The Official MSI GT73VR Owners and Discussions Lounge

I wouldn't use Grizzly Kryonaut on warped/Uneven heatsink. Or with too thick thermal pads. It will pump out. Depends on how bad it is. Noting will stay put if the Heatsink fits is awful in the long run.

 

@Falkentyne ive got phobya and i think i will order one more again because i think this paste has a potential in modest OC i think personally it can show off in harder overclocking it goods temps but i might be wrong its just from what i experienced in past 3 repastes.
in higher oc that temps on lower cores doesnt change maybe very little or not at all. althrough i had to order one from france its hard to find them i found only 2 left on amazon (for uk)

 

Then Try cooler master maker Gel Nano, it is same as ICD 7 and doesn't scratch the die at all.

 

https://www.amazon.co.uk/Cooler-Master-MasterGel-Performance-MGZ-NDSG-N15M-R1/dp/B018WJZDR0
@Falkentyne @Papusan what do you think about this?

 

I don't know. I use liquid metal now. I got tired of pastes pumping out and cores 0 and 2 being hotter than cores 1 and 3.
Unfortunately, I sanded my heatsink AFTER I started using liquid metal, not before.

My cores have remained 1C after 1 week.

I think you should pay attention to @Shehary 's results. He used sanded heatsink with traditional paste.
So my advice is still the same:

1) fix heatsink first.
2) after core temps are close and remain close, then change paste if you want lower temps.
a 'bad' paste should have all 4 core temps rise badly by the entire paste drying out.

 

Worth a try.

 

I can not find any better that Noctua paste in my country. I need to order for UK ebay and wait. I also think that paste is not the problem Core 1 and 3 never reach more than 72 even in prime. But other to cores with low pressure always getting much hotter.
Now it is 22 C difference, crazy...

 

Noctua NT-H1 is a Low viscosity thermal paste. Will never work properly on uneven/not flat notebook HS cooling plate.

 

@Papusan I love you
I didn't know that NT-H1 is low viscosity.

So he needs Phobya Nanogrease Extreme or Gel Nano ?

 

It's a reason it is easy to apply.
I would as well put Gelid Extreme in the line of good paste to test. Easy to find and good performer. Thicker paste, the worse HS you have. Not all can fix their HS to perfection. Zero temp between the cores shouldn't be the main goal. Rather go for good temps and lowest possible temp between the cores.

 

I can try only with cooler master paste. I can buy it here in my town. but still I need to sand my HS or no? I do not need perfect temps up to 10 C I can live it, but now it is 22 C that is too much...

 

Here is the 3rd (today) test result, room ambient temp little higher as compared to previous 2 tests,

First 72/72/69/70
2nd 71/72/69/71
3rd 72/74/71/72

 

Still perfect. I will be more than happy with those temps

 

For best results... Sand heatsink (be sure you know the risk if you shave of too much or do it wrong). Maybe try first without. Cooling in this Msi notebook is easy to maintain vs other Jokebook models. Aka disassemble the cooling, repaste and sand HS doesn't take too much time if the results wasn't as hoped first time.

 

I second that

 

I just afraid that I can shave too much but if I start doing that I will use sandpaper 2000, it will be very hard to shave too much. I mean I already bought 2 pastes and it is the same. I can buy one more but if is the same, I will just trow my money away lol

 

You used Noctua NT-H1 With almost equal temps on all 4 cores. If HS was very bad, you would not have experienced such results.

 

I had such a good results for 1 day Tomorrow 10 C difference and now 3 days after is 22 C difference My paste probably dry because of low pressure on cores near VRAMs. I would like to avoid sanding HS but I can not see any other way. I can order gelid extreme from Uk ebay and try but... I think my only way is sanding. I will just waste money on gelid extreme.

 

There is a very easy way to find out how much copper you are shaving, get a blue or black marker (not permanent) draw a grid or table on HS Copper, so with each lap you will start seeing the marker lines start disappearing and you will know how much you shaved and how much you need to shave this also tells you where the heatsink is uneven, where lines are disappearing quickly and where not, also same trick can be used when you have to shave a part of the heatsink, so draw a grid on that part and lap....

If you want to shave copper quickly, use dry sand paper, if you want to to shave copper slowly, use wet sand paper (make sure to keep the sand paper in water for atleast an hour before using it)

 

I think it's possible that @Papusan , @Ivan994 and @Shehary are all correct.
Low viscosity paste + weak pressure + uneven heatsink.
I have Nanogrease Extreme but no reason to use it since I am using LM. So I cannot test.

@Shehary your results are near perfect. I think your heatsink is flat. If you use LM, you should have 0C/1C temps now.


Non permanent marker: That is a good idea.
That will work nicely. VERY IMPORTANT NOT to use a permanent marker!!!

But I found with this heatsink, you don't need to really do that because with 2000+ grit sandpaper, it will get very shiny first, the part that you are shaving. Because it will get shiny quickly. But its safe to take shehary's advice and use the temporary marker.

HOWEVER if you already did initial top layer sand and got it polished first and are still having problems, you should use the marker.
This is a good advice.

Please remember, start at the bottom (colder cores) first.

*Edit* I can't spell.

 

i had a same results on first and second repaste on start as well temps very good between cores then next few days boom 14-15 now 20 as well a week after 3rd repaste with phobya all the time different method i used
i think i had a difference about 4-6c max after repaste.
im gonna take @Shehary advice thx man

 

@GENOCID You're using Phobya Nanogrease Extreme, right?
the thick paste.
So I think the sand the heatsink needs to be done.

@Papusan
I Think @GENOCID is using Phobya. And has same problem as @Ivan994
so paste is not the problem. Cancer heatsink is

 

yes im using phobya nanogrese extreme.. great paste if i fix that hs ill stick with that

 

Advanced Stream Detect is the bandwidth monitoring, yes, but I'm not sure if the Killer Network Manager, or the Killer Control Center, will stop I/Oing to that one file so long as its on at all. You might have to stop the service and process.

-- Anthony with Killer Networking

 

According to that post it will stop the file access... so it's worth a try

@leo zefei - you can also uninstall the Killer Networking Network Manager, or never install it, and that will stop the file access.

If you have the same version Killer Suite installer, re-run it and you can unselect elements and it will uninstall them while leaving the drivers installed.

If you don't have the same version Killer Suite installer, then download the latest Killer Suite installer - or the Drivers Only installer.

Uninstall the current Killer Suite using the Windows uninstall control panel (or a 3rd party uninstaller), then reboot.

You can then install the Drivers Only package, reboot and that C:\ProgramData\Killer\BWC.txt file access should be stopped - as the Bandwidth Management function isn't installed.

 

Thank You guys for an advice! As soon as I done something, I will report here my results.

 

I already want an Ampere GPU. I'm never satisfied.

I'm going x080 level card next time.

 

You can try it. Amazing paste I have come across.
You will have great experience with it, I reckon it can give good temps below 80C even on OC upto 4.4GHz on all 4 cores. Make sure your ambients are under 30C. I have very high ambients >30C and even turbo boost clock produced 80C on 31x multiplier.
If you want to go above 4.5GHz, say 5GHz then your best hope is LM, there's no other way like @Falkentyne said.

 

and what about gpu anyone runnin oc on gtx1070? ive tried core & memory +100mhz ive seen only little improvement but temps didnt cross 67c gpu after hour gaming. clock on 1936mhz and 143W
im not gonna play with it anyway just wondering if u can oc gpu in unlocked bios as well i suppose? with voltage?

 

@Falkentyne I increased Cinebench R15 score for 46x as well. Not much, one point but still

 

Hi everyone, I just got a good deal for the GT73VR this week (7820HK and GTX1070) and although overall I am very satisfied with the unit, after checking everything I have discovered what seem to be either tiny stuck subpixels or some tiny particles of dust under the screen (some bluish and one reddish). It can only be seen when the screen is black (easier with the light off) and getting quite close to it, and after checking it more carefully I 've noticed that depending on the angle I watch it whatever it is it can be seen on or not. Two of the bluish ones are slightly noticeable with a black screen (although with normal use I don't notice them at all) and the rest of them are harder to notice.

I don't know if any of you have experienced anything like this, but it's driving me crazy and I would appreciate any advise in order to fix it (I'm already trying PixelHealer but no success for now).

Thank you!

 

A problem with where some of the panels are assembled. If it bothers you too much, ask for a RMA.
Also I suggest you ask your supplier to test the panel before shipping it out, by paying for any dead or dirt behind the panel warranty. They should test the panel for any dirt or foreign particles if you ask them to.

 

I have the 1070 with the voltage problem with the .8A Vbios fix and I'm running +200 core and +500 memory with no problem. Have been playing witcher 3 for a few weeks now maxed and runs fine.

 

So with LM and non conductive pastes it seems they lower temps greatly but don't last long before the temps start creeping up again. I don't know how long but it looks like the extreme stuff lasts about a year before people are re-pasting again. I also don't know if that is because the people using that stuff are also running their clocks to the extreme and cooking the paste.

Which non conductive paste gives the lowest temps and holds those temps the longest?

 

If you apply as much paste as MSI does it should last longer

 

Thermal Grizzly Kryo will last min 3 and max 4 months with OC, I'm using it and had seen temp start rising after every 3 to 4 months and since I bought the laptop, after every 3 to 4 months, I open it, clean the Fans / HS, re-paste and good to go for next 3 to 4 months, if you are looking for long lasting paste, LM is the only option with good HS

Note: on GPU the paste last longer as compare to CPU

 

@Falkentyne

Today's test after cold boot, core temp difference remain same, I agree with you that LM is the only option to close the gap but 72 to 74 @ 4.1 with stock voltage, do I really need to use LM, all the effort and still do not have piece of mind when moving around the laptop, may be LM get out and start wandering on MB.....I did use LM on this laptop once for 3 to 5 days, didn't see any difference in temperature, only 2 to 3 degree drop as compare to non conductive paste.

First 72/72/69/70
2nd 71/72/69/71
3rd 72/74/71/72
4rth 72/73/71/71

 

is ur gpu " deffective"? my runs ok with stock vbios . whats ur max mhz u using? i went up to 2000mhz with 150W power. i never had a problem with my gpu and it never reached more than 67c in anythning i was doin. so thats why i was wondering if i can add a bit oc but improvements not that great only really few fps up. maybe it would be better if id increase voltage in bios? dont wanna play with it if difference is small.

btw i bought another tube of phobya nanogrease i could find only two left on amazon from eu so i couldnt be bothered.

 

I used grizzly kryonaut about 4 days ago I believe and temps were good st first and then started to progressively get worse, I just got Phobya yesterday and decided to repaste and the temps immediately lowered(i was checking heatsink and noticed it wasn’t completely flat as many have mentioned and kryonaut probably got pumped out, couldn’t even run Aida 64 without temps getting dangerous) and are I believe either the same or better than before when I first pasted with kryo. I need to OC first and see how the temps do, but gonna try that later, but so far so good, hopefully this paste lasts and temps don’t progressively rise like kryo.

 

If it does rise, you need to do what @Shehary did and polish the heatsink. Start off with doing a flat even polish of the entire surface first, since this is the exact easiest to do with no risk at all. Try what shehary said, mark a grid with a temporary (NOT permanent) marker, then start polishing with very fine grit sandpaper (2000 grit or higher). You should get it to a nice reflective shine (not mirror, just reflective) then clean up and repaste. See if that helps.

If it does, and the temps remain perfect after 1 week, then mission accomplished.

If the temps rise again with cores #0 and #2 getting far too hot, then you have to do some more sanding work--as I showed in previous pictures, you would start with sanding the bottom half (away from the VRM section) first (again, using fine grit sandpaper). Then once you do that, then do another FULL polish and then repaste.

 

4th repaste done with phobya nanogrease extreme. first test straight after repaste auto fans another test on max fans.. 4c difference for NOW.
did a spread method this time and made sure all screws were same pressure when tighten.

not much difference between max and auto fans
@Falkentyne @Shehary @Papusan @hmscott @Ivan994
ill give it a week that difference in temps rise about 10c in between cores. hopefully ill be wrong

i used this time smaller amount of paste but covered a corners as well a bit.

 

Looks good. What other changes than repaste?

 

pretty same.. 0.5mm arctic pads, but smaller amount of paste used and i spread it with a small plastic around cpu corners as well... then i attached heatsink and screw all screws same but not too tight just as they allowed me. no pressure. temps still stays around 4c but tbh im doubt they stay like that after few days

 

Genocid
did you find any 2000 grit sandpaper at your hardware store yet?
If temps rise, then do what shehary said.
1) find a NON PERMANENT MARKER
2) draw a grid around the heatsink.
3) sand it to a nice polished finish with water (alcohol works well too).
4) repaste.

 

not yet but gonna order it when i order new hs. this repaste is still on old hs. im waiting till 23rd for an order for new heatsink. then i do as suggested if temps gonna change which probably will anyway

 

I would run several Cpu benchmarks runs within the first days [Cinebench R15 - WPrime 1024M stress tests ]. This way you can have cold - hot - cold - hot Cpu temp cycles to put thermal paste in place.

 

I have the defective 1070 yes. You aren't going to get much more past 2000MHz if at all (MAYBE 2050MHz but you wouldn't notice it and it may even crash). No point going any higher for core but for the memory you can do +475-500. Seems 2000-2050 core and 450-500 is the sweet spot for the 1070. You will have to test but +475 should be safe for memory.

You can try the .8a Vbios as it gives better control and less erratic clock speeds. People with non defective 1070s switch to it also apparently.

I've never modified voltage and I wasn't aware you could without an unlocked bios which I have no idea if there is one or not. I wouldn't even bother for the minuscule gains and potential harm to the GPU.

 

@Falkentyne I’ll see how these next few days go and hopefully the temps don’t rise and stay consistent, haven’t tried a small OC yet but will later as I want to game a bit before trying to stress the Cpu, also I know you’ve been saying the heatsink isn’t flat and should be slightly higher away from the VRAM, but when measuring the heatsink with different rulers, etc. I found my heatsink actually has a very very small hump/bump warp in the MIDDLE of the heatsink, or is it supposed to be like that or COMPLETELY flat.

Edit: well decided to run a small OC at 4.0 and temp difference was dramatic core 1 and 3 up to 83-84, and core 2 and 4 around 70-72, ridiculous difference, guess I’m gonna have to sand after all,btw how much sanding is too much, as in how much margin of error do I have before I have to order another heatsink from MSI if I mess this one up?

 

Needs to be completely flat.

There's your bump.
you need to sand that thing down.
Make it shiny finish. Not mirror, although mirror is not bad.
That tiny bump is what is causing the problem.
I bet you that bump is RIGHT over cores 2 and 4. Causing cores 1 and 3 to get less pressure, the paste 'moves' outwards towards the less pressure cores, then evaporates.

Some people believe in sanding the CPU slug also I did take some sandpaper to it, only polishing of course, and ALWAYS ALWAYS wet (i use alcohol for wetness). But I only do enough to try to remove obvious marks and any leftover oxidized liquid metal. My CPU "Did" have a pressure mark on it before I ever started using LM, right over the cooler cores. Clearly that was from the uneven heatsink. I sanded that down and removed that mark. (Yes I sanded my "slug").

If you do want to sand the CPU AND the heatsink, be VERY VERY VERY careful with the CPU. Some people sanded slugs back in the old days before we had IHS's, because they tended to be slightly warped.

Margin of error?
Just use polishing sandpaper like 2000+ Grit.
Do what he said. Use a non permament marker and draw a grid, then sand the entire heatsink down to a polish. Don't use 400 or 800 grit unless you have a very good reason for it. it isnt necessary. Copper can be sanded just as easily with 2000 grit. The key to controlling sanding is wetness. the wetter the sandpaper, the more polishing and less sanding effect you have. So always keep it wet. Dry 2000+grit can polish quickly but dry should only be used if there are difficult marks you are trying to remove. Sandpaper is cheap. heatsinks are not. Best to take it slow.

Look at @Shehary's results.
He has had steady core temps for 5 days now, with normal paste, because he sanded the entire thing flat.

 

Thank You for your results mate. Look great. I had this kind of perfect temps also day after repaste and I used DOT method. Tomorrow my temps got worse and difference was 10C. Now it is over 20 C. I think I am going to try to sand NS tomorrow, I will share my results.