1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

[Help] Problem when OCing MSI GTX 1080 Ti Gaming X

Discussion in 'Videocards - NVIDIA GeForce' started by Mott, Aug 5, 2019.

  1. Mott

    Mott Master Guru

    Messages:
    816
    Likes Received:
    138
    GPU:
    MSI 1080 Ti GamingX
    I recently built a new PC for a friend and he wanted my ROG STRIX 1080 A8G and I had a chance to snag this MSI GTX 1080 Ti Gaming X for what he was paying me to put together and build / overclocked his new PC for him, so naturally I jumped on it as I have wanted that extra boost the Ti gives for 2k gaming. Anyways, the card seems to be in perfect condition and performs exactly as it should as long as I do not touch the voltage slider on MSI Afterburner. On my GTX 1080 I had voltage set to +80 so it would hit 1.08v which was the sweet spot for that card's overclock. If I up the voltage at all on this card something happens after a couple minutes (usually really quick... sometimes it takes 30 mins to an hour) and even though it says it is sustaining the clocks and voltage I have set it is rendering at about half power (my FPS is cut in half). This will persist until I reset my PC and then everything is fine unless I mess with the voltage slider. It seems that everyone else overclocking this card has no problem with upping the voltage as long as the temps are fine (which they are on this card as I never go over 74C no matter what I do and am usually around 66C). So, is there something broken on this card? Could it be bad VRMs or something?

    I have no clue what do in order to troubleshoot this. I would love to OC this bad boy with some extra voltage but this card isn't having it. I can put a mild OC on the core and memory (2ghz core and 12ghz memory) without touching the voltage and just sliding the power/temp target to the max, although I want to get the most out of it. Problem is that even though I have the receipt and there is well over a year left on the warranty I just can't see them RMAing the card as it performs perfectly on stock settings while I don't see them replacing it just because I cannot up the voltage without issues. In their eyes the card performs as specified so why would they replace it?

    I have tried flashing the stock BIOS back on the card unless there was any corruption to no avail and do not know if there is another BIOS I could try without bricking the card.

    Has anyone come across this issue before or have any advice for me? Anything would be appreciated.
     
  2. HeavyHemi

    HeavyHemi Ancient Guru

    Messages:
    6,222
    Likes Received:
    571
    GPU:
    GTX1080Ti
    More voltage does not equate to higher clocks or stability. I undervolt using the voltage curve at 1.012v @ 2050mhz. Runs all at day at 2050 mhz below 50C and 2037 above that. I never hit 60C. Temps are key on Pascal, not volts.
     
    Caesar likes this.
  3. Mott

    Mott Master Guru

    Messages:
    816
    Likes Received:
    138
    GPU:
    MSI 1080 Ti GamingX
    Even though that goes against everything I have done with my GTX 1080 I will give it a shot. Will post back with results. Thanks, I hope!

    Edit: Yeah.... no. Anything under 1.05v trying to reach 2ghz on the core and I crash so hard I lose video until a hard reset. It definitely needs more volts and when I give it more the card throttles itself super hard until reset. What the heck is going on?
     
    Last edited: Aug 6, 2019
  4. A M D BugBear

    A M D BugBear Ancient Guru

    Messages:
    2,784
    Likes Received:
    163
    GPU:
    2 1080ti-XOC BIOS
    More then likely the card's tdp is going overboard and thus throttling down and other issues.

    Every tried using other bios, like evga 353w bios???

    I do know that the xoc bios I am using is perfectly stable around 2038/2050 give or take with absolutely no throttling down whatsoever, before it was throttling down like there was no tommorrow.

    2ghz on core, good job friend. I think that's prettie good, although I have seen reports higher but I think 2ghz is good, personally I would be more then happy at that frequency but that would be me.

    First please read an old blog of mine:

    https://forums.guru3d.com/threads/1080ti-bios-mod-very-good.415526/

    link to evga's 358w bios.

    https://www.techpowerup.com/vgabios/201451/evga-gtx1080ti-11264-180309

    I would start doing the cmd feature first before using this bios to be safe.

    here:

    This is a mining blog but still same rules apply:

    https://cryptomining-blog.com/tag/nvidia-tdp-limit-nvidia-power-limit/

    I have used this method first before the evga's bios, wasn't good enough then I choose the xoc bios, never went back but please lower the tdp first before entering the xoc bios zone as XOC bios is absolutely no joke, basically just as powerful as the shunt mod but without damaging the pcb of the card.


    Flashing method + Xoc bios download(Don't use unless others are done):

    https://www.overclock.net/forum/69-nvidia/1627212-how-flash-different-bios-your-1080-ti.html
     
    Last edited: Aug 6, 2019
    Mott likes this.

  5. Mott

    Mott Master Guru

    Messages:
    816
    Likes Received:
    138
    GPU:
    MSI 1080 Ti GamingX
    Hey, thanks a lot for all the information. I will take a look at it and see if it would be beneficial for me. Also, I have been reading a lot more about the 1080 Ti cards and it looks like OC I have now (+100 core and +500 memory) is around the max anyways so maybe it isn't worth the trouble of doing anything extra; I just thought it was super strange that if I exceeded the original voltage the card throttled itself until I rebooted windows. I guess I was worried it had some sort of defect, although now that I can reach the common OC with ease and no throttling maybe I will take it apart and throw some IC Diamond on the core for better cooling (thus voiding my warranty) and not worry about needing to RMA it. I just didn't want to open the card up to improve the thermals like I did with my 1080 if it might need to go through an RMA first.

    Again, thanks for all the info and I will definitely take a look. Maybe the XOC bios will be useful if I can take down the temps a bit with a better paste job. IC Diamond tends to give me a good 5-10C temp reduction depending on the original paste used and how well the job was done. I will report back with what I decide to do and how it worked out when I am finished.
     
  6. A M D BugBear

    A M D BugBear Ancient Guru

    Messages:
    2,784
    Likes Received:
    163
    GPU:
    2 1080ti-XOC BIOS
    Go with thermal Grizzly kyronaut, you will be more then happy, trust me.

    I have looked into the diamond thingy but passed on it, not to say I might never try it but the gain difference compared to grizzly isn't by much from what I understand.

    Personally and this is just me, if its not aio/hybrid or custom loop, I wouldn't go to the xoc route, just to be safe, then again I seen reports that people have no problems but this is just me.
     
    Last edited: Aug 7, 2019
  7. BuildeR2

    BuildeR2 Ancient Guru

    Messages:
    2,827
    Likes Received:
    121
    GPU:
    EVGA 1080 Ti FTW3
    @Mott Dude, consider yourself (or your friend, rather) lucky! I'm on my 4th, yes FOURTH 1080Ti FTW3 from EVGA. First 3 were not even stable out of the box. RMA'd each of them about 4 months apart (paying shipping each time.........) and this 4th one I'm on is finally stable at stock, but won't OC worth crap. I'm just assuming at this point they harvest all the good dies for Kingpin, while leaving the ~1900MHz dies for the FTW3 cards.

    I've tried for days with different drivers, every voltage step for every 12-13 MHz OC, but anything over 1938MHz ends up freezing or CTD for nearly all games. And no, it isn't my PSU or CPU OC. Tried this GPU in 3 different machines and it behaves the same. I've since given up and I'm going the other way, undervolted to 0.950v locked to 1900MHz with a +300 memory clock. It is less than 2% slower than stock clocks, but is about 10C-12C cooler depending on the game. And that is with much lower fan speeds to boot.

    Not buying a 2000 series card, so I will just settle until big Navi or 7nm Turing shrink. :/
     
  8. HARDRESET

    HARDRESET Master Guru

    Messages:
    380
    Likes Received:
    150
    GPU:
    1080Ti G1 GAMING OC
    MSI OC SCANNER works well for me, Gigabyte gaming oc 1080 Ti.
    [​IMG]
     
  9. Caesar

    Caesar Master Guru

    Messages:
    628
    Likes Received:
    213
    GPU:
    GTX 1070Ti Titanium
  10. Mott

    Mott Master Guru

    Messages:
    816
    Likes Received:
    138
    GPU:
    MSI 1080 Ti GamingX
    Well now it is even doing it without touching the voltage limit. After an indeterminate amount of time of gaming it just loses like 15-30% of its power, even though it says it is still at the same clocks, the card benchmarks considerably lower on 3dmark. I reset windows and all is well again.

    I guess I might have to RMA this card... .ugh so annoying. The temps are fine, the clocks are the same and it just kinda dies a little. If it was truly hardware why would restarting windows make a difference? If I let it sit for a while then game again it will still have lost its power but if I reboot it goes back up to where it should be.
     

  11. HeavyHemi

    HeavyHemi Ancient Guru

    Messages:
    6,222
    Likes Received:
    571
    GPU:
    GTX1080Ti
    What are your temps? The first temp bin is all the way down ~40C. In other words at that temp, your clocks will drop. I actually start out at 2063 until the GPU hits ~39C then it drops to 2050. Generally I am under 50C. So I stay pretty solid at 2050.
    Now of course there are games that really bounce around with the power, and on those I will occaisionally hit the power limit. I don't use the voltage slider at all. Strictly voltage curve.
     
  12. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,221
    Likes Received:
    510
    GPU:
    Redacted by NDA
    I had the same problem until I changed my cooler to the Accelero Xtreme III.
    Now it boosts to higher clocks and will take a voltage increase simply due to being cooler.
    (Oh and I used Gelid Extreme paste)
    Lower voltage didnt help my card one bit.
     
    Last edited: Aug 10, 2019
  13. Mott

    Mott Master Guru

    Messages:
    816
    Likes Received:
    138
    GPU:
    MSI 1080 Ti GamingX
    My temps are mid 60s. The temps and clocks don't seem to be a factor in this because I set my OC and it sticks to 2ghz. It starts off at 2012 for a couple seconds then settles into 2000mhz for the duration of my gaming and doesn't budge. the strange part is that after an hour or two of gaming, even though the core is still at 2000mhz and temp hasn't changed, it just drops significantly in performance while all the metrics of the card stay the same; it is not hitting the power limit or any limit for that matter. I can't think of anything else other than sort of broken part that just can't handle the stress. Also, if I try and up the voltage at all it just tanks the performance like 80% down even though it shows higher clocks. Maybe a VRM or something is faulty. Something is definitely faulty. This is unlike any behaviour I have seen on several pascal cards I have overclocked for myself and others.

    Were your clocks lowering with higher temperature? The problem is that my clocks and voltage don't drop; they are steady as a rock and isn't the problem here. I get certain performance with the same clocks and temperatures for about an hour or an hour and a half and then something trips, like a bad capacitor/VRM or something, and then with the same clocks, temps, and voltage I get substantially worse performance until I reset windows then the same clocks, temps, and voltages give me substantially better performance again. Like rebooting windows resets whatever gets tripped up over an hour or two of gaming. It is so strange. I have never seen behaviour like this on a GPU and I have OC'd several GTX 1080s before.
     
  14. BuildeR2

    BuildeR2 Ancient Guru

    Messages:
    2,827
    Likes Received:
    121
    GPU:
    EVGA 1080 Ti FTW3
    Have you tried monitoring via HWinfo, to track your core clocks and the video clocks? Maybe it is doing the built in Pascal core restricting that Buildzoid pointed out, so games and most apps still see 2000MHz core, but the video clock may very well be 1620MHz or so. Just an idea, never had your exact issue. :)
     
  15. Mott

    Mott Master Guru

    Messages:
    816
    Likes Received:
    138
    GPU:
    MSI 1080 Ti GamingX
    Well, I had a flash of genius right before falling asleep tonight (which is why I am still up). I thought about how even though my PSU is single rail that my friend's PC I just built with a 1300w EVGA supernova PSU in it recommended to use two separate cables for the GPU. Now, I specifically remember asking about this exact situation and if it would benefit a GPU if a single rail PSU were to use one cable or two and I do recall the consensus being that as long as it is single rail then one cord should be more than enough. So, I decided to pull another PCI-E power cable out of my Corsair RM850x box and tried running dual lines, one for each 8 pin connector, instead of one line feeding both 8 pin connectors and guess what. I seem to be able to game for as long as I want and also up the voltage and clocks significantly without issue. It seems I was starving the card of power somehow by only using one cord for 16 pins of PCI-E power.

    Does this make sense to anyone? I have not come by a single post advocating for doing this in all the reading I have done on GPUs and power before and after I got this MSI Gaming X card. Whatever is going, though, it is working and I am happy. Looks like no reason to RMA! Hurray!

    Right now I am just running a custom fan curve with MSI Afterburner and the power/temp limit to max and a moderate +500 on the memory. Playing with the core gets me maybe 1-3fps in the real world so I think it just isn't worth it; I don't even see a difference in 3dmark by raising the core. I seem to be steady at 1949 on the core while gaming and I can game for hours with no performance degradation.
     

  16. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,221
    Likes Received:
    510
    GPU:
    Redacted by NDA
    Its been a while since I saw a high power PSU that has 2 PCI-E power connectors on a single cable.
    My 8 year old Corsair AX750 uses 2 cables.
    My new Seasonic Prime has at least 2.
    The builds I have done in between with many PSUs (all good quality) have used 2 cables where they have been needed.
     
  17. Chastity

    Chastity Ancient Guru

    Messages:
    1,825
    Likes Received:
    437
    GPU:
    Nitro 390/GTX1070M
    @Mott If the cable is thick enough to handle the current AND if the PSU doesn't have current limits for each PCIe port (which seems yours does) then you can use 1 cable with 2 8pin connects. :)
     
  18. Mott

    Mott Master Guru

    Messages:
    816
    Likes Received:
    138
    GPU:
    MSI 1080 Ti GamingX
    It looks like I have been woefully ignorant. Strange as I have never had a problem in the past and I am told that a single cable from my RM850x should be able to easily deliver the necessary power. It is strange because it works for some time and then flakes out until a restart. You would think that if the cable couldn't handle it then it wouldn't handle it for an hour or two. Very strange to me although I am really happy that it occurred to me to try out dual cables even though it is a single rail 12v line. It works perfectly now! Also strange that not a single tech support thread ever mentioned that could be the issue. Not even MSI offered that as a troubleshooting step when I was going to RMA it. No one ever asked me or anyone else with similar problems if they were using a single or dual cables for their PCI-E power. I am happy the thought came to me!

    @Chastity how can you see that each cable on my PSU is limited? This information will be invaluable to me in the future. Thanks!
     
  19. HeavyHemi

    HeavyHemi Ancient Guru

    Messages:
    6,222
    Likes Received:
    571
    GPU:
    GTX1080Ti
    You *can if you have to. It it not advisable on a 1080 Ti. They will spike upwards of 300 watts or more OC'd. You will get a voltage drop under load with one cable versus using two. If you're a casual, simply running at default settings, you'd be fine. For over clocking, on a 1080 Ti, I've seen this at least a dozen times from using a single cable: instability.

    Pretty much the same answer applies to your question.
     
  20. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,221
    Likes Received:
    510
    GPU:
    Redacted by NDA
    There is a small voltage drop across any cable length due to its resistance.
    When a cable is pushed hard it heats up which increases its resistance.
    This increases the voltage drop and the current delivered to the card drops too.
    As the cable heats up further this gets worse and is likely the cause of your problem.

    Using 2 cables for the same current halves the overall resistance so there is half the voltage drop initially.
    Because each cable is carrying 1/2 the load it heats up much less, decreasing the increasing voltage drop as well.
    This is why 2 cables at high current are better than one.
     
    Last edited: Aug 13, 2019

Share This Page