Question Thermal limit reached afterburner

[coene]

Hello,
I got an msi 3080 suprim x graphics card, installed the beta version of msi afterburner and maxed my power limit and temp limit (116% and 91°c)

In games and benchmarks i reach 58°c max but still afterburner gives me sudden temp limits (0 to 1) on the temp limit graphic.
I double checked with log files from hwinfo64 and gpuz and the temps never spike up but they both say i reached temp limit at the same time i get the temp limit 1 in afterburner.
I thought i fixed it by going into a folder of afterburner in my c-drive and manually putting in 91°c next to the thermal limit of each profile but that didn't do it.

I never see any throttling happen and everything plays smooth so i am not sure why 3 programs give me temp limit reached when there is not a single temperature value going higher than my temperature limit...

Is this a faulty temp sensor or is there something else i could check ?
Thanks in advance !

 

[McLovin_99]

Hello,
I got an msi 3080 suprim x graphics card, installed the beta version of msi afterburner and maxed my power limit and temp limit (116% and 91°c)

In games and benchmarks i reach 58°c max but still afterburner gives me sudden temp limits (0 to 1) on the temp limit graphic.
I double checked with log files from hwinfo64 and gpuz and the temps never spike up but they both say i reached temp limit at the same time i get the temp limit 1 in afterburner.
I thought i fixed it by going into a folder of afterburner in my c-drive and manually putting in 91°c next to the thermal limit of each profile but that didn't do it.

I never see any throttling happen and everything plays smooth so i am not sure why 3 programs give me temp limit reached when there is not a single temperature value going higher than my temperature limit...

Is this a faulty temp sensor or is there something else i could check ?
Thanks in advance !

I have the same question! Hopefully you are able to find an answer!

 

[coene]

I have the same question! Hopefully you are able to find an answer!

I am trying to find a solution with msi support so hopefully i can figure it out...
Do you also get random blue screens since installing your 3080 ?

 

[McLovin_99]

I have a 3090 and I haven’t had that problem on my end. I just see the GPU temp limit spikes in Afterburner during gaming when the actual temps don’t ever go over 53C. My performance doesn’t seem to be impacted at all. I’m just worried that I’m degrading the chip by hitting temp limits and not sure whether to believe the temp graph or the temp limit graph.

Hopefully MSI support can provide a resolution!

 

[McLovin_99]

I have a 3090 and I haven’t had that problem on my end. I just see the GPU temp limit spikes in Afterburner during gaming when the actual temps don’t ever go over 53C. My performance doesn’t seem to be impacted at all. I’m just worried that I’m degrading the chip by hitting temp limits and not sure whether to believe the temp graph or the temp limit graph.

Hopefully MSI support can provide a resolution!

As an update, I just had HW INFO running at the same time as Afterburner during a gaming benchmark, and HW INFO didn’t show any thermal limits being hit. I’m wondering if this really is just an MSI sensor software issue.

 

[coene]

As an update, I just had HW INFO running at the same time as Afterburner during a gaming benchmark, and HW INFO didn’t show any thermal limits being hit. I’m wondering if this really is just an MSI sensor software issue.

Ok i will try later today to run the log of gpuz, and hw info again to see it my DDU driver install changed something. Still getting temp limit in afterburner tho...
And msi support is no help at all, i gave them all the info i could and they replied to do a ddu, wich i did and told them that my problem is still not fixed.
Their reply to that pissed me off: wish you a nice life.

What kind of stupid reply from a support team is that ?????

So i wrote back that i would like professional help from someone else in their facility, will let you know if this has helped my case when they reply to that.

Maybe you could also try emailing their support with the same problem, who knows you would actually receive some usefull information

 

[McLovin_99]

Ok i will try later today to run the log of gpuz, and hw info again to see it my DDU driver install changed something. Still getting temp limit in afterburner tho...
And msi support is no help at all, i gave them all the info i could and they replied to do a ddu, wich i did and told them that my problem is still not fixed.
Their reply to that pissed me off: wish you a nice life.

What kind of stupid reply from a support team is that ?????

So i wrote back that i would like professional help from someone else in their facility, will let you know if this has helped my case when they reply to that.

Maybe you could also try emailing their support with the same problem, who knows you would actually receive some usefull information

WOW they wished you a nice life?!! That is insane. I’m sorry to hear that. Maybe MSI stands for “Meh, Suck It.” I’ll email them too and see if I can get a friendlier response

 

[coene]

WOW they wished you a nice life?!! That is insane. I’m sorry to hear that. Maybe MSI stands for “Meh, Suck It.” I’ll email them too and see if I can get a friendlier response

After my long email back after the wished me a happy life, i got the following respond from the support team:
Content:
This is normal.

They wrote back to me that my problem with thermal limit is normal. Great job MSI

 

[coene]

WOW they wished you a nice life?!! That is insane. I’m sorry to hear that. Maybe MSI stands for “Meh, Suck It.” I’ll email them too and see if I can get a friendlier response

Sorry for double replying but i have something for you to test also.
Annew afterburner betaversion came out yesterday ( beta.4) on guru3d.
After fully uninstalling en installing afterburner and rivatuner i did stresstests and benchmarks for 2 hours with +130core and +500memory.
The only times my temp limit went from 0 to 1 was when opening or closing a stresstest/benchmark.
It didnt occur during those tests like it did when i had betaversion.2 of afterburner.

I will have to check this during a gamesession but if it only occurs when opening/closing tests then for me it is more acceptable. Will reply here if i tested it some more but for now you could also test betaversion.4 and give me your findings

 

[McLovin_99]

Sorry for double replying but i have something for you to test also.
Annew afterburner betaversion came out yesterday ( beta.4) on guru3d.
After fully uninstalling en installing afterburner and rivatuner i did stresstests and benchmarks for 2 hours with +130core and +500memory.
The only times my temp limit went from 0 to 1 was when opening or closing a stresstest/benchmark.
It didnt occur during those tests like it did when i had betaversion.2 of afterburner.

I will have to check this during a gamesession but if it only occurs when opening/closing tests then for me it is more acceptable. Will reply here if i tested it some more but for now you could also test betaversion.4 and give me your findings

Thanks for this update! I will download that beta version and give it a try. I’m willing to try anything at this point.

 

[coene]

Thanks for this update! I will download that beta version and give it a try. I’m willing to try anything at this point.

Did you get a chance to try this ?

 

[fozzybeat]

greetings folks, dont want to necro this threat but its not too old.
i have a MSI RTX 3090 Suprim X and i have the same behavior.

what is your current status on this.

some research i did found that the drivers from the 456.x do not do those thermal perf-cap/limit spikes.
all rivers i tested since 457.x branch do those spikes.

i stumbled upon this evga forum thread

https://forums.evga.com/m/tm.aspx?m=3124192&p=316

it suggest that those spikes come up on load change.

Question - False Report of Temp Limit

I'm running a Zotac Amp Holo RTX3080 with a reasonable overclock. I have amended fan profile to be a bit more aggressive. For metrics I'm using MSI's afterburner. Even while gaming (example: Call of Duty Cold War, Ultra settings at 1440P), I see my GPU temp evens out in the 60s. However, I...

forums.tomshardware.com

another one with what seems the same behavior.

i also opened multiple web tickets with MSI which are still pending

 

[coene]

greetings folks, dont want to necro this threat but its not too old.
i have a MSI RTX 3090 Suprim X and i have the same behavior.

what is your current status on this.

some research i did found that the drivers from the 456.x do not do those thermal perf-cap/limit spikes.
all rivers i tested since 457.x branch do those spikes.

i stumbled upon this evga forum thread

https://forums.evga.com/m/tm.aspx?m=3124192&p=316

it suggest that those spikes come up on load change.

Question - False Report of Temp Limit

I'm running a Zotac Amp Holo RTX3080 with a reasonable overclock. I have amended fan profile to be a bit more aggressive. For metrics I'm using MSI's afterburner. Even while gaming (example: Call of Duty Cold War, Ultra settings at 1440P), I see my GPU temp evens out in the 60s. However, I...

forums.tomshardware.com

another one with what seems the same behavior.

i also opened multiple web tickets with MSI which are still pending

Hello and thank you for replying and sharing your own research.
I emailed a lot with alternate about this issue and they told me that it's normal when a card makes a load change it could give temp limit reached spikes.
( Did not see these spikes on my gtx 1080 tho.... )

With the newest version of hwinfo i can read my tjunction memory temperature and it never goes above 66°c during a gaming session so i can leave this one out if the sensor of hwinfo is correct

I undervolted my card to .925mv and 1920core clock , temperature never goes over 52°c but yet temp limit spikes, but they always occur when the no-load limit jumps from 1 to 0.
Playing the witcher 3 for a few hours my no-load limit stays almost all of the time at 1 and only spikes to 0 at random moments together with a temp limit spike, and not all of these temp limit spikes are found in hwinfo/gpuz shift log.

A friend of mine has the exact same card and has never seen a temp limit spike during testing/gaming....

The only difference between our cards is the bios version.
On this mather i do not have many experience but perhaps my card was loaded with an early bios version and this causes the temp spikes ?

I hope we can find out why more and more people get these spikes....

 

[coene]

Ah i just asked my mate with the same card, he already updated his bios to the same version i have so that's not it....

Hopefully it's just a driver error like you said with the 456.xx and nvidia will fix this

 

[fozzybeat]

Ah i just asked my mate with the same card, he already updated his bios to the same version i have so that's not it....

Hopefully it's just a driver error like you said with the 456.xx and nvidia will fix this

the last couple of days i am testing drivers again and no 456.xx driver gets those spikes. i even uninstalled MSI Afterburner, RTSS, etc.
ddu of the driver and also tested the dcu driver that windows installs automatically (456 .71)

as soon as i try a newer driver (post 457) i get those spikes again on load change.
i also use HWInfo and GPUz Sensor for monitoring.
while sometimes HWInfo would not trigger the Performance limit thermal to yes, GPUz shows thrm as perfcap reason for the fraction of a second together with Vrel and PWR as well as VOp


MSI support threw me a little of the rails, after explaining my testing they just sent me a one-liner

"in order to save your time, suggest send it for repair service. thanks! "


wtf?!? of course i asked for reasons for such suggestion and if i should be concerned about the health of the card.
no reply yet though.
in a previous reply on the vram temps (hitting 102 during memory torture test) they told me that those temps where safe and all cards are tested

otherwise i have no issues with the card, except Boost clocks not holding 1860 MHZ during Timespy Extreme and dipping down to 1815, no clue if that is normal, timespy normal holds higher clocks than the advertised 1860 but not by much and on others they even go to 2000+ from time to time

in games my clocks always stay above the 1860, i think only RDR2 brings the clock down a notch

not sure what to do, i will wait for msi response, if they give me an explanation on why i should send it for repairs (and what would you repair on a gpu ?!? ) i will get in touch with the local vendor and check for a warranty claim.

i really hope that this is a driver issue, however the new hotfix driver 461.51 broke more than it fixed for me (gb SIV interface stopped working) including the same spikes.

456.xx and 457.xx and later are different driver branches, i didnt test studio drivers yet though

edit:
forgot to ask, would you be willing to test the 456 driver as well? just ddu and clean the old driver and let windows install automatically.

btw. im on 1909 with no hardware scheduling, 2004 gave me too large fps dips

 

[coene]

Yeah msi support was of no help to me also as you can read in my previous posts here...

I will do the testing with doing ddu and install a 456.xx driver, i am on windows 20h2 just fyi.

Tomorrow/day after i will have the results of the driver and will reply here.

I have very good scores on 3dmark with my gpu and i dont see fps dips caused by this temp limit so my conclusion is that it is not harmfull for the gpu but only software/driver related. Did enough time with those logs of afterburner/gpuz/hwinfo and never have i seen something that worries me

BUT those goddamn temp limit spikes need to go away.... XD

 

[fozzybeat]

i will wait for your results, but i expect that you will have the same findings

as for MSI, i work in tech support and if we would work like that .... damn that would not end well for us.
subpar is an understatement, and that for their flagship products ....

i have two open tickets, one via tech support and the other regarding the warranty as i wanted to know the production date of my card, if its from before or after the delivery delay due to the recall, installed bios is from end of november, so after the delay.

will post updates here, and yes those spikes are the sole reason i am not 100% happy with my product.

 

[coene]

Maybe just a few questions:

1. Did you try it with both gaming and silent bios ?

i only tried the gaming bios, gonna first try the silent bios and then will do a DDU and use driver 456.71.

2. What type of PSU are you using, maybe there is an issue with the voltages going from PSU to the GPU.
I had a corsair 650x for my GTX 1080 and never saw temp limit spike.

I bought my 3080 along with a new PSU corsair RM1000x so the only 3 items from my pc that have changed are those 2 + added an M2 SSD.

3. What beta-version are you using of MSI afterburner ? I still use beta.4 but i saw there is a beta.5, maybe this changes something for the frequency of the temp limit spikes.
But gpuz and hwinfo still record the temp limits yet less frequently so i think we can exclude that it's only an issue with msi afterburner..

 

[fozzybeat]

Maybe just a few questions:

1. Did you try it with both gaming and silent bios ?

i only tried the gaming bios, gonna first try the silent bios and then will do a DDU and use driver 456.71.

2. What type of PSU are you using, maybe there is an issue with the voltages going from PSU to the GPU.
I had a corsair 650x for my GTX 1080 and never saw temp limit spike.

I bought my 3080 along with a new PSU corsair RM1000x so the only 3 items from my pc that have changed are those 2 + added an M2 SSD.

3. What beta-version are you using of MSI afterburner ? I still use beta.4 but i saw there is a beta.5, maybe this changes something for the frequency of the temp limit spikes.
But gpuz and hwinfo still record the temp limits yet less frequently so i think we can exclude that it's only an issue with msi afterburner..

a few answers

1. nope i didn't bother with the silent bios

94.02.42.00.F5 vBios Date: 2020-11-23

this seems to be the production bios, however their so called recall on the suprim X cards was around the 27th of november

2. its a BeQuiet Dark Power Pro 11 1200w, 4x12v Rails, 12V1/2(CPU/Mobo/disks etc) 35 Amps each Rail, 12V3/4 (pcie) 45 Amps each Rail, GPU is connected to two 12V rails, according to GPU Vendor Specs, i ahve a DPP 11 850w as well with the same 12v Rail config, however 35Amps on the 2 pcie rails, maybe i will test it

3. i was using the latest stable to be honest but updated the RTSS to the latest beta, however for my testing i removed MSI Afterburner and RTSS completely and used only the GPUZ sensor readings to check for the flag trigger and HWINFO

i started another thread over on overclock dot net

https://www.overclock.net/threads/p...n-load-change-rtx-3090.1776791/#post-28735777

a guy seems to confirm the driver bug suspicion and mentions that EVGA FTW3 (evga has good temp monitoring build in to their cards) registers a erroneous temp reading of 6500C

so digging further with that new pointer i found the forum thread regarding this

https://forums.evga.com/Erroneous-S...d-Unusual-Fan-Behavior-In-GTA-5-m3192504.aspx

fwiw i will go back installing 641.40 and Afterburner/RTSS for overlay and wait for MSI response for the time being.

 

[coene]

a few answers

1. nope i didn't bother with the silent bios

94.02.42.00.F5 vBios Date: 2020-11-23

this seems to be the production bios, however their so called recall on the suprim X cards was around the 27th of november

2. its a BeQuiet Dark Power Pro 11 1200w, 4x12v Rails, 12V1/2(CPU/Mobo/disks etc) 35 Amps each Rail, 12V3/4 (pcie) 45 Amps each Rail, GPU is connected to two 12V rails, according to GPU Vendor Specs, i ahve a DPP 11 850w as well with the same 12v Rail config, however 35Amps on the 2 pcie rails, maybe i will test it

3. i was using the latest stable to be honest but updated the RTSS to the latest beta, however for my testing i removed MSI Afterburner and RTSS completely and used only the GPUZ sensor readings to check for the flag trigger and HWINFO

i started another thread over on overclock dot net

https://www.overclock.net/threads/p...n-load-change-rtx-3090.1776791/#post-28735777

a guy seems to confirm the driver bug suspicion and mentions that EVGA FTW3 (evga has good temp monitoring build in to their cards) registers a erroneous temp reading of 6500C

so digging further with that new pointer i found the forum thread regarding this

https://forums.evga.com/Erroneous-S...d-Unusual-Fan-Behavior-In-GTA-5-m3192504.aspx

fwiw i will go back installing 641.40 and Afterburner/RTSS for overlay and wait for MSI response for the time being.

did my DDU and used the driver version 456.38 and tadaa no more temp limit spikes.

i guess we have to wait for nvidia to pick it up and come with a new driver....

still find it odd that my mate with the latest driver version does not have those tmp limit spikes....
Or maybe he does have them but is not doing the tripple HWinfo/gpuz/afterburner check like we do

he plays shooters 99% of the time so maybe there is not much load change going on there, i play RPG's most of the time and going in or out of a cutscene gives me a frequent temp limit spike due to the load change

 

[fozzybeat]

did my DDU and used the driver version 456.38 and tadaa no more temp limit spikes.

i guess we have to wait for nvidia to pick it up and come with a new driver....

still find it odd that my mate with the latest driver version does not have those tmp limit spikes....
Or maybe he does have them but is not doing the tripple HWinfo/gpuz/afterburner check like we do

he plays shooters 99% of the time so maybe there is not much load change going on there, i play RPG's most of the time and going in or out of a cutscene gives me a frequent temp limit spike due to the load change

i am running 641.40 again and sometimes Afterburner does not register a spike when HWinfo does...
guess what ! MSI still didn't answer

 

[coene]

i am running 641.40 again and sometimes Afterburner does not register a spike when HWinfo does...
guess what ! MSI still didn't answer

I was playing around and setting -300 for the core clock when not running any game or benchmark, hwinfo registered a temp limit but afterburner did not. Same happened when going back to stock and doing -300 again.
Temperature was 35°c at that time.

So even with a sudden load change when not using the gpu it will give a temp limit... This gives me more reason to think it is indeed software/driver related than an actual problem. But still wish there was no temp limit at all when not exceeding the set temperature limit

 

[fozzybeat]

I was playing around and setting -300 for the core clock when not running any game or benchmark, hwinfo registered a temp limit but afterburner did not. Same happened when going back to stock and doing -300 again.
Temperature was 35°c at that time.

So even with a sudden load change when not using the gpu it will give a temp limit... This gives me more reason to think it is indeed software/driver related than an actual problem. But still wish there was no temp limit at all when not exceeding the set temperature limit

i dont remember if you said this but when you registered the tempspike in the Afterburner Graphs, did you also notice a drop in the Power % and wattage graph ? i can also see a sudden spike in the FPS and a drop of the GPU usage.

 

[coene]

i dont remember if you said this but when you registered the tempspike in the Afterburner Graphs, did you also notice a drop in the Power % and wattage graph ? i can also see a sudden spike in the FPS and a drop of the GPU usage.

Your gpu usage drops to 2% ? was this during a quick alt-tab or maybe doing something on a second sceen ?
But your fps changes from 60-ish to 256 so perhaps you died in your game resulting to a much lower gpu usage and higher FPS ?
If this happens during your testing and it drops from 100% to 2% with nothing special happening in your gaming/testing then there is indeed something wrong....
i am currently at work but tonight i will check this again.
( i say again because i already deleted my previous logfiles of afterburner -_- )

 

[fozzybeat]

Your gpu usage drops to 2% ? was this during a quick alt-tab or maybe doing something on a second sceen ?
But your fps changes from 60-ish to 256 so perhaps you died in your game resulting to a much lower gpu usage and higher FPS ?
If this happens during your testing and it drops from 100% to 2% with nothing special happening in your gaming/testing then there is indeed something wrong....
i am currently at work but tonight i will check this again.
( i say again because i already deleted my previous logfiles of afterburner -_- )

i honestly dont remember, i believe this to be alt tabbing out, but i kind of lost track of everything i did.

yesterday i called bequiet support, doublechecking if i connected the gpu correctly, i did. it is connected to rail3 and rail4 while the connection to rail 4 is via split cable, i specifically asked if it would make sense to connect a third cable to the psu but he declined as it would still use the same 12v rail,
my card would need with its 450w 37.5 amps and each rail delivers 45.

anyway the support guy mentioned that i can switch my psu to single rail mode (99 amps with 107 peak) so i am currently running like that.

i dont know if its placebo effect but rdr2 seems to stutter less that way, but now i read that stutters in that game are more related to engine issues .... go figure.

i didnt have issues in other games anyway so for the sake of being protected i will probably switch back as OCP is just on the singel rail in that mode while in multirail mode its on each.
the new model keeps the per rail ocp in its single rail mode (dark power pro 12).
running my psu like that is still safe according to the support tech and he seemed quiet competent (i work tech support too and can smell incompetence, at least id like to think so ~~)

i am drifting off topic now .... i will continue monitoring