The Official MSI GT73VR Owners and Discussions Lounge

I will try, this is my 3rd time in the row to repaste in 1 week I am little tired of it. I do not have any temps problems just that core diffrence

 

Test the fit with soft cream or toothpaste before the final application. I used soft thin medical sgel for my testing of the fit. Use what you have in hand forum.notebookreview.com/clevo-overclockers-lounge

 

Haha I am very impatient to solve this and only way is to order from Ebay and wait for shipping 2-3 weeks Is there any other good brand? Maybe I can find that one in stores in Serbia.

 

hmm good Idea. I will try. Thank you for help

 

Anything that is 0.5mm thickness is fine.
And please don't complain about the # of times you repasted.
I repasted 8 times with Kryonaut and with Phobya Extreme, 8 times, before I gave up and went to Liquid Metal. And yes it was always cores 1 and 3 getting higher temps. Because there is a pressure problem. This is a design issue with this laptop. The only 100% fix would be to use a very thin copper shim, but you must always be careful with shims. Good temps but you don't want to crack the die.

And then I did about five or six LM repastes before I finally got the best application I could which has now lasted for 1 month without getting worse, after I switched to 0.5mm pads and put a little extra LM on the CPU (since some does get absorbed by the copper):

(prime 95 small FFT, AVX/FMA3 disabled, 4.7 ghz, 1.271v (1271mv) manual voltage, IA AC DC Loadline = 5 (in Bios-->CPU VR settings->Core I/A Domain), core temps 87/84/87/84 or 88/84/88/85 after 15-20 minutes).

Note: if you can buy a shim that is thin enough, you can fix this problem permanently, but then you will have to be creative in finding a way for the VRM's and chokes to get heatsink contact (may require thicker thermal pads or stacking pads).

 

you have one perfects temperatures, well done mate . I am expert in repasting and I do not wanna try with LM. It is too risky for me. I will order this pads. I need to wait 10 days, it is not so bad . How much I can get with this pads? I am good with 7-8 C difference. If I leave like this 15 C difference because I am not overclocking more than 4 ghz and there is no way to throttle because I am reaching 80 C just in prime 95, in games max I saw is 77 C. Is there any harm with this 15 C difference? I mean lower life of cpu or anything like that. Thank you for help!

 

Two things are of concern about GT73VR and GT75VR.

1) VRM cooling requiring heatsink contact and pressure.
2) what the hell is this? Pressure squares on the left side of this heatsink picture, to attach to "Exposed" part of the mainboard (nothing is cooled or short circuited by this; mainboard section is empty there)? why? This can't be good for pressure...

Look at the left side of this heatsink:



The only way to fix something like this is a shim.... but again you need to stack thermal pads to cool the VRM's.

 

You can probably shave of a bit and instead use K5 Pro or thin pads.

 

Never seen K5 pro but I saw you talking about it in the Alienware group.
This core temp issue is a design problem with these MSIbooks.
I don't know why no one ever discussed this in the past.

I think @Donald@HIDevolution 's IC7 repaste uses thick enough paste to prevent dryout problems like this.
Maybe Arctic Ceramique 2 (designed for very high temps; resistant to dryout) should be tested on these BGAbooks too....even though thermals won't be great, core temps should be closer with such thick paste.

*Edit*
maybe someone can try sanding down the posts (on the left) with sandpaper slightly. Is that aluminum? can that be sanded down?

 

Any engineers or smart people here?

Anyone think this would work ?

0.5mm pads plus this: (sanding down those posts)?

 

Thermal Paste Product Details

What I suggested in previous post.

 

The paste is fine, it's only been a few days - it looks like tension on the heat plate / CPU heatsink shifted, which happens when flexing / moving a laptop.

"Movement" or "Settling" of the cooling components can happen even if you aren't physically moving it, with the heating and cooling causing physical pressure changes through expansion there will be shifting.

You are going to at least need to get in there and re-tighten the fasteners around that CPU cooling assemblage, and see if that's enough. If not you'll need to disassemble and re-paste. If it lifts enough it will introduce gaps that when re-tightened may introduce voids in the contact through air "bubble's".

When it's fastened down at the factory they use torque limiting wrenches to tighten enough, but not too much. But, if too loose or too tight the lack of tension or too high tension can shift the parts and cause a lift on one or all corners.

That lets the contact pressure drop.

That's also why I don't recommend opening the laptop and doing this yourself, most people don't know they have to find the right tightness balance to have the assemblage stable long term. There is also a particular order of tightening that gets around any uneveness built into the design and physical implementation.

When you tighten one bolt that takes "wiggle room" away from the other 3 bolts, so you want to figure out the order to start and tighten 1/2-1/4 turn to "bite" to hold a corner and then start each of the other 3 corners, and slowly roundly tighten them for even contact.

Try backing off each screw, not completely as that will "pop" up the whole plate and trap air bubbles in the paste, just back off enough so you can get moving room to even out the contact pressure all around the plate.

This is the craft or technique you need to perfect, the feel or eye to see the movements as you slowly turn the screw(s) and watch for movement of the point tightening as well as the effect moving other components. It takes calmness and focus and notes to write down what you did exactly and then see what effect it has on the temps, then move only 1 thing and see what effect that has, etc. Leave off the bottom cover while doing this - otherwise you'll strip the back cover screws

I wouldn't try "grinding" any of those "protrusions" / tabs, you don't know how thick the material is or if there is heatpipe / vapor chamber under them. Plus the "plating" is likely thin too and you might expose metal you don't want exposed.

I've used NT-H1 myself, it's good, but if you can find a modern paste - take the time to get it before re-pasting, but you can re-tighten while waiting for delivery of the paste and thinner pads.

Or, you can RMA to MSI to re-apply the paste, and they can tighten it properly (hopefully they are equipped to do so in that office), and test if before shipping to make sure it's working right. Tell them you tried to do it yourself, but couldn't get the core temperature differential down below 5c, and your hottest cores are going into thermal throttling.

Be honest, they may still charge you but less likely if you are honest from the start - if they catch you in a lie they may even refuse to service it. You need to get / keep the service technician on your side; they are the ones that decide if you pay or if they will do it at all.

They are likely going to give you crap about trying to do it yourself, but remain calm and polite and demur (don't try to gain the high ground) to them and ask for help.

Please post back and keep us up to date on your progress. Good luck.

 

Thank you very much for very detailed info. I will try everything. Just one question, when I am putting pads, do I need to put them on cooler or on VRAMs and chockes directly ?

 

Depends on the application. When you disassemble look to where the existing pads stick - some stick on the components, some stick on the plate. Be sure and remove all remnants of the original pad / glue / cruft before affixing the new pads.

I wouldn't put pad's anywhere else than where you find original pads, at least not now, focus on getting the heatplate CPU level contact.

While you are in there, you can check around for other components under the cooling hardware that could do with a pad to make contact, or K5 Pro to reach the gap, if any.

Keep in mind that the original design of the fitment and pad placement / thickness are *supposed* to work - they are supposed to make even contact when assembled correctly, so great change, sanding, different sized pads *shouldn't* be necessary.

If you only knew the order of assembly, order of screw tightening, and torque value for each screw, that *should* get you the correct results.

What usually happens to some units is someone goes out of order in the assembly, doesn't carefully align the motherboard in the push tray - off canter paste square placement, or forgot to use the torque wrench - or they lost / misplaced it or ran out of them for the number of people affixing cooling hardware.

Try to look at it from the point of view of the mechanical designer, thermal designer, and assembly instructions - which you don't have, and after some thought it should become clear how to put it together, perhaps after a few misstep's, so don't be afraid to backup and retry / reorder what you are doing before locking it all down.

Good luck

 

Off topic here but I noticed on the 7th gen 7820hk GT73 page that MSI released true colour 2.0 and of course the 6th gen 6820hk has nothing. Are those compatible?

Same for audio/nahimic drivers?

 

Generally yes..., but with True Color there are model specific ICC profiles that match the display, and if there isn't a match in the pkg for the other model, it won't be quite right.

I run without TrueColor, I just load an ICC profile into the OS and skip the filters.

If you have the installers for what you have installed, I'd feel comfortable trying out other newer versions from MSI.

The Nahimic drivers and Realtek drivers need to be matched to work with each other, and you want to uninstall Nahimic first, reboot, uninstall Realtek, reboot, install the new Realtek, reboot, then install the new Nahimic, reboot, and see how it works.

Generally I don't worry too much about upgrading drivers unless I have a specific problem, and haven't for a long time, but then again, I don't use Windows 10...which updates randomly breaks things and fixes things, at various times.

 

iv

i got exact same temps after two repastes with phobyia although first wasnt very good job but same here in both times temps good few days now around 10-15 core temp difference sometimes but im not bothered yet with it as gaming doesnt exceed 86 in extreme demanding games like assasins creed origins, im on 0,5mm arctic pads and im running 4.1ghz -100mv with icc max to 1000 in bios and have not thermal throttling so ill keep with that for a few weeks and then i try repaste again but dont think it will help

 

@Ivan994 @GENOCID

Try to increase pressure on the screws closest to the VRM's (tighten well) and have less pressure (slightly less of a full turn at the end--dont overtighten) the screws on the farthest end (by the RAMS).
See if this helps.
The problem is caused by the chokes and VRM's. Only way to avoid this completely is to use something even thinner than 0.5mm pads (like a thick nonconductive material like K5 pro) or to remove the pads and use a nice thick paste (like noctua, phobya, etc, directly on the VRM's and chokes, but I do not know if that's safe. @leeloyd did that on his Alienware (direct paste on the choke and VRM's, no pads) which helped but he had a warped CPU slug or heatsink so so he still got high temps at high overclocks but it helped greatly at lower speeds.

This is why you should do a pressure test with something like toothpaste (Yes, toothpaste, because everyone has it, unless of course you have a lot of spare thermal paste). If you don't have pressure paper (pressure paper is best), you can apply a dab of toothpaste to the very center of the CPU, then tighten the heatsink fully, then remove the heatsink and look for the spread pattern. If you have less pressure on cores 1 and 3 compared to cores 2 and 4, the spread pattern should show an uneven spread.

 

I do not have any temps issues, higher I saw in games was 77 C. Without undervolt and on 4Ghz CPU. I am very happy with it just this difference. I will try with 0.5 pads and to screws better like everybody advice here

 

You have uneven temps. that is temp issue. Uneven core temps: core 1 and 3, compared to core 2 and 4.
Look at this picture.



This is AW turdchip 7820HK
@Papusan (thank you Master).

Cores #1 and #3 are closest to the VRM's at the top. Cores 2 and 4 are closest to that light green thingy at the bottom.
On the MSIbook, cores 1 and 3 are also closest to the VRM's, but in below images, are at the BOTTOM rather than the top (Look at the orientation of the BGA). You can tell from this video:


This picture is blurry, but you can see the same exact layout here: (look at the top left of the slug, see those group of circles? they match the circles in the above AWbook picture.


If that's too blurry for you you can look in this video at 40 seconds and see it more clearly:


So, in both the second picture and video, cores 1 and 3 (hottest) are closest to the VRM's at the -bottom-, cores 2 and 4 (coolest) are closest to the RAM at the top.

So you need to find a way to increase pressure on cores 1 and 3.
unless the "metal posts" on the heatsink (on the GT73, to the left of the CPU area ; it touches the exposed parts to the LEFT of the CPU, as a mounting balance spacer) is causing interference, which it probably is not, as core 4 might be hotter too (This is just a GUESS), you could try, if you have extra pads left, removing the pads off the VRM and chokes, applying a thick layer of thermal compound on the VRM and chokes, then re-duoing thermal paste on the CPU and reapplying. If I were going to try something like that, I would keep the 0.5mm pads on the VRM's, and remove the pads on the (Grey) chokes and apply paste on the chokes, but if you REALLY want to test this 100%, remove both pads and apply paste only.

I TAKE NO RESPONSIBiLITY IF THIS DESTROYS YOUR LAPTOP. An alienware user did this and was successful.

If this causes perfect temps, then you know the problem. Perhaps try softer, even thinner pads later.

Also look at this picture again:



Since core 3 seems to be ALWAYS hotter than core #1, maybe ALSO try sanding the FIRST TWO parts i circled slightly (the square at the top, and and then the rectangle below that, just like half a mm, if you have sandpaper. You can try that too. Of course it's best if you had a spare heatsink (They are available on ebay and aliexpress, and MSI will sell it to you if you can contact "Tom".

 

So can I use Conductonaut on the GPU as well?

 

@Phoenix I'm working with @Donald@HIDevolution on purchasing this pc. What do you think about it? and will you help me tune it if i decide to get it?

 

Of course. Works well. Just use foam dams protection and proper prep (either super 33+ tape or transparent nail polish).

 

I was lucky enough to get THAT same laptop when they had them in stock last year with the CM238 chipset which has better performance and Audio. After that, MSI decided to release other variants with the other chipset which had no Sabre DAC Audio and not as good of a chipset in preparation for the release of the GT75. I've had this laptop for 1 year and 2 months now and this is the longest I've ever owned a laptop, just comes to show you how great it is, starting from the keyboard, to the screen, the performance, the cooling, the fact that it just works out of the box without the need of any modification. Also when you get it from HIDevolution now you can get the extra Fuji Monopoly pads or whatever they call them for even better cooling and you can get Conductonaut.

Once you get it, tag me again so I can help you tweak it, here are the Phoenix Tweaks and I'll also hook you up with an unlocked BIOS if you're brave like Phoenix

1- Uninstallation of all the Windows Store Garbage Apps like 3D Paint, Print 3D, Alarms & Clock, Feedback Hub, Maps, Groove Music, News, Weather, etc.
2- Removing all Windows 10 Privacy Invasion stuff like Telemetry, sending your keystrokes to Microsoft, automatically installing suggested apps in the background, allowing Microsoft to conduct experiments on your laptop, disabling automatic driver updates through Windows updates, disabling Windows 10 ads, disabling Cortana, Cloud Search etc.
3- Removal of nVIDIA and Intel Telemetry
4- Adjusting a lot of options in PC Settings like disabling Windows Tablet Mode (you have a laptop not a tablet), disabling suggestions appearing in the start menu, disabling automatically connecting to Paid WiFi Hotspots, disabling Game DVR which reduced performance in games, and a lot more.
5- Replacing Cortana with a much better search tool that finds results instantly called Everything which finds any file you want as soon as you start typing even a few letters of its name and it doesn't even rely on the Windows Indexing Service to do this!
6- Restoring the classic Windows Photo Viewer
7- Restoring the Classic Windows 7 Start Menu
8- Removal of the new and useless 3D Objects Folder that appears when you open "Computer"
9- Completely disabling Windows Defender from its roots (including all scheduled tasks and startup files) for those who want to install their own AV. ESET NOD32 Antivirus is highly recommended if you want the utmost security with the least system performance impact.
10- Updating all drivers to the latest versions in case they weren't up to date.
11- Disabling the performance hit that is caused by the Meltdown and Spectre patches recently released by Microsoft (optional, only if you don't mind having this security vulnerability and would rather have your CPU's full performance back)
12- Overprovisioning your SSD anywhere between 10-20% depending on how much space you can spare to ensure optimal and consistent performance at all times.
13- Blocking the Google Chrome Software Reporter Tool (which causes a high system load when it is scanning files)

 

ok I'm lost there, once I have Conductonaut I'll watch the video to see what is all this nail polish crap My wife better not see me entering the house with a bag that has women nail polish in it

 

Nail polish (Cellulouse based ONLY), transparent, is a very good insulator for the exposed SMD resistors, and it's very popular on the Delidded CPU group on overclock.net. It also gives lower temps than tape for obvioius reasons.

Use 3 coats. coat (brush stroke 1), let it dry, coat 2, let it dry, coat 3.

Foam dams are for people who transport their laptops around often, to prevent any stray LM from escaping the CPU/GPU area and getting on the mainboard. Kapton tape and nail polish will NOT prevent that!! the LM will just happily run over the tape or polish and move towards the PCB...

 

1) How long shall I leave the nail polish to dry?

2) do I apply the nail polish all over the PCB except the CPU using a brush?

 

5-10 min per coat.
you only paint over the SMD resistors surrounding the GPU or CPU only.

Here's a video:



pictures of GPU: (on the right).

http://forum.notebookreview.com/thr...al-grizzly-conductonaut.799343/#post-10411462

Basically just the same places that would be insulated with tape instead.

 

Look on the CPU video. The nail polish he marks where he puts it clearly. You only need small dabs there.
On the GPU you basically have to paint all around the GPU (look in the forum picture link) because there's a lot more resistors all around it.

 

Ask your wife Or better, You can practice on your wife's finger nails to achieve a perfect application and same time receive all needed data Or let her do the job

 

Jeez would you mind helping me with mine? lol

I haven't done most of that, most of that stuff I had no clue about. Win10 sucks.

 

sorry bro, only for HIDevolution Gaming Team members, it's a 4 hour tweaking session so I can't be doing it for every one

 

@ Falkentyne Thank you for info! I will try to repaste and screw more Vrms side as soon as I get pads form Ebay. Are you sure that 0.5 mm pads are not going to destroy laptop?

 

Ah ok. Thanks anyways.

 

I'm using 0.5mm pads right now....how would they destroy the laptop?
and I've put more than 1.35v into the CPU....

 

Can someone beat this?

 

I'm going to google down all the stuff in your list. Time to learn some crap about Win10.

If I run into problems or have any questions do you mind if I message you?

 

I don't do PMs. To me, a forum is a forum for public discussion. No personal or private anything.

To save you the hassle of searching, most of what I do is already here:




NBR Windows 10 Clean Installation Guide
Windows 10 Tweaks and Fixes (Index post #1)
O&O ShutUp10 - Do not use Windows 10 without it!
Macrium Reflect Installation/Usage Guide

 

I have the GT73VR 7RE Titan sli 4k 285ukl and was looking to use the spare m.2 slot. I bought the 2 pin head version but have found it is to big for the slot and the pins dont match. Is there a adaptor for this or can someone show me what I need please?

This is what I currently have

https://www.scan.co.uk/products/1tb...lc-nand-read-1800mb-s-write-560mb-s-155k-?v=c

 

I can not believe how I am bad luck I have manage to undervolt CPU just -50 mv ((

 

Awwww, well, it could be worse, but I know it can get better over time. I ran one CPU at -15mV initially but after 18 months I got it to -45mV, so maybe you can get your's to -75mv or -80mV over time.

Is that -50mV at stock CPU settings, or OC? Higher clocks 4x will require reducing the undervolt a bit more the higher you go, and tuning for lower clocks will increase the undervolt available too.

Hows the temps at -50mV at 100% 4x CPU load?

 

It is OC to 4 Ghz. Hmm I can not see some big difference maybe 2-3 C. Cinebench with turbo fan is about 77 C and prime 95 82 C max. In games is about 77 C but first time ever I saw in games 80 C in new Assassin Origin, it is so cpu demand game. I saw 100% cpu usage in that game, crazy

 

What kind of undervolt do you get at stock speeds? That's where I recommend starting, and start with -100mV, maybe that will still work for you? At least -75mV, maybe more.

Your temps are good enough at the 4.0ghz OC to stay there with -50mV too, but higher OC, like up to 4.5ghz with no undervolt might start getting toasty.

 

I have never tried with stock speed because I just undervolt using XTU and I am using Dragon center profiles and I need undervolt to work with all my profiles. I do not think I can push 6820 HK to 4.5 GHz. I think 4 Ghz is enough for games . I am not render guy or anything like that. I am just PHP Developer, I can work or 1 GHz CPU

 

Hey, cool then you are done.

The 6820HK has been known to be stable and run find at 4.2ghz, some at 4.3ghz, but it sounds like you don't really need to push it any further.

If you change your mind, the trick to going higher in multiplier than DGC allows, is to continue on in XTU. Set the multiplier there after DGC is set to maximum.

I did that on a 7820HK, 4.3ghz in DGC, and 4.5ghz in XTU with DGC running. If you change DGC performance settings you will need to start over in DGC then XTU again.

Or, eventually, drop DGC and do it all in XTU + MSI Afterburner + RTSS (frame limiter) + Silent Option (fan tuning) + Windows Power plan.

Have fun

 

Thank You for info, If I start playing I will report You result here! I have plan to fix uneven core temps before that I am not doing anything, 1 thing at the time

 

Ivan you ignored my screenshot.
@Ivan994 @GENOCID

Look at those core temps.
If I can do this, you can too.
Trust me.

(This is from sanding the metal posts about half a mm with 400 grit sandpaper, then 0.5mm thermal pads and Conductonaut

 

Sorry, but this is not the problem with this heatsink.
I tested my theory and dropped core temps to 0-2C delta (only got 0C once, usually its 1-2C) at 80W CPU power, and 2C at 100W.
The heat doesn't shift. It isn't tripod. It's very secure.
The problem is PURELY a combination of two things:

1) VRM+Chokes cooling (causing slightly less pressure on cores 0 and 2, compared to cores 1 and 3).

2) metal standoff support posts reduce pressure on core 0 and even more on core 2.

As a result, everyone with this laptop will have higher core temps on core 0 and 2 than 1 and 3.

I fixed this problem by doing 2 things:

1) 0.5mm thermal pads on the VRM's and the grey chokes. (helped a bit).

2) sanding down the posts that are 1mm higher than the "long" post that isn't connecting anything, with sandpaper. Sanding off about 0.5mm of metal.

As a result of this, I managed to get core temps to 0-1c (usually 1-2c) at 4.5 ghz, small FFT prime95 (AVX disabled), 1.175v vcore (got 73/73/73/73 but its usually 73/71/73/72 or 74/72/73/73 or 75/73/75/75!), and 2c delta at 4.7 ghz, 1.271v, prime95 small FFT (AVX disabled) 87-85-87-85 or 88/86/88/87.

The heatsink is extremely steady, but it is imbalanced because of these things:



As you can see, standoffs on right side of the heatsink (left side in the picture, right side when HS is installed), cause reduced pressure on the top left in the picture.
VRMs and chokes cause slight reduced pressure on the entire top.

Unfortunately, to fix this fully requires modding

Here is core layouts (cores 0 and 2 are the hottest).
This from an alienware but its the same on msi:


And here is our heatsink:
I have marked on it so you can see, although it's inverted:



So you see the problem.
Insufficient pressure on cores 0 and 2 cause thermal paste dryout, which is why those core differentials increase!
The heatsink does not move even one iota. This is no tripod AWbook heatsink!.

So that's the only way to improve the problem. Thinner pads and sanding those posts. Just the two square posts. The rectangular one does not go down as far.

 

Sorry mate. I have not seen first post. I am waiting for pads and I will try to fix it. I will be well with 6C difference. If I make it like you, It is going to be superb Thank You for all knowledge shared with ous! I have friend with 6700 Hq msi and He has also 10 C Difference and everything is stock, he has never opened laptop. You are totally right about everything!