XPS 15 9550 temperature observations (undervolt + repaste)

Thanks P. That guy measured the 9550's VRM with a handheld temp sensor showing 105*C on GPU and 91 on CPU

- First temp measurement of VRM by section I have seen; that indicates how much hotter GPU mosfets can get

- With the bottom cover on, you would expect temperatures could be higher; if he just ran the test with his laptop upside down and the bottom cover off, temps would be a lot higher as there is virtually no air circulating in the VRM area

- I assume he is measuring the hottest parts of the VRM (the mosfets, which are highlighted in red and mislabled VRM)

I would not recommend using putty as you really want the thermal pads to contact clean surfaces.

I tested thermal pads to ensure there were small rectangular indents from the heatsink. GoNz0 used paper to test for friction (and does not recommend VRAM pads unless you see a problem with heatsink contacting pads). He may be most expereienced guy here so do a search on his posts...

 

You can clean the surfaces. Many engineers use tolerance putty (or plastigage) to measure contact surfaces like that. It's a great idea. I did that stuff all the time as a mechanical engineer. The only bummer is with the VRAM it may be different on every system depending on how flat the heatsink is...

I recommend doing VRAM if you have poor contact (which you will likely have). It's true that VRAM doesn't get as hot, but why not sink them.

 

Yes I've seen these images. FLIR would be better, to see the worst offenders and their relative magnitude - with a heatgun a lot of manual documentation is needed. Also the VRMs under the black shield and various other chips around the board. The idea is to use least pads (leaving a bit of air) for best heat dispersion towards the cooler areas.

For VRAM pad thickness one could use paper or plastic foil I guess, but the issue is with the uneven surfaces.

 

Any issues with the solution suggested here?

 

I've been running with the chokes sinked for months without issue. I posted before and after graphs somewhere in this thread - it helped prevent my CPU from throttling.

 

A slightly dried plumbers putty or heck even a dry dough should be more than enough to accurately measure the thickness needed for all pads...

This is exciting! The bottom of the laptop is sure going to heat up more with this like gonz000000 has said...however, with proper airflow it can basically act as a horribly inefficient active/passive cooler depending on airflow...

One thing to note... I think there's a second section of VRM... not sure if its for the memory or CPU VRM but its near the right fan and they're covered...
Only checking temps will be able to tell if those extra 2 VRM are for the CPU or GPU...

 

Eason - I think custom90GT's trick was to use cheap, low W/mk thermal pads. I think couple of others did the same with good results.

If you can get the VRM very cool, you might consider jumping to XTU and looking at the turbo watt & timing boost levers. They worked on my 9550 and made a big difference for power limit throttling. Danger - monitor temps carefully. Use at your own risk. Also XTU seems to have a few bugs + conflict with ThrottleStop so you need to exit XTU after settings are made.

I tried with 17 W/mk thermal pads to the 6 chokes and the keyboard got uncomfortably hot very quickly. Thermals-benchmarks were much better but bottom case was so hot GoNz0 said it was superheating intake fan air so causing more issues than solving. A few others had the same issues.

The mosfets (mislabeled VRM in red in photo above) generate most of the heat and are likely most sensitive to heat so heatsinking them is really the better solution - but they are tough to reach. Ginglymus just did that a few pages back on a 9560.

Philaphlous' post today on p. 75 indicates (at least on a sample of 1!) that the GPU mosfets looks a lot hotter so if you have time to burn maybe try just heatsinking the GPU side. But, the sample is tiny and the CPU mosfets are hot too. It seems when the VRM area gets above about 80*C I run into some type of throttling

Finally see the earlier post by Brian Anderson in this thread. He used a $1 USB fan blowing on the f8 key with good documented results.

 

I've got a bunch of fans, all different types, and I'm planning on building a custom cooling pad. Basically my thought is I really just need 1 fan blowing a fair amount of air into the area right by the CPU and GPU. That way it'll cool the case and components inside... I would think with repasting, the blowers should do a good enough job to get that heat taken care of from what I've seen... The XPS isn't limited on throttling due to lack of conventional CPU/GPU cooling but more VRM cooling...as we've discussed in the past 5-10 pages here... Just too bad we can't link the cooler heatpipes with the VRM somehow...

 

You could, using thick thermal pads and perhaps some copper. But I think it wouldn't work well. That the heat from the CPU and the GPU might spill into the VRM area. And that it is better to offload VRM heat elsewhere, to case bottom, preferably wider.

 

Check this out:



I think this is overboard, and if it isn't sitting on an active cooler then I think it might heat up the chassis too much.

 

Unfortunately after spending many hours over the span of a few days experimenting with different thermal pad configurations and settings, I am unable to get through the ROG real bench test without throttling. Even with a maximum temperature (which is typically my CPU integrated graphics) of 77 C. The throttling does not occur while outside, despite the appearance of the temps being fairly similar, again concluding it is thermal related. Below I will include HWInfo graphs both inside and outside. I do want to mention in addition to my VRM pads which have been working great, i've also padded my heatsink to the case with the ultimate goal of trying to defeat real bench.

My thoughts/observations:

*Despite my low temperatures, I believe that the throttling might be appearing sooner with my chassis temperature rising. This may suggest there is another sensor measuring the bottom of the case, or case in general which is possibly related to the intel thermal framework? This is just speculation. Since my clean installation a few days ago the only software ive been using to monitor is HWInfo64.


1) and 2) below are without Voltage offsets using XTU (default 0.00) - but with the turboboost timer at 96 seconds and turbo boost power max up to 65 per Pressing's recommendations. Also make sure to compare the temps of 2) and 3) as they are pretty damn similar but with 3) throttling.

1) Inside: (XTU but no default voltage offsets)
http://imageshack.com/a/img923/5343/qwf08Y.png

Max Temps for:
CPU Core: 74 C
CPU Graphics: 84 C
VRM: 64
GPU: 67



2)Outside: (XTU but no default voltage offsets)
http://imagizer.imageshack.us/a/img923/3590/yPAYjT.png

Max Temps for:
CPU Core: 67
CPU Graphics: 76
VRM: 59
GPU:59


3) Inside: (XTU with -140mV voltage offsets - on a side note I believe this is causing my system to be come slightly unstable as I've locked up 1-2 times with these settings)
http://imageshack.com/a/img922/692/bbLBwv.png

Max Temps for:
CPU Core: 70
CPU Graphics: 77
VRM: 64
GPU: 66

Despite my temps being pretty damn low, #2 did not throttle, yet #3 did. I also want to note i've had my CPU temps up much much higher without throttling previously, so I wonder if temperature is being measured from the chassis somehow with the Intel Thermal Dynamic Framework.

 

You should be able to pass ROG stress test without throttling.

My 6300HQ throttled out of the box. It required a repaste, undervolt, "ultra performance" setting in dell command-power manager (and probably the THREE turbo tweaks in XTU I mentioned previously). And maybe a few other small tweaks. Now my XPS won't throttle on that ROG stress test (and ROG benchmark score is 10% higher).

The case bottom is a terrible heatsink. It failed with just pads from my VRM chokes. By also padding the heatsink to the case bottom, I think you are not giving the VRM heat anyplace to go and may be sending heat there and the motherboard. There are lots of heat sensors that are just waiting throttle your system. Remove the heatsink pads to the case bottom.

Your case bottom is superhot. After removing the heatsink pads from the case bottom, Is it comfortable to the touch for 30 seconds? are you still seeing throttling & instability at -140mv undervolt?

If you are seeing instability, try a less aggressive undervolt, like -130mv or less if necessary. That instability is telltale of an overly aggressive undervolt. You will see here, the Skylake processors can be undervolted by up to -170mv depending on lottery. KabyLake processors generally need to be undervolted more conservatively.

If you case bottom is still too hot, remove some of the VRM thermal pads. custom90GT got the right balance there

Make sure you are not running XTU & throttlestop at the same time.

 

Honestly ever since I padded my VRM to the case it has been great, I can now play every game I have tried without throttling (Overwatch, Warcraft, Doom). I also don't throttle with unigine valley or heaven. I just ran heaven for 20 minutes on 4K High settings with a max temp of 68 with no issues, and my case bottom doesnt really seem all that bad either with my cooler. However real bench on the other hand, is another story for some reason.

The copper heat sink to the case is just an experiment and so I'm all for taking that off, which Ill probably do in the next day or two. What is the other XTU tweak you are referring to besides turbo boost time and turbo boost power max?

Honestly I'm pretty content that I can run the games I will be playing on and off with no issues, however I was determined to have a successful run through a 15 minute real bench. I'm also using the 32mb RAM option since I have that much system ram and am not sure if that would have anything to do with it.

Thanks for the advice as always

 

Don't pad the heatsink to the case. It should cool the CPU and GPU (whose temperatures are not the worst offenders) a little more, but lead more heat to the overheated VRM area, rather than out through the heatsink fins and fan exhausts.

I don't think there is a sensor on the case bottom, but a hot case likely also means hot inside of the VRM area.

If you want to remove heat from the VRM area, put the laptop on a cooling stand, so that a fan blows into the VRM area from below. A big external heatsink below the rear of the laptop along the rubber foot might help too.

 

My biggest offender seems to be my CPU at the moment, and mainly in the case of real bench ( so far). my VRM temps haven't been an issue for me since padding. I don't think my VRM temp has even made it to 70 since that time, If the sensor is an accurate representation that is. My temps seem to be good pretty solid in general though.

 

Hi i did some testing with the following setup.
i7 9560
Thermal grizzly kryonaut repaste
arctic 1.5mm thermal pads explained in the picture
-120mv undervolt

Results and padding picture

It looks like i should be getting better temps but i'm not sure how i'm supposed to get 65° under load like the other guys in the thread.

 

You're not using a cooler, are you? I'm not a fan of any mod that requires more than the laptop itself.

 

I'm not using a cooler, what made you think i did?

 

I guess different sensors have different thresholds. I think CPU can run at 90 without throttling in Prime95 when the GPU is not loaded. But your Temp3 52 looks higher than what I was seeing, also Ambient (row 6 of Dell EC) 69

 

After seeing these results I think I'm going to go BTTW and repaste my CPU/GPU with liquid metal TIM... Definitely going to have to use a fair amount of nail polish to protect the surrounding components but it sounds like its going to be a must of I want temps I'm shooting for...

 

Kiss goodbye to your warranty then.

 

High-temperatures certified nail polish anyone?

 

I think you will get better performance and overall lower temps for the troublesome components with fewer thermal pads. The case bottom is a terrible heatsink so you want to use it sparingly.

- Remove the RAM thermal pads. The RAM runs cool. You are just taking heat from the case bottom and returning it to the inside.

- Try removing the thermal pads between the heatsink and the case bottom.

- Consider running the ROG RealBench stress test also as it provides a rigorous broader test.

 

Liquid metal has inherent dangers although can boost cooling significantly. That said, premium paste + a few well documented tweaks work fine on the XPS for most users.

Also, as GonZ0 notes, using fingernail polish can void your warranty. I think that is an insane proposition given the XPS' track record. If you are committed to liquid metal tim, read around the forums as some people have identified a decent electrical tape brand for this.

With the tweaking steps noted in this thread, my laptop runs cool and won't throttle in any normal usage scenario. I would try with premium paste and if that does not work out consider alternatives.

 

the very low temps!

 

I'll definitely start out with the Cooler Master nano paste that I have here first... still debating if its worth it with the liquid metal TIM...I could always use nail polish remover to get that stuff off...but I may have a much better and easier way to protect everything around the die's...

 

I did a 15 min rog stress test and it passed, these are the aida64 screens. It looks like it doesn't throttle but the temps still look quite high compared to others people tests.

Also after running unigine heaven for more than 6-7 mins gpu's memory clocks keep jumping up and down from stock 3500 to 2500, the core clocks remain the same. Any suggestion on how to solve that?

People in the thread talk about 65-70° load temperatures, mine are much higher.

 

Remember the Kaby Lake is an "overclocked" i7 so you will likely be running a bit hotter than 9550s quoted here.

Two testing goals were to run the Prime95 small ffts & ROG stress test (not together) with CPU clock running at max the whole way through and with both CPU & GPU well below 80*C. No thermal or PL throttling.

Typical ROG max stress test temps were 72*C and 76*C for CPU and GPU, respectively. Prime95 small ffts CPU was max 57*C . That was with Skylake i5.

Your aggressive Kaby Lake i7 & 1050 should handily outperform my i5. But I don't know if you can quite reach these temps.

I tried a bunch of small tweaks (like SpeedShift & boosting turbo watts & time via XTU) that helped. Read the thread for some ideas Just don't keep ThrottleStop & XTU open at the same time.

 

Alright i reduced the voltage to -144mv (it's the limit, less than that makes it unstable) and removed the thermal pads from ram and heatsink.

Removing the thermal pads didn't do anything so i guess they didn't do anything in the first place.
Reducing by another 0.24 the voltage helped a tiny bit.

This is a 15 min realbench stress run measured with aida.
(the green line in the temperatures tab is the gpu temp, the label is only shown when the dedicated gpu is being used)

What i don't understand is why the cpu is always at 100% usage during the test but at some point the temps drop because the clock drops even tho it doesn't look like it's throttling.

Still trying to figure out a way to avoid gpu memory throttling; temps never exceed 80° (gpus could safely stay at 90°+), why is it throttling? There must be some software issue.

 

I apologize g0dl3ss I am just briefly skimming the posts before class, and may not have read yours correctly. But to ensure you're not throttling download HWInfo64, run real bench and watch the PL1 parameter. If it ever drops below 45 which is the system default then you're power limiting likely due to thermals. Once it occurs it typically drops to either 15 or 7 W from my experience.

 

Follow Ginglymus' tips for looking at items like temps, speeds and PL indicators in something like HWiNFO64 during the stress test. It is boring but you see what is happening real time

 

When i said that less than -144mv was unstable i meant that... anything less than that number was unstable (eg. -150, -155 was not ok)

Thanks for the suggestions, i did another 15 min - up to 16gb ram - realbench run with hwinfo64 this time (still wondering why no one ever made a really complete and user friendly hardware monitoring software).

This is the result

Looking at PL indicators suggests that there's no cpu throttling but temps are still ****.

To sum up for whoever is looking for benchmarks the above realbench was run on an i7 9560, grizzly kryonaut repaste, arctic repad and vrm padding and -144mv throttlestop undervolt.

Gpu memory throttling in unigine heaven is another story and still a mistery.

 

I'm surprised your CPU temperatures are in the 90s on the RealBench stress test. I might be tempted to ultra clean the CPU & GPU and redo the thermal paste again. Interestingly, even GoNz0 will does a couple of pastes to estimate the min amount of paste for full coverge.

While you are at it, for the heaven debacle, double check every one of your VRAM pads are properly mating with the chips & heatsink...

FYI - my ROG BenchMark score is about 79,000 (~72,000 out of the box) on 6300HQ FHD. Your laptop should kill that score but run it just in case.

EDIT - my Heaven 4.0 scores
A. 2230
Preset: Basic
Render: Direct3D9
Mode: 1280x720 2xAA windowed

B. 1179
Render: Direct3D11
Mode: 1920x1080 fullscreen
Preset: Custom
Quality High
Tessellation: Disabled

All cases - GPU max 78*C, CPU max 64*C, no throttling (even running several times in a row).




EDIT - Should have asked if you are running a 4k screen as that matters!

 

Alright i did a third repaste, this time instead of using a drop of paste (also ran out of kryonaut and used the slighlty worse hydronaut) i manually distributed it on the cpu and gpu; i also squished the thermal pads under the heatsink to reduce the height a bit because i suspected that they prevented the gpu part of the heatsink to make good contact (now i suspect that one of the squished pads doesn't make good contact with the heatsink -_-)
This is the repaste

From a 15min realbench stress run i gained 9° on the average temp, now it looks in line with what people tried to get.
This is the hwinfo of realbench stress run ( old run to compare)

These are the realbench benchmark run results:

Spoiler

Image Editing: 136232
Time: 37.348
Encoding: 135651
Time: 70.681
OpenCL: 37693
KSamples/sec: 776
Heavy Multitasking: 120973
Time: 80.844
System Score: 100098

I tried running unigine heaven again (set it to basic): the score is 2013 and temps were between 75-78 all the time (cpu was at 55 most of the time) but in the last part of the run the memory clock shown in the top right corner keeps jumping from 3500 to 2500 and back up.

 

gained? or lowered 9c?

 

lol gained as in "positive increase". Technically the temps were lower by 9°.

 

According to that pic you put WAYYY tooo much paste on both the GPU and CPU. You should have either a TON of pump out or something is wrong with the cooler seat. Thermal paste for a laptop should be so minimal. Judging by the looks of it, your cooler might be bent slightly and thus its not making proper contact... I'll be able to tell once I get mine next week.

 

It may not look like that but the paste on the die is really thin.
What made you say that the heatsink looked bent since there's no heatsink in the picture?

edit: by the way the gpu memory throttled because i ran unigine in windowed mode.
I tried running it in fullscreen multiple times and temps were the same (75-78) but it did not throttle once.
It must be some kind of driver/software issue.

 

This is the best pic. This is from a member here on the 9550. The GPU heatsink is raised slightly...you can easily see it with the paste.

Spoiler

It also gives you a good idea of how little paste is needed for the CPU and GPU... if seated properly, you basically should be able to see the die and copper underneath the paste... The paste is more of a gap filler than a thin layer... That's why people recommend the dot or line method because it gives the paste a chance to squeeze out and fill the gaps...Only exception is a liquid metal paste which has to be applied to the entire die and heatsink... I'm really hoping my heatsink comes flat but it might be a challenge trying to get it perfectly flat...

This pic here shows I think the stock paste but see how thick it is compared to above? It formed a layer which is baddd! Especially on the GPU! I bet those temps were terrible!

Spoiler

Ah man! Finding all sorts of good pics! Thanks google! here's another bad paste job. Notice how much paste is left on the GPU...hence an uneven cooler...Still too much paste, you could probably cut the paste in half in this pic and still have enough....

Spoiler

I'm pretty skilled with thermal solutions. I'm super excited to get my hands on the 9560 next week. I've mostly had desktops but heck I was able to get my GTX 670 back in the day running on full load in BF3 maxing out at 35C! lol!


mmmmm! Thermal pads!

Spoiler

 

I'd spoiler those images.
As for the repasting it should be ok since i applied even less grease than on your first pic, the difference is that mine is evenly distributed and in that pic the paste is split between the heatsink and the die while on mine it's only on the die.

 

Gob-

Thanks for posting your detailed results - looks like you are making good progresss.
Dropping thermals by 9*C is a big deal and fixing the dislodged thermal pad might help with throttling.

Your realbench score is about 25% better than my 6300HQ's is and seems reasonable given Kaby Lake overclocking from factory + nvidia 1050 bump + i7 bump - heat penalty for all that performance in the same small XPS chassis.

However...your Heaven "basic" score of 2294 (posted below) should be killing mine (2230). I have taken some pictures of the Heaven scores with the description so you can compare apples to apples...

 

Thank you for the update.
I ran some other tests and by the scores it looks like my 1050 performs the same as your 960m (and it shouldn't).

High preset - 1920x1080
Score: 926
More details

Basic preset
Score: 2294
More details

Temps never exceed 79° and average at 74° (which by my experience with desktops is fresh for a gpu).

At this point i can only imagine there's some kind of power saving mod activated or it's a software issue.

edit: in this video the gpu temp doesn't go above 68° but that's a really low temp for a gpu working at 100%... need more 9560 detailed benchmarks

 

Are you running in high performance mode? Look at your CPU and GPU clocks and ensure they are pinned at turbo.

 

I believe I got a ~27xx on basic heaven with a max GPU temp of 71 C (my GPU memory has never down regulated as far as I am aware) and CPU max somewhere in the 70s. However, I throttle every attempt in real bench, even with the sensors reading a max of low 80s or occasionally high 70s. On average, I would say my CPU max in real bench is ~84-85. I'm going to repaste again soon and see if anything changes. Starting to run out of ideas =).

 

It took people here a while to tweak the 9550 to run flat out and only a few got them to run without throttling in more aggressive situations. Dell could have engineered better but the chassis is near its thermal limits.

Maybe the 1050 card and overclocked Kaby Bay CPU is just too much for the chassis.

We have only a couple of people struggling with basic 9560 stress tests but thereare plenty of tips and tricks in this thread which could help. And it seems like you are both getting close, so keep trying!

 

I'll get to the bottom of this once I get my XPS 15

 

@pressing
as requested!

Unigine Heaven Benchmark 4.0
FPS:
108.0
Score:
2720
Min FPS:
24.2
Max FPS:
191.3
System
Platform:
Windows NT 6.2 (build 9200) 64bit
CPU model:
Intel(R) Core(TM) i7-7700HQ CPU @ 2.80GHz (2807MHz) x4
GPU model:
Intel(R) HD Graphics 630 21.20.16.4574/NVIDIA GeForce GTX 1050 21.21.13.7306 (4095MB) x1
Settings
Render:
Direct3D9
Mode:
1280x720 2xAA windowed
Preset
Basic
Powered by UNIGINE Engine

 

Thanks for the favour GoNz0!

INTRO

Heaven is a CPU intensive benchmark. Preliminary reported results only. Not updated. Other benchmark results are posted by users here and in other posts.

Use caution and observe all temp sensors and other indicators to prevent damage

Note - no longer updating this consolidated post so updated-corrected details can be found in newer individual posts. Data and observations may be a useful starting point...
_______________________

Ok - now we have 6 Heaven "basic" scores for 9560 i7 users:

RockStar75
*2834
7700HQ
(Kryonaut repaste, -120mv undervolt. Reduced thermal pads to 6 chokes (GPU and CPU)
==> Reported no GPU throttling in Heaven (1680mhz flat throughout test)
Reported GPU at 70*C
Max CPU 81*C
Reported vrm sensor at 78*C
==> FYI - Throttling Watch Dogs 2 @ 15 minutes

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-82#post-10464427

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-82#post-10464578

Philaphlous
*2960
7300HQ (i5)
(CLP repaste, thermal pads on VRM mosfets, VRAM dense thermal paste)
==> No throttling in (ROG) real bench at all with CLP. I think I might have hit 75C but fans were lower RPM.
==> No throttling on Heaven benchmark with +260 on memory (assume based on score?)
- Can loop Valley benchmark without any throttling stable at 1730MHz on the GPU and I think +100 on the memory...I'll have to go back and check.

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-94#post-10470662

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-94#post-10470633

Eason
*2894
7700HQ
(undervolted -120, vrm pads, pads on ends of pipes; clamped 35W via ThrottleStop to prevent throttling Heaven)
Firestrike Max CPU 75C / GPU 76C
Uningine Heaven Max CPU 85C, Max GPU 78C.
==> throttling on the CPU after a few minutes of Unigine heaven (score = 2372)
==> Fresh unigine basic nets me 2894. Didn't TDP throttle (though I clamped at 35W with TS)

http://forum.notebookreview.com/thr...observations-undervolt-repaste.785963/page-96

GoNz0
*2800
7700HQ
(Gelid GC Extreme repaste, no undervolt, no repad, raised back of computer from desk to improve airflow)
==> Reported GPU throttling in Heaven (voltage limit at test start; temp limit 2/3rds of the way through test)
CPU around 80*C (3400-3500mhz)
Room temp 28*C
==> vrm sensor at 99*C

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-80#post-10463408

Ginglymus
* 2813
7700HQ
(artic silver 5 repaste, -144mv undervolt, large thermal pad on vrm,...)
==> Reported no GPU throttling in Heaven
Max GPU temp 70*C
Max CPU temp 72*C
vrm sensor <70*C
==> FYI - ROG RealBench StressTest 30 min - no throttling
==> Throttling - largely combated by disabling turbo, disabling hyperthreading, reducing CPU to 80% (via power management). Used Zalman fan cooler

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-85#post-10466965


g0dl3ss
* 2968 best
7700HQ
(repaste performed, -144mv undervolt, artic repad vram (0.5mm), 3mm pad between case bottom & heatsink @ GPU. This was to increase mating pressure; also uses case bottom as heatsink)
Max GPU temp 69*C (?)
==> No throttling in Heaven(?)
==> FYI - ROG RealBench StressTest throttling begins at 8 minutes

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-85#post-10466738

Preliminary thoughts:
==> Very small sample, but it seems like the 9560 should be able to hit ~2900+

==> GoNz0's 9560 scores near average with only repaste & raising back off desk for airflow

==> FYI General conclusion for 9550 was that case bottom was delicate heatsink with very little room to help before overheating. Hot case bottom could hurt active cooling system by heating fan intake air.

==> FYI Undervolting 9550 CPU seemed to provide on average 2x reduction in temps (~10*C) vs. repaste (~5*C). So should provide some additional headroom for CPU intensive tasks. Benefit for 9560 might be less due to CPU overclocking

==> Hot VRM mosfets may be limiting current to GPU. GonZ0 said his Heaven benchmark had voltage drops on the 1050. Without the voltage drops, I think his Heaven performance could improve (my words, not his!)

==> Heaven benchmarks on i5 and i7 don't appear significantly different

==> May want to reduce VRM thermal padding to bottomcase. Then retest. Delicate balance of taking just a little heat out of mosfets and not overheating case bottom...

** = current version of Aida64 vrm temp sensor mislabeled at DIMM. Should be obvious by unusually high dimm temperatures. GoNz0 reported and next version should be fixed (maybe versions dated after 14 feb)

 

Not quite stock, I already put Gelid GC Extreme on

 

Give me 2 weeks and I'll be able to post my results! The CLP is back on! Repasting my CPU and GPU once I get it with Coollaboratory Liquid Pro thermal paste...hehe New Vram pads, and VRM pads also... m.2 drive getting swapped out for a Sandisk X400 512GB drive also...Those will be my mods.