*** The Official MSI GT75 Owners and Discussions Lounge ***

I still don't know why your PCH is getting that hot.
I again decided to run atto disk benchmark in a 30C apartment with auto fans, and the 960 pro got up to 75C, but the PCH did not exceed 54C. I really wish i knew what was going on.
There was a GT75 disassembly video somewhere; I'll try to look at it. I can only guess that the PCH is doing extra work and on the GT73VR and GT75VR, it just stays frosty ?

And CPU getting to 82C and GPU to 75C on auto fans (without custom fan curves) is normal, @heliada Even with liquid metal. Mine does the exact same thing.
That's because the GPU fan first speed ramp is at 65C (like 2000 RPM), and the next speed ramp is at 76C (like 3500 RPM), where it actually starts blowing air. And the CPU ramps up first at 70C then at 81C (and only at medium speed). It doesn't ramp up to actually cool properly until 90C! That's why you need custom fan curves. Use MSI Silent Option.

Note; on MSI silent option, the slider goes from 0 to 150%. But on these laptops, 95% is maximum fan speed so there's no point in 100-150%. At least thats how it was when using the "skylake' Silent Option on the GT73VR/GT75VR. It's probably similar on yours.

 

The cores that are hot are ALWAYS the ones closest to the VRM's because the pads elevate the heatsink more and cause less pressure. That's why almost every single GT73VR, GT75VR and newer Titans from literally everyone's results always see cores 0 and 2 hotter than cores 1 and 3, and usually core 2 is the hottest core (top right position when disassembled on 4 core processors). You see the exact same thing with Alienwares. Always cores 0 and 2. Tilting won't do a thing. And if you use a runny thermal paste (like Kryonaut), you will have eventual pumpout/dryout on those 2 cores because the pressure will be less than the other cores. And a 6 core laptop is going to fare even worse.

Even sanding the heatsink won't fix THAT problem. Yes you have to sand to fix 'convex' issues, but the very first thing users need to do is to reduce the thermal pad thicknesss on the VRM's. That will prevent that reduction in pressure and help more. Removing the 1mm pads, using Arctic 0.5mm pads and compressing them with your fingers and then trimming them will fix that problem. Keep in mind that the MSI 16L13 RAN just fine without any thermal pads on the CPU VRM's (the CPU heatsink in that system actually cooled the GPU VRM's (yuck)). I would personally not remove the pads completely (if you did for the love of god, use a very thick thermal paste like Phobya Nanogrease extreme or Coolermaster Mastergel Maker nano), but 0.5mm Arctic pads, and then compressing them a bit manually even more, will work.

So use Coolermaster mastergel Maker nano thermal paste for repaste, and 0.5mm compressed Arctic thermal pads on CPU VRM's and chokes.

 

Same as the Alienwares. Some PCH chips just run Hot. You know... Bin quality.

 

Thanks for the quick response. That's what I kind of expected with regards to needing to remove the heatsinks for motherboard removal, but I was secretly hoping that wouldn't be the case.

I've already repasted with quality paste, presently I am using Kryonaut on my CPU and Phobya Nanogrease Extreme on my GPU (I am using the Phobya paste on the GPU because I ran out of Kryonaut after a few repaste attemps, damn tiny tube!). I am very happy with the temperatures and fan speeds I am seeing presently while gaming, hence my lack of enthusiasm with regards to disassembling things again. Combined with undervolts on the CPU and GPU I see GPU temps in the 70-75 degree range while gaming and fan speed around 3000 RPM, which I find to be quite reasonable. The CPU runs nice and cool as well, and it's fan never seems to go over 2000 RPM in gaming use. The GPU paste and pad job out of the factory was pretty poor, as the fan would spin up to astronomical speeds (sounded like cooler boost was likely kicking in), and even then the GPU temperature was shooting up to 85+ degrees.

I did already replace the CPU chokes and VRM's with Gelid 0.5mm pads as they were indeed damaged the first time I removed the cpu heatsink. I actually also found that some of the GPU components were actually missing thermal pads altogether, so I installed pads anywhere that it appeared they were missing.

For gaming performance do you think it's actually noticeable to add a second RAM stick to enable dual channel?

 

For gaming, dual channel benefits are negligible at best. You get more benefit from dual channel in things like video editing and encoding.

Edit: Phoenix's videos show that my information was outdated.

Edit 2: I ordered those 0.5mm thermal pads and some extra memory since I will have to dissect the laptop anyway. I didn't see an "MSI Silent" option in Dragon Center so I changed the CPU and GPU fan curves to 30/40/50/65/75/85 to try and keep the PCH from burning out.

 

I did some quick googling and came to the same conclusion. I'm not sure I'll bother with adding another stick as all I'll be using the laptop for is gaming. It came with 16GB out of the box, which seems to be plenty for my purposes.

 

Cant answer since I don't use single channel. Sorry.
Also the PCH is behind the motherboard, not in front, so it requires full disassembly.

 

Holy smokes, the second set of videos, Watch Dogs 2 in particular, makes the difference seem very significant! I guess I have my answer!

 

The thing is I never saw any post on the GT73VR thread where anyone ever complained about high PCH temperatures, and I went through all the posts.
So unless Azor gave MSI bad PCH's to save money, it's possible either:

1) there is more airflow on the GT73VR backside than the GT75 Titan
2) The PCH has to work harder because of all the new components. I mean a user had 80C PCH temps just gaming on the GT75 Titan. But if I can put 230W through a 1070 and not get higher than 55C on the PCH, that isn't what is causing it.

We need a larger sample size of people to test PCH temps.
@Phoenix @heliada @Pedro69 What do your PCH temps get to after doing a "4 queue depth 64 MB size" SSD benchmark test in ATTO benchmark?

(Atto disk benchmark is free).
Use HWinfo64 to look for the PCH temperatures.

 

Bin quality if there is only a few with boiling PCH chips . @Phoenix post your PCH temp results and firmware version.

 

Yep. So let's see if brother Phoenix has a boiling chip. Because he's good at managing his computer. If he has boiling temps, then I think we should make a new company called "AliewareSIthrottelvomitp-p-p-powerbooks".

Ok, that was sad. Where's that new vomit mannequin....

 

Tested using ATTO Disk Benchmark v4.00.0f2

Ambient Room Temp = 24C

Fans = Auto

Remember, I have the Fujipoly thermal pads and the bottom panel mod:



Result:

 

With the bottom cover mod it will normally be a bit higher PCH temp as long you don't use a notebook cooler.

Your results should show a bit higher PCH temp than same models with stock bottom cover (as long you didn't run your test with a notebook cooler). All with same or higher temp than this from same model have bad binned PCH chips. With same tests and ambient temp.

 

I haven't a cooler

so are these temps good or bad? I have not the abiliy to understand your post sometimes

 

+-84C is lets say more than enough But you use the bottom cover mod and without a notebook cooler. Aka your PCH temp should be a notch higher vs. others without the mod. If others without the mod have higher temp... See my previous post

 

Why would the mod make them hotter? I thought the bottom panel mod with more air flow would make the temps lower?

 

Bottom panel mod won't affect this model for PCH temps. The PCH is on the backside of the motherboard. There is no airflow there anyway. Some models have PCH in the frontside, but this one has it behind the motherboard (e.g. Keyboard side)
I think all models of this GT75 series have burning PCH temps then. GT73VR PCH temp never exceeded 56C, ever, and its "physically" in the same location as the GT75 from the disassembly video.
Funny that no review of GT75 Titan ever mentioned 90C PCH temps....

 

Completely off-topic, but.... would a 330w power supply work for the i9-8950HK model instead of the two 230w power bricks that come default?

 

Pure logic bruh. With the bottom mod the fans will we get more of the air from the modded bootom holes and not from the inside of chassis. Many notebooks is designed this way... The fans will get some of the air from top of the fans (It will drag/pull air from between keyboard and MB. Aka there will be more circulation of air around the chips.

Normally always some air circulation bethween keyboard and MB. Its designed this way on most models with hardware on other side of MB. Ram etc.

Msi could also added higher/screwed up PCH Core Voltage for the Coffee lake models . Everything is possible with bad firmware/software engineers. My best bet is still binning quality of the chips.

 

Probably because I am one of a small number with that problem. Everyone else seems to be in the low 80s. I placed a support ticket in with MSI about the issue.

 

I would not recommend it. It is very likely that you will try and push more power through the 330W brick than it can safely handle. There is a solid but expensive 780W solution if you want a single power brick.

 

@Phoenix , with that mod from HID you will be get better temps in cpu/gpu but worst temperatures in ssd.
@Falkentyne Will post that bench in minutes.

 

Low 80's is bad if the GT73VR never exceeded 55C....(I am assuming the GT75VR is the same). But on alienware, the PCH I believe is in front of the laptop. You cant even access the PCH on the MSI without a total complete teardown. If all GT75 Titans are getting 80C PCH, I will no longer be able to recommend MSI to anyone. Even 80C PCH can cause stuttering and problems with SSD access speeds.

 

My 960 EVO is on fire but only in the sensor 2...

 

Have always recommended notebook cooler with bottom cover mod. Some notebook models will get problematic High internal chassis temp due less air flow... + you save your notebook fans. Longer lifespan.

 

The 330W PSU would work as long as you don't overclock. The 330W Delta can pull up to 380W directly from the wall. It should be safe, but unless you can get it for cheap, I would not spend $200 buying it. Might as well just get the 780W unit. But if you already have access to the 330W unit, use it.

The 8950HK can go up to 150W if you overclock. Then 200W GPU+150W CPU is getting into limits area, because you still have 30W for USB, mainboard, LCD and DRAM/hard drives. That would be like 420W from the wall with the inefficiency, and the PSU will just shut off before that point.

As long as your CPU uses less than 120W at full load, you should be fine.

However make sure that your motherboard isn't throttling (in throttlestop, ->CPU PL1/CPU PL2/EDP OTHER in red) prematurely if you are using the single 330W.. You need to check this and I can't do it for you. I know that on the GT73VR SLI with the 2x230W (for gtx 1070)., if you unplugged one of the two 230W PSU's, the CPU would be capped to 45W TDP because somehow the EC would detect that one PSU was not plugged in. I don't know how.

So you'll have to make sure you are getting equal performance with your 2x230W vs one 330W. However a person here did test the 780W on a GT75 Titan and there was no throttling.

 

Yuck.
91C on the SSD RAM...
>_>

That's bad.
I'll test my 960 Evo I guess.

 

which taptop cooler shall I buy bro?

 

Search U3 Mod (from me... Pooopusan) in the forum. I’m out with Ivan The Terrible walking the long mile

See... CoolerMaster Notepal U3

 

The 780w power supply you speak of is this one right?

http://www.hidevolution.com/780w-la...oc-p775dmx-p870dmx-p870tmx-msi-gt-models.html

It is expensive indeed, but yes I was just looking to have a single brick to carry around since two is not ideal. I will see if there is any cheaper versions available but if not then I'll probably go with the 780w brick.

 

I don't know what MSI is doing to your guys laptops but this is mine.

 

If your PCH is on the other side of the board and getting airflow, that would make all the difference.

Also: Ouch at those 960 temps. My 970 barely hit 50C at most.

 

Auto fans?

 

And this is mine... http://forum.notebookreview.com/thr...ers-welcome-too.810490/page-155#post-10685128

Edit. How SSDs temps is in the Thin modern Alienware’s... http://forum.notebookreview.com/thr...review-by-phoenix.801314/page-9#post-10486496

 

Every review I read said they didn’t have any throttle issues. They said even though it got a little warm inside, it still maintained its high clock speeds. That’s kind of disappointing that a unit this massive with this type of cooling has throttling. I would be very nervous repasting due to being worried I would break something. I mean I’ve opened up the back of a laptop before but how hard is it removing the heat sink to repaste? How far did you undervolt?

 

Did you run ATTO disk benchmark?
That benchmark heats up all SSD's.
And the PCH on the GT73VR, GT75VR and GT75 Titan are in the exact same location, behind the motherboard.

 

It throttled under a heavy synthetic load and then only barely. I also seem to be having general heat issues so it may just be my unit. Removing the heat sinks are very simple and straightforward, just watch one of the many good video guides in this thread. I have undervolted as far as 150mV and was still stable with a significant reduction in temperatures. Your mileage may vary but you should have no trouble doing, at absolute minimum, 100mV

 

I wish I worked for MSI. If I had a GT75 Titan to tear down I could find out exactly what was wrong with it. Assuming I didn't make the magic smoke come out first.
But loaner laptops for stress testing are only for people who either have jobs, professional gamers, or those with VERY VERY close connections in the company, so I guess i'm done until I get a 9900K.

 

Soooo guys I tested atto bench as @Falkentyne suggested. It did get a bit cooler inside the house though, "only" 27C room temp. I have to say crystal disk bench punished it more than this. Fans were at idle speed the whole time, 1200rpm AND I only had firefox open in the background, otherwise I wasn't doing anything (you can see that on my perfectly idle gpu temps and power draw hehe).
Max PCH: 75C, max ssd: 71C. I do have to say the PCH seems to idle around 58-59C mostly (so pretty much where it is at your highest Falkentyne)
EDIT: I do have it propped up on the coolermaster notepal u3, but it has NO fans attached, so it shouldn't influence anything.(one of those little fans was making noise so I gave up on it)

 

Am I seeing this right though...? My raid setup seems the fastest?

 

I guess my little drives aren't enough to tax the PCH but here's a data point for anyone that's interested.

 

SATA ssd's don't run Hot Only PCIe NVMe drives.

Its just good for the lifespan of your ssd if the NAND memory chips don't run too coold. The problem is the overheating controller (see hottest temp sensor). That's why the thermal throttling happens and not from the memory chips.

 

Changed the thermal pad from HS of ssd and in first test i got 71ºC, second test that is in the image i got 78ºC...i think that we cant compare sdd with diferent capacity...all tests are done with auto fans.

Also, anyone knows what means the letters S,P,P?

 

Well I've done some research and have found that an exact match for the ram in my laptop is pretty scarce online, not to mention awfully expensive. What I am going to end up doing is purchasing a 2x8 GB matched set of ram. I'll be going with a more performance oriented ram than what came stock with the laptop. Does anyone know whether 3000 mhz ram will run at full speed in the GT-75 Titan? It's not like there are many kits out there that run at this speed anyhow, but just looking for compatibility information.

 

The letters probably means the placement for for 2 pcs M.2 SATA drives. But I mean I have seen @Phoenix talked about he couldn't get it to work with two of them But 3 NVMe drives worked.

 

You would have to test it yourself. I remember there was *ONE* user who had RAM faster than 2666 mhz running in the GT73VR and I could also be completely wrong about that too.
Regardless, you would HAVE to unlock your Bios, gain access to the DDR RAM multiplier (since 9 is 2400 mhz and 10 is 2666mhz), and to the DDR memory voltage setting (I believe which is in overclocking performance menu) to get more than 1.20v.

You also gain access to the "tertiary" RAM timings section which is completely useless to anyone except maybe @Meaker@Sager as I doubt most people outside of world record benchers know what those timings even do. Plus a RAM timing fail=hello pressing the power button down for 45 seconds to wipe the CMOS

Besides if you're going to be trying 3200 mhz or faster DDR4 memory in a BGAbook, why not just get an 8086K or 9900K and a LGA system that can actually run fast enough to show its legs with such expensive memory? BGA isn't fast enough to make expensive memory worthwhile (4 core BGA=1 speed bin clock for clock lower than LGA 4 core, and 6 core BGA=2 speed bins lower than LGA 6 core. Proof= @Papusan ).

 

So letter S is for 1x M.2 SSD slot (NVMe PCIe Gen3), and letters P and P are for 2x M.2 SSD Combo (NVMe PCIe Gen3 / SATA )?

Because i have the OEM ssd(PM981) in one of "P" letters and my 960 EVO in the another "P" letter...the OEM ssd was disabled by me in OS so the ssd dont heat anymore but should i put the 960 EVO in "S" letter? Any speed diference on it?

@Phoenix

Need your opinion on this.

 

MSI Support confirmed that I likely have a defective unit. It is getting shipped back to the retailer and a replacement should arrive on Wednesday.

 

I hope you have more luck with the new one. Also, I re-read your previous post:

You should have totally had better results on liquid metal, I went ahead and tested coolerbooster now. I get similar results on stock cpu paste and gelid gc extreme on gpu (decent paste but pretty average these days). With coolerbooster on and 28C room temp I get cpu in prime95 max 80 degrees on hottest core (undervolted using ac dc loadline=1 setting) and in unigine heaven bench I get max 75 degrees on the gpu (with no undervolt, totally stock and pushing 180-190W steady power draw with 1848MHz).
When I undervolt & OC my gpu (the way I like it): max 67C and steady 1860MHz...
Now yeah you may have had 8850H but on stock they are very similar (mine is 3.9GHz all cores, yours 4.0GHz) and given that you did not mess with bios they should throttle just the same in stuff like prime95 and give pretty much the same result in temps/frequency.
PS. with coolerbooster my PCH temp stays below 70 when running ATTO disk benchmark (65-69C) and when gaming it would stay at 60C with max fans.
Your laptop clearly had some weird temp issue on all components unless you live in sahara and have no AC and I think if your new laptop is okay you will be a lot happier with it ^^