHz; -0.070 Undervolt)
GTX 1080 (+150/500MHz Overclock)
32GB DDR4
2x Samsung 960 Evo 500gb in Raid 0
1TB HDD (Included with the laptop)
G-Sync QHD 120Hz Display
External Alienware AW3418DW Monitor when home
BIOS v1.2.4
Windows 10 - 1803 Update

Now for the results:

OCCT Run #1 - Fans were set to performance mode in the BIOS and undervolt was -0.08v in this one, but I ran into stability issues occasionally with OCCT. Sometimes it would stop the test saying "Error Detected on Core #", but I never saw any issues during gaming or normal use. . I changed it to -0.07v and had no issues.



OCCT Run #2 - Fans in normal mode, Undervolt @ -0.07v




For the Unigine Heaven 4.0 Benchmarks, I ran these settings with the only change between the two being the resolution:
API: DirextX 11
Quality: Ultra
Tesselation: Extreme
Stereo 3D: Disabled
Multi-Monitor: Disabled
Anti-Aliasing: x8

Unigine Heaven 4.0 (1920x1080) - I included the HWInfo64 screenshot. At the time of this screenshot I had been running Heaven tests for about 25mins straight. I haven't done a PCH mod yet, as you can tell by the temps, but it's next on my agenda.



Unigine Heaven 4.0 (3440x1440) - I didn't take a screenshot with HWInfo64 on this one, though this was ran during the 25mins that I mentioned earlier.




Here's a quick pick with the ultrawide (I'm in love with it). My fiance uses it with her XPS when I'm on the road and she hates to give it up when I get back. Note: I don't leave my laptop shut while gaming, it definitely gets warm. Temps are ok during normal use though.

 

3DMark Firestrike and cinebench R5 scores?

Be nice to see temps after 2hrs of a demanding game. Occt and unigine are not representational of hard use. Try BF1, PUBG and we can compare as I have same unit as you

 

Is Firestrike and Cinebench free? If so I can give them a shot. Here is a screenshot of HWInfo after a 6hr gaming session on Diablo 3 if that counts:



I did notice a few stutters in game but they weren't regular. Looking back at the clock speeds in HWInfo, it looks like it did throttle some, but the CPU temps look decent enough. I'm guessing it didn't like the PCH getting so warm and dialed it back a bit? Hopefully my small heatsinks for the PCH and SSD's will be here soon.

 

Yes firestrike and cinebench are free so try them.
Diablo 3 doesn't really count since it won't work the laptop enough. You need a modern AAA FPS game really. Yes PCH can often cause throttling but it's really hard to detect although in your case it's obvious given the temp and minimum clock frequencies.

See mine here: http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-106#post-10749291

85C at 42W and unable to sustain 4.2 is very under par for LM I'd suggest. Running mine solid at 4.3ghz and 52W in PUBG (very very demanding game) is 78C max and 66C average. Also your average CPU power consumption is only 20W vs my 40W. My fans are spinning a little less than yours as well. Looks like it's being throttled heavily?

That's quite a high GPU temp as well with a monster overclock probably adding 10C to GPU and PCH for very little gain <3% if any due to throttling on GPU (I have actually undervolted my GPU @ 0.937V vs your max of 1.063V and I still get only 72MHz less than you on average as well as better stability and lower temps). Undervolting your GPU is really important. You will get lower temps and better stability for CPU (small difference) , GPU (big) and PCH (big)

The stuttering is due to little or no contact between VRMs and heatsink for the GPU area. A more demanding game will generate much more stuttering which will quickly become unplayable. It will also get worse in time like a few months as well from my experience.

 

What resolution are you running? I assume if you're maintaining ~110fps in PUBG you're either on 1920x1080 and ultra settings, or 2560x1440 and high settings. I get almost exactly 110-115fps on full HD and ultra settings in there.

The 85C at 4.2ghz is because of my GPU being so hot more than likely. If you see on my OCCT test at much higher watts with 20 degree lower temperature.

As I said earlier in the first post and thread title, I am running a 3440x1440 monitor. That's why the last screenshot had temperatures on the higher side. And when you take in account that I'm running over a million pixels more than built-in display per frame, the temperatures really aren't that bad. I just need to get my PCH cooled down if I'm going to game on the ultrawide. When gaming at full HD or QHD, my results are about on par with yours.

I copied your V/F graph inside of MSI afterburner and this helped with the temps some. I just did a back to back comparison with 3440 vs 1920. Sitting in PUBG lobby for 30 minutes, both on ultra settings, I averaged 62fps with 3440, 165w of GPU power (183w Peak @ 0.975v/1900MHz) and temps hit a peak of 72C on the GPU and 76C on the CPU. With the same exact V/F curve in MSI, and the only difference being the full HD resolution I averaged 113fps with 1920, 139w of GPU power (163w Peak @ 0.975v/1900MHz) and temps hit a peak of 67C on the GPU and 72C on the CPU.

 

Thank you for linking that other thread. I pretty much copied your V/F graph, except my curve is a bit different, and it helped tremendously I believe. I started up Heaven 4.0 and OCCT together to try see if they both together would push the temps more than a game would. The GPU power during the stress test was about on par, however the CPU was averaging 55w which I never pull during gaming (I typically only pull 25-30w on Witcher 3).



Now I just need to try to do the PCH cooling mod and I think I'll be good to go for a while. Thanks for your help!

 

yo pete, i've found out that my cpu uses around 60/65 watts when gaming (LM, 4.2 Ghz max temp 82C, GTX 1080@ 180Watt on 2560x1440 DSR). Do you know a way to get my cpu temps down, without bringing the clocks back. i'm currently running 4.2 Ghz with a static of 1.15 V, so voltage cannot be the culprit. as soon as my GPU goes over the 190W+ my cpu rises towards the high 80's low ninety's but no thermal/power throttling. i've yet to mod my laptop with the SSD/PCH mod. i was thinking about putting a extra fan underneath for more airflow, but i cannot seem to find a power source (searching for a USB connection, or AC connection, where i can get power from) it would become something like a towerfan but then something that could be embedded within the chassis (choosing a towerfan over a normal fan blade, cause a tower fan blows air sideways. just what we need in our laptops (so i can create myself an low pressure and high pressure area under the keyboard.)

 

Hey sorry, just got round to this. I'm running QHD and have high settings with a couple on low / medium. (I researched values to change and some of them appear to do absolutely nothing but reduce FPS)

I doubt this very much, pascal GPUs do not get that hot, certainly way less hot than the CPU will under full load. 85C for the CPU at 4.2GHz is not normal. Whats the V supplied to each of these cores? This vital info needs to be in you HWINFO. Eitherway, the temps for your OCCT test at 4.2GHz seem normal

Ah, I see your point but I'm not sure it makes a difference. Those extra pixels do not change the fact that at 42W CPU usage your temp is 85C! Perhaps you can run a test for me? Run the same game (or better yet a AAA FPS title like BF1 at QHD, just leave it in first person view of spectator mode for the whole map then re-do it on new map) for at least an hour, then we can look at the power draw and also the core voltage consumption to be added to your hwinfo as well please. Your CPU usage in % would be useful to see as well, I bet you'll notice this is working no / very little harder on average between ultrawide and QHD. If you were running with V-Sync off and you weren't capped at a framerate then the difference between the screens is irrelevant, you'll just be pushing out less frames on the ultrawide vs the QHD, but your CPU and GPU will be working just as hard for both. FYI I have a lot of data for BF1 benchmarks so be interesting to compare

Cool, we could compare PUBG with a few games at QHD but actually, just do the BF1 test above, PUBG is not a good way to compare because the maps can end real fast and you'll spend most of the times in the menu.

 

Aha interesting! 81C at 60W is much more normal. I reckon this must be down to the GPU outputting less heat with my undervolt curve, good work, glad it helped!

 

Two things really since those temps look about normal for correctly applied LM and balanced heatsink:

1. Lower the ambient temp of air in the laptop
   1. Cooler room / Aircon in room
   2. Notebook cooler
   3. Extra fan in chassis
   4. Remove SSD drives blocking airflow and heating ambient air
   5. Improve PCH cooling
   6. Ensure fans are clear and free from dust and hair etc
2. If you're using throttlestop there are some small gains to be made by undervolting not only the CPU core but also the iGPU, CPU cache, system agent and analog I/O

What temp is your GPU and whats voltage / clockspeed on average after gaming for a while?

 

I think point 1.1, 1.2 and 2 are the ones i can tackle immediately. i'm going to see how much it improves by tackling 1.1 and 2 first. the Netherlands is now currently experiencing a mild heatwave so cooling my house down is going to be a problem but not impossible !

depends on which game i play, but mostly it now hovers around the 75/80 c (which is too high) and at first it was around the 65 max. 70 max. When benchmarking it can go up to 85 C peak which again is way too high. ambient temps are between the 24.5 and 26.5 c

i'm thinking about redoing my LM repaste (have LM'ed a few latitude laptops and they still work fine, and are not used stationary) i'm fairly confident with LM but before doing that i wanted to explore all my options and hear another notebook enthousiast his opinion.

i'll start with undervolting, and buy or build a notebook cooler.

 

Fair enough. FYI I regularly game at 4.3GHz with CPU temps in the high 70's, low 80's. It's not a problem at all for these chips so I'm curious as to why you think this is "way to high"?

 

Yeah 80~85c are perfectly acceptable temps and desktop GPU's run at the same temperatures in general apart from the aftermarket models with really big coolers. All FE boards are running at 80~85c as well and that for the GPU. CPU though depends the cooler but in general a CPU can handle more than a GPU can in terms of heat. The flip chip connections are more fragile on larger substrates and Intel has their manufacturing very well under control. I would be worried more if the GPU is too hot than a CPU in general. But 85c is fine in my opinion and still below stock Intel CPU's in a desktop with a stock cooler.

 

i don't like heat that's all lol. you and @rinneh are definitelly right, i just don't like how my fans ramp up to 4200 rpm. it makes a hell of a noise (not the fan itself, but the air that flows from the fans) most of the times it would run on 35 to 3700 rpm, but since the weather started to become hotter and hotter my temps started to creep slowy. it's not that it's bad (i can game for hours without overheating etc) it's just that the sound @4200 rpm is annoying.

before my temps started to creep i ran my chip daily @4.5 but now it's back @4.2. when doing cinebench @4.2 60 watts of heat equals to around 65/70 max temp. i can do a run if you want. it's just when my gpu heats up over the 180 watts everything rises likes the sun