***The Official MSI GT83VR Titan SLI Owner's Lounge (NVIDIA GTX-1080's)***

@johnnobts Listen to this guy, he knows what he's talking about. Falkentyne >>>>>>>>>

 

Hey guys, if I did a Liquid Metal repaste - whats the suggestion here in terms of a repaste? every 6 months or so ? Just wondering how long shall I wait before repasting. Temps are still below 84C (@4.4ghz) just wondering (at stock clocks and no overvoltage max temp is 76-78C). If you got any suggestions let me know - thanks.

 

You don't need to re-paste until the temps become worse significantly. Liquid Ultra can last anywhere between 1-2 years. I have no experience with other Liquid Metals just talking about Liquid Ultra

 

If you have a very good mount between CPU and heatsink, and core temp difference is less than 2C after repasting, the liquid metal should last for 1 year or longer.


Read:
http://forum.notebookreview.com/threads/after-1-year-of-thermal-grizzly-conductonaut.799343/

If the contact is this good like this, you won't need to repaste at all.
You can see judal's heatsink is PERFECTLY FLAT and perfect contact with CPU. You can tell by the shape outline. Because the contact was so good, the paste did not "Oxidize" and crystalize on the heatsink or CPU and remained good. those are nicely flat heatsinks.

But MSI has been known to make worse heatsinks.

Look:
Can't you tell this heatsink is worse quality machined? Compare it to Judal's heatsink!!!!





if the contact is bad (core temp difference >3C, without AVX or FMA3 instruction test), then it will become worse with time, just like regular paste (except IC diamond).

I would use 3 coats of nail polish over CPU SMD resistors on the green BGA package. Same thing (even more important!!!!) if you want to LM the GPU's too. If you do the 3 coats of nail polish (let it harden for 5-10 minutes after each coat), you won't need to use the "Kapton tape" or "Super 33+" tape, because the nail polish does the same job.

I also prefer to use foam dams around the CPU for 'extra' protection, very compressible, very light, must compress down to a MICROMETER with your fingers!!! and you can 'attach' them to the BGA housing with dot of nail polish at each corner too: Foam dams are NOT TO BE USED IN PLACE OF NAIL POLISH OVER SMD RESISTORS, but IN ADDITION TO IT--foam dams are insurance to 'trap' any extra LM fully inside the CPU or GPU area, it cannot penetrate the dams.

Here is an example of a foam dam cut out in the shape of the CPU.



As far as correct foam, the left side is BAD FOAM, this is TOO THICK, will add too much resistance and interfere with heatsink pressure! RIGHT SIDE is correct foam. There may be other variations of this foam too (like, darker color) but importance is density. Must be very very lightly dense, very compressible to micrometer with fingers:

 

Thanks @Falkentyne . My last (and first LM repaste) was done with both nail polish and super33+ tape.

I remember an older post of yours saying the tape can actually increase temp by 1-2 C (small variation but you never know).

I guess the best way to find out if I have flat copper surfaces on my heatsinks is to open it up and see how the LM is doing after 3 months.

I might wait for it to reach 5 months and do a repaste (just to see how the copper heatsinks look like). Fingers crossed they are proper flat heatsinks that sit on CPU/GPUs properly leaving no angled gaps.

 

Hey guys! I recently got my GT83VR through the GT80 trade in program (I only have the 1070's, so I'm not sure I should even post here.) Much to my dismay, it didn't come with an SSD installed, instead having two large standard drives. As you can all imagine, it's just way too slow for my tastes and I've purchased some SSDs to install.

My question is twofold, first if there's an internal blueprint/schematic anywhere available so I know where everything is (just because I think it's cool), and second...the important question, is how do I go about entirely cloning the boot drive to the SSDs? Will it preserve the factory reset partition? Any must-have programs to help with the process? This is the first time I'll have attempted anything of the sort so any help you could provide would be awesome.

 

@Ardiem Just youtube or google GT83VR disassembly. Should be some guides.
And Macrim Reflect can image drives (if the destination partition size is larger than the source, i don't know how to resize partitions though).

 

Not sure how to answer this, but there isn't room in the laptop for 2 standard 2.5" HDD drives, so if you are seeing 2 partitions as mounted volumes, it's all from 1 2.5" drive.

There are 3 M.2 slots on top under the top panel just below the screen, there are screws on the bottom of the laptop, one under the "void if removed" sticker, that need to be removed first, then you can slide the top panel to the right and it lifts off. The M.2 drives are just left of center, although this has changed over model release's I think yours has 2 there and one further right to the left of the 2.5" bay (single 2.5" bay).

If you bought 2.5" SSD's, one will fit, but you will want to use the MSI BurnRecovery to make a bootable flash recovery thumb drive - 32GB USB 3.0 (don't get 2.0 or 1.0, they are too slow), and create the backup.

Then I'd also get used to using Macrium Reflect to make a image of the boot drive (all partitions) as you update and install new software, maybe a new image after every install session of note - so you can restore from it if something goes wrong.

Please keep us up to date on what you do.

Note: As @NuclearLizard points out there is a slim 7mm optical drive slot too, and if you pull that optical drive out you can fit a 2nd 2.5" HDD, but it's not a standard configuration and wouldn't come that way from MSI.

 

You got it right, under the panel it should be some m.2 Nvme drives and the 2.5 inch Sata.

I brought it up way long ago, I wanted to swap the optical drive for an ssd in the bay but I think it was on a different controller or not the same speed.

Sent from my SM-G930W8 using Tapatalk

 

Yup, the optical is Sata 1.5...I think that's how it shows up in hardware inventory programs, not SATA 3. It's about 250MB/sec vs 550MB/sec.

 

Yea, it was too bad, storage is starting to become a problem.

I haven't seen the new ones yet though I'm not sure if I'm ready to jump from my 6920hq to the 89xxhk.

When the new gpu's come out of course.

Sent from my SM-G930W8 using Tapatalk

 

There are 2TB M.2 SATA drives now, cheaper than the M.2 PCIE x4 NVME 2TB drives.

WD Blue 3D NAND 2TB PC SSD - SATA III 6 Gb/s M.2 2280 Solid State Drive - WDS200T2B0B - $493
https://www.amazon.com/Blue-NAND-2TB-SSD-WDS200T2B0B/dp/B073SBW3VD

Samsung 860 EVO 2TB M.2 SATA Internal SSD (MZ-N6E2T0BW) - $649.99
https://www.amazon.com/Samsung-860-SATA-Internal-MZ-N6E2T0BW/dp/B07822SVMS

 

We don't have those ones in Canada. And the western digital one is 780 dollars. Lmao

Sent from my SM-G930W8 using Tapatalk

 

That's the first I've seen them on Amazon US too, before this it has been OEM dealers that offer them, so hopefully they will show up eventually in the Great White North.

$780CDN =>$618 US, yipe!...

 

Yea it's pretty brutal up here. It's almost like they don't want us buying stuff. Lmao


Ah well maybe once you guys move on to fancier stuff we will get it at a good price.

Sent from my SM-G930W8 using Tapatalk

 

It's not "U.S. Guys", it's the South Korean guy's

And, it's probably some darned tariff thing, speaking of that, it might be prudent to order anything you need for the next year, before it goes up 25%+ in cost, yipe.

 

Lol it's everyone not in our borders. Lol


I dunno I'm thinking of just abandoning my current 3-4 year cycle and moving towards more middle of the road hardware again and just running it till it dies.

And yea they are gonna kill everything in the economy, hopefully amd and nvidia want to open some fabs here. Lol


It's a little closer to letterkenny in my neck of the woods.

Sent from my SM-G930W8 using Tapatalk

 

Thanks so much guys! I'm pretty worried about the cloning of the drive as it's something I've never done before, but I'm already several steps down the path you laid out for me.

I purchased this Toshiba THNSF5256GPUK - 256GB, NVMe PCIe SSD (M.2 2280) For use as my SSDs, for $65 I couldn't go wrong. I did double check and you guys are absolutely right, the storage drive is just partitioned strangely. I believe the m.2 drive I purchased is a B key type connector, am I reading things right? Are B key and M key types compatible? Otherwise, I'm not sure which of the three slots in the 83vr are for which. I know according to the specifications on MSI's website it has Two M key type and one B key type connector though.

I also watched the XOTIC PC teardown video of the Laptop to get a feel for doing it.

Thanks again everyone! You're all wonderful.

 

I gotta find some stuff for my setup.

Anyone know of a stand that would fit this laptop or a similar sized competitor


Also I want to find a reliable HDMI switch that can do 4k and multiple outputs.

Maybe I'll have to do a battlestation build log. Lmao.

Sent from my SM-G930W8 using Tapatalk

 

I'm pleased to announce that i've decided to join the MSIGT83VR club . Purchased a used GT83VR-6RE (1070 sli version) for 1800 and I love it so far sans some issues.

Hopefully someone might be able to help me out and steer me into the right direction.

The CPU temperatures are pretty high. I get thermal throttling on sport mode and any turbo mode when I benchmark in XTU. I'm pretty new to this whole overclocking thing but from the impression I get from this product, it should be able to hold 4.0 ghz no problem in turbo mode while playing games on ultra without hitting temps of 95c. I am the third owner of this laptop(possibly 4th). The warranty sticker was still completely intact so this unit had never been taken apart. I'm wondering if the CPU needs to be repasted. In the event it does need to be repasted, I would prefer to use liquid metal - Thermal Grizzly Conductonaut(saw the linus tech tips video where it lowered cpu temps by 20c). My question is are the heatsinks in the 6re aluminum? If so, then conductonaut can't be used ;(. If anyone has any idea how I can fix the temperature issue I would be forever grateful. I'm considering taking this laptop to a computer repair shop to have it looked over(I don't know much about computers, I just know the moment I saw this thing in 2015 I wanted it and finally could afford to buy one).

On a side note, the laptop came with 16gb of kingston 2133mhz ram. I just purchased 32gb of kingston hyperx impact 2400mhz ram to upgrade. The MSI website states the 6re supports 2400mhz ram but when I googled my processor(i7-6820hk), intel states the highest it supports is 2133mhz ram. I'm hoping I bought the right ram for this upgrade, anyone know for sure what the 6re will support?

Cheers

 

First, Congratulations on the new GT83VR!!

And, good news is you loaded the right tool to do the next tuning that will reduce your CPU temperatures at load, undervolting.

XTU has a dropdown or a text entry box for setting the CPU Voltage Offset, and recent CPU's will undervolt in an approximate range between -50mV and -220mV.

An undervolt of -100mV will drop 100% all core temperatures by 10c, and most CPU's will be stable at -100mV, so it's a good value to start with.

If you aren't stable at -100mV, you BSOD or hang or otherwise experience instability, try -50mV.

If you are stable at -100mV (or -50mV), try increasing the undervolt by -10mV (or -5mV) at a time, test each setting under load - like XTU Stress or Benchmark - and when you BSOD / Crash / unstable, back off +5mV and try again, backing off +5mV until stable.

Once you have a stable undervolt with XTU Stress / Benchmark, then run other CPU stress / benchmark and game for a while on a favorite game to make sure you are stable. If not backoff another +5mV.

Once you are happy with the load stable undervolt, the last test is the Idle stable test.

Exit all games and software apps (browser, explorer, task manager, etc) and let the laptop sit "Idle" for as long as you can stand it, up to 30 minutes should be enough.

Then if you BSOD or hang at Idle, back off another +5mV, do another Idle test, and you should be fine. You could back off +10mV to be extra sure, especially if you don't have time for another Idle test right now.

Then Save the final value in a named Profile, and either load it manually or let XTU do it at boot time for you. If you crash for any reason or interrupt normal Shutdown (power button off) XTU will notice and not load the Profile at the next boot, so you'll need to go in and select and Apply that or another Profile manually.

Windows 10 screws with shutdown / startup through Fast Startup which is a Hybrid Shutdown using Hibernate, and in the past has messed with XTU's monitoring of the shutdown, so I disable Hibernate:

Run As Administrator cmd window:

powercfg /h off

That disables Hibernate, deletes the 1x memory sized C:\hiberfil.sys hidden file (32GB if you have 32GB, 16GB if you have 16GB ram), and stops Fast Startup / Fake Shutdown.

There is also a thread on how to set up a script to apply XTU settings at startup through a script if Windows 10 messes up loading XTU profile at boot:

How to automatically under-volt in XTU without opening it
http://forum.notebookreview.com/thr...y-under-volt-in-xtu-without-opening-it.802143

Have fun

Here's a sample XTU -100mV settings view before Apply'ing, you can ignore the other changed settings:

 

Having more trouble now. Swapped out the 2 X 8gb 2133hz kingston ram cards for 4 X 2400mhz kingston hyperx impact ram. Computer own't boot up now. Keyboard lights up, on light comes on, i hear it start up and the monitor is black. No idea what to do

 

You know the drill, unplug all the memory and start out with 1 SO-DIMM at a time, check them all for function - make sure they allow you to boot, you may find a bad one - or more.

Then move to 2x, and try all the combinations to find the incompatible pair, if there is one - or more.

If you boot at 2400mhz on 1 and 2 SODIMM's, that may be the best you can do, otherwise try 3, then 4, sometimes that will even work - just working the SODIMM's in and out of the sockets can help make better connectivity.

Otherwise those 2400mhz SODIMM's may be too fast for 4x configuration, so try running them at 2133mhz.

Good luck

 

I took the new ram out and put the old ram back in. Still having the same issue . I saw a video where a guy used an eraser on the edges where the connecting points are. Is this something i should actually try? Will it leave marks on the ram?(I'll have to return the 4 new sticks I just bought if they don't work so I don't want to jeopardize a return). I guess i'll keep fiddling with the old ram in the slots to see if I can get it to boot up properly

 

And, the only change between working memory and now is the removal of the OEM memory, plugging in of the new memory, and now restoring the old memory?

Try leaving in the original memory, and do a CMOS reset. You can try unplugging the CMOS battery, or first try the power reset - from off or on, hold down the power button 60 seconds or so till it powers off, and start up again. (thanks to @Falkentyne for the sequence, still check your manual for model specifics).

If that works, you could try removing the original memory and plugging in 1 new SODIMM - pick the correct slot, consult the manual - and then do the power on reset again, and see if that resets and retrains for new memory.

I haven't tried the reset on the GT83, so check the manual for a sequence, or a CMOS reset "hole" in the underside.

It's a pain, but methodical and studied (manual or other source) patience will prevail, eventually.

 

That trick of using a "gum" eraser - soft kind that rolls up little bits as it takes away the corrosion is for "old" memory and old sockets, and has worked a treat in the past many times for me. I also use the harder pink eraser, but I have a studied hand and pressure so I don't pull up traces and pads.

I wouldn't do that in this case, it's unlikely corrosion, but it might be a new socket not filled before had some debris in the socket that got pushed in and shorted between pins - also unlikely - but possible, it's happened for me a few times before. That's why I blow out the socket as a matter of procedure when adding memory - or at least look closely for dirt in the socket.

It's more likely the memory timing issue, the laptop noticed new memory and tried to change setting to match, but it doesn't work, and now it's in a static state - not resetting again for the new old memory, which is why you do the CMOS reset.

Good luck.

 

The "hole" is EC reset. Basically disconnects the battery from powering the EC and resets all EC registers.
Nothing to do with bios.
Bios reset: press power, hold it down, laptop will shut off, keep power button held down for 60 seconds. Then release.

That will wipe the entire bios NVRAM and reset time/date to jan 1.

 

MSI advertises that the msi gt83vr6re is ddr4 2400 mhz compatible so the 4 8gb sticks I bought should work right? Is this a bios issue or a cmos issue? Was I supposed to do some sort of bios changes before I installed the new ram? Or am I completely off and don't understand? lol. i'm a huge computer newb, installing an ssd yesterday and the ram today is my first hardware project ever LOL.

 

I have 4x8GB in my GT73VR 7RE and theyre also overclocked. Works fine.
I still don't know what issue you're having.
Are they all the exact same sticks/brand identical?

BTW you DID wait the full THREE MINUTES before you gave up, right??

It takes 3-5 minutes to retrain the RAM. You will have a black screen for TWO MINUTES before the MSI logo appears!!!!

 

They are Kingston Hyperx Impact sodimm ddr4 2400mhz cl14 8 GB sticks. 4 of them(2 kits of 2). They were factory sealed when I got them in the mail this morning.

The issue I am experiencing is that I pulled the old ram out/installed the new sticks and my computer display won't turn on when I boot up the computer. Keyboard lights up, computer makes normal start up sounds, power light is on. USB mouse won't work either. I tried taking the newly installed ram out and putting in the old ram but the problem persists, I cannot boot up my computer now . Everything was working fine till I put in the 4 new ram cards.

 

I was not aware I should wait 3-5 minutes to retrain the ram, thank you for sharing that information. I think I waited maybe 2 minutes. I suppose I should put all 4 new ram sticks back in and try it back out?

 

Yup, can take a seemingly forever time. Go make coffee or watch a show, then come back. A watched RAM training never ends...

 

So then what i'm experiencing with my display monitor not turning on at all/usb mouse not working is normal when installing new ram on these msi models?

 

USB mouse typically won't work unless you go into the BIOS or OS.

Can you get into the BIOS? How many SODIMM's do you have installed?

Unless you did something else, putting the original memory back in should work, but may need retraining if you aborted retraining while the new memory was installed.

Did you do the CMOS power reset, CMOS battery reset, or the EC reset?

You aren't feeding back much helpful status / progress info that we can follow to give you specific instructions.

Do your laptop status lights come on when you power on? Which ones?

 

On the GT73VR, when new RAM is installed or even when just ONE RAM stick is pulled, the monitor will black screen for 60 seconds.
Then the power light will turn off and on quickly
Then it will black screen for another 60 seconds.

If the RAM timing succeeded, you should then see the MSI logo appear, then it will turn off again
then the logo appears. Then it turns off again. Then on. Then off Then on.
Then boots to windows.

If there was a problem, the power light should KEEP turning off and on once every 60 seconds. Please check this.

If the power light turns off and on for 60 seconds then REMAINS ON for the next 4-5 minutes nonstop, then something is wrong with a DIMM, a RAM slot or the contact between them.

My usual practice is putting Deoxit D5 on the RAM contacts. That fixes bad contacts fast.
Was able to fix a MSI steelseries keyboard connector this way as well as my radeon r9 290X giving the "beeps of death" in the Bios, after a liquid metal paste job. PCIE slot was dirty.....needed some Deoxit on it.

 

I didn't know how to get into the bios. I originally had (2) ddr4 8 GB 2133mhz kingston value sodimms installed. Went to put in (4) ddr4 8 GB 2400mhz kingston hyperx impact sodimms. Then this issue happened. I didn't do the CMOS power reset or CMOS battery reset or the EC reset yet(wanted to wait till my friend came over, he offered to look at it but he hasn't arrived yet and neither of us know if he will be able to do anything as he is more familiar with desktop installation than laptop.

Messed around with this laptop for hours straight today, even traveled to a friend to try his windows usb drive in an attempt to repair/install windows since it wasn't booting.

Update: After a TON of trial and error with the sodimms I was finally able to get the original 2133mhz sodimms to work + 2 of the 2400 sodimms for a total of 32gb ram. They are automatically downclocked to 2133mhz. After this was stable I tried swapping out the original kingston value 2133mhz cards for the other two 2400 hyperx impact cards, no luck, wouldn't boot again.

Then windows wouldn't boot at all and the computer was going straight to bios from the start. Couldn't repair windows etc. Tried to even boot windows from a usb device and that wouldn't work. Eventually I went into the bios and loaded default settings. Tried changing the ram speed to 2400mhz in the bios, didn't work after I put all the 2400mhz cards back in, no boot. Ram is back to 2133mhz and i'm using 2 kingston value 2133mhz and 2 kingston hyperx impact 2400mhz all at 2133mhz.

I'm thankful for everyone chiming in to help me, you guys rock!. The system boots windows fine now but i'm stuck using 2 different sets of ram.

I'm not sure what my next step is to try and get my 2 matching sets of 2400mhz working at 2400mhz. Any advice or instructions from here on out would be awesome. If someone wants to guide me through the process from start to finish and it's a success i'll gladly tip you via paypal.

 

Yeah, at this point take a break and use what you have working, and think about putting together a test plan to either get the 4 x 2400ghz SODIMM's all working at 2133mhz, at least then you'd have compatible memory in all 4 slots.

Where did you get the memory? Did you check the compatibility matrix for the MSI GT83VR before purchasing? Are they supposed to be compatible at rated XMP profile speed?

Right now you are relying on the 2133mhz memory profile to set the speeds on all the SODIMM's, but when you put in the 4x 2400mhz memory and want to set it to 2133mhz you are going to have to have all the settings available from the 2133mhz to put in manually - test the manual settings with a stick of the 2133mhz first, and get that to work, then swap in a 2400mhz with the manual settings and make sure that works, etc.

You need to know the compatible 2400mhz detailed memory settings to input manually after that, as for whatever reason the XMP profile data isn't working - it might not be compatible - but looser timings might work.

Or find compatible 2400mhz memory for the GT83 that is known to work, and return what you have now for the right stuff.

 

I bought the Kingston HyperX Impact DDR4 2400MHZ SODIMM kits on ebay from a reputable source. Both packages were factory sealed. Paid 300 shipped total for both kits(would have been 400 direct from kingston). In terms of compatibility, The MSI website states the GT83VR-6RE model supports 2400mhz DDR4 so i figured I couldn't go wrong. The Kingston website states DDR4 Impact @2400mhz is compatible with skylake HQ/HK processors(I have an i7-6820HK). As far as XMP goes, i'm not even sure how to select a profile to begin with ;/. At this point i'm pretty thankful that I was able to get the computer to recognize my ram again and boot windows.

I'm thinking that I might bring this laptop to a local shop and have them put it on the bench to figure out what is going on with the ram + high load temps when OCed. I'm at least the third owner of this unit and it had never been taken apart or upgraded(warranty sticker was still in tact). I have the feeling this laptop was cranked on turbo by previous owners for awhile and perhaps the factory thermal paste needs to be replaced. I think this could explain the higher than expected temperatures while OCed to 4.0 with a heavy load(for example Crysis 3 on max settings + DSR 4k CPU temp was at 95c with thermal throttling-fps goes down to around 40-47). Sport mode temps without the DSR 4k mode enabled for Crysis 3 range from 75ish to 85ish. In general, other core intensive games like starcraft 2 on ultra settings really rev the CPU temps up as well. GPU temps are always below 60c(according to dragon center). I took your advice for undervolting in an effort to reduce temps at high load. I can stable undervolt around .100 on comfort/sport modes but I have yet to find a stable undervolt for turbo mode when oced to 4.0 ghz(crysis 3 will sometimes flat out crash).

All in all my goals with this laptop is to get the ram up & temps down and game on comfort/sport and use turbo when I want to really crank the settings while streaming via twitch without needing to use the coolerboost feature on a regular basis. My second m.2 NvMe ssd install went fine the other day.

My current storage set up is as follows:
256GB Toshiba m.2 NvMe ssd(MSI's default option) - Using this for my OS & utility programs
256GB Samsung m.2 NvMe OEM PM961 SSD - Strictly for gaming programs
1TB standard HDD - Downloads/documents/videos/pictures etc

None of these are in raid, I still need to read up about what raid is/does to decide if I should raid these or not.

This leaves me with one lone empty m.2 sata slot. I might buy a 512GB m.2 sata ssdexclusively for back up/recovery. Maybe I can just automatically replicate everything to that to have a full safety net or something. I don't know what my best back up/recovery options are at this point or what the best storage configs are. Time for me to read up.

 

Only one note for now, you have it under control now, or at least the parameters to work with, the 6820HK pretty much drops off on undervolt as you OC, so try -15mV @ 4x 4.0ghz, or nothing at all, instead work on the fan curve.

Also for most things peaking under 90c is fine, 89c is nicer looking, and 93c is thermal throttling, as long as you are under 93c except for fleeting peaks, you are fine - and so is the CPU.

Re-pasting is a beginning that never ends... meaning you will want to keep at it till you break something, or tire of it, which might not be at an opportune time - it might not even be as good as stock - you need time / experience to do your best technique for best cooling.

And, anyone that hasn't done a particular laptop model before won't be much better off, even if they have done many before this. Plus they are working on a fixed budget, and when the hours run out, that's all you get - all the time they will spend whether the result is optimal or not.

Try to get an undervolt - even minimal at 4x 4.0ghz, and over time you can extend that a bit too. I had -15mV out of the box, and by the time I passed it on I was at -25mV stable.

Have fun

 

Repasting seems like a really attractive option. I saw the liquid metal - thermal grizzly conductonaut laptop repaste video that linus tech tips put out. The thermal grizzly liquid metal lowered highload temps by 20% for the model he used it on(compared to the stock paste on that unit). It wasn't an OCed laptop but It was thermal throttling before the repaste at full load and running under spec because of that. After the repaste his temps were down and his clock actually went above stock settings on it's own because it was running so cool. I'd love to be able to drop full load OC temps by 20% with undervolting + repasting. I'd ride it all day at 4.0 if I could have a steady 70c-75c temp OCed. Ever use liquid metal before? Ever repaste any of the MSI GT80/80s/83/83vr series laptops before?

 

LM is *conductive*, that means if you flip even a little of the "globlet's of doom" onto your motherboard components, or somehow some slips out during use and gets on a component it can damage your motherboard, and ruin your life. This happens enough to not recommend doing this yourself.

The non-conductive pastes work well enough to not risk using LM.

And, if you undervolt and adjust your fan curve, use RTSS to limit your frame rate - the GT83 doesn't have G-sync - you only need 60 FPS to match the display refresh, and even if you OC the screen, which I did on my GT80 to 100hz you can limit FPS to 100 FPS, both enough to limit FPS enough to take the "heat" off of the CPU and GPU.

Re-pasting usually takes a few practice runs before you get it right, and LM is not something to practice with. Even non-conductive paste runs will risk some event - dropping something into the open laptop - that can damage your laptop in ways you need to RMA it.

As long as you are not thermal throttling, there is no reason to chase thermals so low, really there isn't

It's a fascination for many that can't seem to resist re-pasting, and be happy with temps that the laptop will be happy with for it's whole life, so if you can't resist just know you are risking it all for nothing, no functional improvement - I call this "vanity" modding.

Anyway, the problem at hand should be resolved first before adding more variables like re-pasting, and you haven't had it long enough to get to know it and find any other flaws - give it a little time, no need to rush things.

Also the first 7-14 days are the initial return period, it's best not to change anything physical during that time, just in case something in the laptop fails - infant mortality - let that time pass and return the laptop should something fail - otherwise you don't know if it was infant mortality or something you did to cause it to fail.

If the laptop dies while you are modding it, or even upgrading RAM and storage, you won't know what caused it to fail and may waste time debugging what is instead an initial parts failure, and then you have to put everything back to stock to return it.

Best to move slowly with these things when first getting them, let yourself become familiar with it, and spend the time testing for port and functional failures so you know you want to keep it, before investing a bunch of time modding it.

 

There are little graphite/graphene pads that are coming that have similar cooling properties to LM coming.

I have been paying attention to those to see if they are a worthy upgrade.

Though I would recommend the solutions of undervolting and frame limiting first.

Sent from my SM-G930W8 using Tapatalk

 

HMmm, Sounds like LM might be better for a unit that will never move(desktop). I can't risk throwing away money.


I definitely need some sort of increased cooling though. I thermal throttle oced at 4.0 ;(.

As far as the return period goes, I made the purchase April 1st on ebay. Despite the few issues i'm having(which I think can be overcome), I'm still happy with my purchase since a fully built new desktop with the same specs would have costed roughly the same(all parts tallied up).

Overclocking a laptop monitor is a thing? I've ONLY ever played @60hz lol. I'd love to experience 100 or more on my laptop screen. What are the cons to OCing a monitor? Is there greater risk of pixels dying or the monitor burning out entirely?

 

Not to sure about the physical cons but yea when OC anything wear is generally worse and the small chance of blowing out your hardware though that's usually in case of a defect.

I have my panel and monitor overclocked to 95htz and it's been like that for near 2 years now.

Sent from my SM-G930W8 using Tapatalk

 

Did you try a lower undervolt?

Also, a note on MSI / Asus fan curves, they error on the side on trying to be "quiet" instead of "cool".

So when you hit the CPU hard with a benchmark, it takes a while for the fan speed to increase to the level it will cool lower than thermal throttle.

There are 2 solutions I use. Run the fans at 100% all the time while benchmarking, and start the 100% before starting the benchmark. That should stop thermal throttling registering in hwinfo64 even initially.

Or, use auto fans, but be ready to click the "reset" button on hwinfo64 after say 15-30 seconds to get rid of the initial thermal throttle triggered while the fans are spinning up, then you shouldn't hit thermal throttling again.

Also, find out how to turn on logging in hwinfo64 and set the interval to 1 second or 2 seconds and look at that log after the benchmark or gaming run to see how often the highest temperatures are reached and how long they stay at that reading, vs the rest of the time. Usually you will find rare peaks with most of the temperature readings holding much lower.

Sounds good.

The only risk is that you need to wait for the screen to give up on the new Custom setting and sometimes power cycle to get back the original setting.

100hz is typical, with some getting up to 120hz, typically 104hz-105hz is the top refresh, but I am happy running at 100hz.

You use an Nvidia custom resolution setting:

 

I'm so excited! I felt confident enough to properly navigating the bios settings and using the cmos power reset button that I decided to give the ram installation another go. I'm so happy I did. I took the old ram out and left in only the 2 working new hyperx impact cards. Booted to uefi figured out to put the ram frequency settings on AUTO and then I chose xmp profile 1. It booted to windows after that and the 2 sticks were clocked at 2400mhz, YAY!

After that, I put the other 2 new sticks in. No boot. Nuuuuuuuuuuu. I decided to switch the slots that the 2nd pair of cards were in. It BOOTED UP STRAIGHT TO WINDOWS!!!!. I double checked XTU and all 4 sticks are running at 2400mhz yay. Thank you to everyone who gave me advice the last 2 days, I could not have done this without you .

After this I decided to do a full benchmark for the computer on userbenchmark.com.

I put the laptop into turbo mode OCed on all 4 cores at 4.0 mhz. Graphics card only had a +200 modifier on the vram clock offset. I did not undervolt for this test.

Not sure how i'm supposed to feel about these results(Not sure if these are good benchmarks for my model or not), except the website showed I was clocked at 3.8ghz instead of 4.0. Not sure if because of the thermal throttling when the cpu was at 99% usage during the benchmark(the website stated thermal throttling at 99% usage) or what.

Here is the link, also attached photos for those who are on mobile.
http://www.userbenchmark.com/UserRun/8616008



Here are the results:





EDIT:
Does anyone know what is up with MSI's stock toshiba NvMe SSD deep que 4k ratings?
Also, the multicore results on the processor seem rather low when I compared them to the other top 6820hk benchmark results on the website. What could be causing this?

I came in 6th place on the list for the 6820hk processor and 2nd place for gt83vr-6re benchmarks. 1st place did not thermal throttle. I'm wondering if they undervolted. I should probably do another test with a slight undervolt when OCed to 4.0.

Anyone have any additional advice to prevent the CPU from thermal throttling? So far the only options I am aware of is undervolting, custom fan curve and repasting the processor. I've set up a custom fan curve already and I haven't found a steady undervolt when OCed to 4.0ghz quite yet. Still considering repasting, ruled out liquid metal due to the risks and might go with Kryonaut.

Second benchmark test: http://www.userbenchmark.com/UserRun/8617299
Processor showed up as OCed to 4.0 this time. Ran my coolerboost titan for this one. I was also undervolted by .075. I did not use any gpu vram offset modifiers this time around. CPU Multi core ratings were much better this time around. Same issue with thermal throttling though . Came in slightly below my first benchmark.

 

Did you log your temps with hwinfo64 to see how often you hit thermal throttling? Actually you need to look at the temps, once thermal throttling Yes is triggered it stays that way unless you reset or exit and restart the app.

Did you click the reset button on the bottom of the hwinfo dialog after the fans spun up to clear the initial thermal throttling reading?

Try the reset method, about 30-60 seconds in the fans should be caught up (maybe sooner) and thermal throttling should stop.

And, starting logging on hwinfo64 and use excel or just a text editor to read the log file to find the CPU temp current and maximum and see how often current temp is too high vs not too high.

75mV @ 4ghz is a good undervolt. The fan curve is the next beast to tackle.

Great news on getting it all figured out

 

I don't know what HWinfo64 is but i'll download it and do what you just said . Thank you again for all your help, you are a god send lol. Feel free to pm me your paypal address for a tip .

 

Thank you for the kind offer, but I am here to help for free, to pay back a little of what I've learned over the years to help others.

hwinfo64 is a monitoring tool that many of us use to capture data - logging - or just for display when watching the effects of changes we've made tuning, and to use as a visual aid when doing a screen shot for posting results.

https://www.hwinfo.com/download.php

hwinfo64 sensors only mode is what is used to monitor temperatures with that compact dialog. hwinfo64 is also a hardware inventory program to see detailed info on all of the components in your computer. Valuable to find out details on display, memory, storage, etc when looking for replacements or additional data online, you can get the part number and maker info.

 

Here is what sensor monitoring in HWinfo64 is showing from the benchmark test. 4.0 OC, .050 undervolt and cooler boost titan on. 1070s @ stock clock. Still thermal throttling as expected.

http://www.userbenchmark.com/UserRun/8618500

multicore cpu ratings are looking better, deep queue 4k on the toshiba nvme also improved substantially. This was my best benchmark so far.