Question Drastic GPU Stuttering 100% GPU on ASUS 3080 Ti

[dizzyjay]

Hi all,

I'm trying to find alternative solutions to my computer stuttering while playing games. Are there any simple solutions I can utilize? (Solutions/Attempts, Build below)

Games such as Heroes of the Storm, or Wartales, both of which I don't believe should be as GPU-intensive, but I'm starting to notice a drastic stuttering problem while playing.
For example, playing Heroes of the Storm (Blizzard's MOBA game), I'd sometimes be waiting at the fountain and my whole computer freezes anywhere between 0.1 seconds to 2 seconds -- I've noticed this is when GPU usage goes up to 100%.
Similar issue with Wartales (turn-based combat).

When running with a game, my GPU temps can go to 90F (32C).

1. Solution (Attempt #1): Additional fans, total of 6 case fans which are 3 intake and 3 exhaust.
2. Solution (Attempt #2): Underpower GPU: I've forced the GPU to stay at 82F (27C) via MSI Afterburner.
3. Solution (Attempt TBD #3): I'm actively looking for a bigger case to help with air flow and possibly research getting additional air cooling for the GPU.
4. Solution (Attempt TBD #4): Switch out the CPU cooler with a smaller one to move the GPU up. (Top PCI-E slot is blocked by GPU cooler)
5. Solution (Attempt TBD #5): Try to install NZXT Kracken G12 with an AIO LC system onto the GPU. This solution requires modding the G12 to fit the 3080 Ti. (YT Video)

Also, an annoying problem, but I hear my GPU rattle a little bit when things start getting "intense" (high GPU usage) for it.



Here's my current build:

• Motherboard: ASUS ROG Strix Z590-E Gaming Wifi
• Processor: Intel i9-10900K
• Cooling: Noctua NH-U14S 140mm Fan
• GPU: ASUS ROG Strix 3080 TI
• PSU: Seasonic FOCUS GX-1000W 80+ Gold Full-Modular
• RAM: 32GB RAM: G.SKILL TridenZ 16GB (2x8) 288-pin DDR4 4400 (x8 Module)
• SSD: 2TB WD M2 SSD
• Internal Fans: Noctua NF-12 3000 PWM 120mm, 3x Noctua NF-P12 1700 PWM 120mm
• Monitor: Samsung Odyssey G9
• Case: Fractal Design Meshify C White (I'm looking to get a bigger case, as I believe some additional air flow will help)

Software:

• Windows 10
• BIOS at v0232
• GeForce Experience with latest GPU drivers

 

[Phaaze88]

"When running with a game, my GPU temps can go to 32C."
Use Celsius when discussing thermals of electronic devices.
That's really bloody good if that's the hottest the gpu core ever sees...


"Solution (Attempt #2): Underpower GPU: I've forced the GPU to stay at 27C via MSI Afterburner."
You'd be crippling the card pretty hard if you actually did that. The card would have no room to run at base clock, much less auto OC itself.
The hardware is fine, even when running at 80C.


"... GPU usage goes up to 100%."
What causes 100%?
-Cpu, resolution and game settings legit cause the gpu to give its all.
-Got a miner, or some other gpu dependent app taking up gpu resources alongside the game(s).
Gpu is running slower:
-Thermal throttling. The creator of that YT video does not show Vram thermals, and GDDR6X memory on the 3070Ti and up run very hot without proper cooling, even though it uses very little power - like, 3-6w a chip, or something like that.
-The gpu is power throttling; you can observe very low core clock when this happens. Loose psu cable, or VRM thermal throttling.


Suggestions:
-Reevaluate what's actually hot for the hardware. Gpu core: 83C+. Gpu hotspot: 100C+. Memory: 95C+.
-Monitor Vram thermals in game. Hwinfo or Gpu-Z are your friends for this. Possible the G12 mod was worse than the original cooler for this.
-Monitor gpu core clock behavior.
-Good ol' Task Manager Processes, and see what else might be using up the gpu.

 

[dizzyjay]

"When running with a game, my GPU temps can go to 32C."
Use Celsius when discussing thermals of electronic devices.
That's really bloody good if that's the hottest the gpu core ever sees...

Oh, my fault, I mixed it up. Thanks for noticing -- I did mean to write 90C.

My computer on standby (basic apps like chrome, steam, discord) runs at 48C.
When playing games it can go up to 90C.

"Solution (Attempt #2): Underpower GPU: I've forced the GPU to stay at 27C via MSI Afterburner."
You'd be crippling the card pretty hard if you actually did that. The card would have no room to run at base clock, much less auto OC itself.
The hardware is fine, even when running at 80C.

I've removed the underpowering setting for now.

"... GPU usage goes up to 100%."
What causes 100%?
-Cpu, resolution and game settings legit cause the gpu to give its all.
-Got a miner, or some other gpu dependent app taking up gpu resources alongside the game(s).
Gpu is running slower:
-Thermal throttling. The creator of that YT video does not show Vram thermals, and GDDR6X memory on the 3070Ti and up run very hot without proper cooling, even though it uses very little power - like, 3-6w a chip, or something like that.
-The gpu is power throttling; you can observe very low core clock when this happens. Loose psu cable, or VRM thermal throttling.

For the record, I do not have any mining programs on my computer. It's strictly intended for gaming.
I will run some tests further below.

Suggestions:
-Reevaluate what's actually hot for the hardware. Gpu core: 83C+. Gpu hotspot: 100C+. Memory: 95C+.
-Monitor Vram thermals in game. Hwinfo or Gpu-Z are your friends for this. Possible the G12 mod was worse than the original cooler for this.
-Monitor gpu core clock behavior.
-Good ol' Task Manager Processes, and see what else might be using up the gpu.

Thank you for the feedback!

I had CPU-Z, but I've installed GPU-Z now to test it out and include logging.


Setup:

• Running GPU-Z
• Running HWInfo
• Running MSI Afterburner
• Windows Task Maanger confirms everything listed here is what will generally take up the resources.
• Basic apps that'll be running in the background:
  • Steam
  • Battle.net
  • Discord
  • Chrome (Eating up ~6GB of RAM, but I'll leave it on as it's a likely app to keep running when gaming)
  • Kasperksy Antivirus

For this test I've removed the following apps from running:

• Wallpaper Engine (this seems to have been taking 20% GPU)

Games will be running on HIGH / VERY HIGH / ULTRA settings where applicable (I'll try to document settings before trying it out).

• Wartales
• Heroes of the Storm
• Baldur's Gate 3
• Counter-Strike GO
• Mount & Blade II: Bannerlord

Additional settings games will be running:

• Screen Resolution 5120x1440 (native display settings)

I will report back once I have sufficient data, it may take a few days as it's already 11:30PM where I'm at.

 

[dizzyjay]

I will report back once I have sufficient data, it may take a few days as it's already 11:30PM where I'm at.

I ran against two games so far. I'm not sure how to make heads/tails of the information while peeking at GPU Clockspeeds. Google Spreadsheet of HWInfo of Heroes of the Storm run.

• Bauldur's Gate 3
  • Did not notice stuttering, but some objects were missing (floor) on one level, but could be because the game is in development still. (never had a problem with it before in previous software builds)
  • Ran ~1hr test
  • Ultrawide, 5120x1440 @ 60hz
  • Fullscreen
  • All settings enabled / set to Highest possible
• Heroes of the Storm: 1 game just lasted 15min and I stopped
  • Heavily stuttered 4 or so time during gameplay
    • About ~1min into actual gameplay
    • ~10min into gameplay
    • ~15min into gameplay
  • Ran ~30mn test
  • Ultrawide, 5120x144 @ 60hz
  • Fullscreen
  • All settings enabled / set to High (NOT Ultra / Extreme), just "High"

I actually only noticed stuttering during Heroes of the Storm gameplay, which is starting to lead me to believe that the lack of DEV support on it may be part of the reason my current setup may be stuttering (ultrawide?). I will try again later with lower settings -- watching the replay of the game did not show any sign of stuttering.

Will post more when I get to it, in the meantime, any feedback is appreciated -- but I may chop this up to old/incomplete games without direct support of Ultrawide settings.

 

[dotas1]

Did you try with non ultrawide resolutions?

 

[Phaaze88]

Oh, my fault, I mixed it up. Thanks for noticing -- I did mean to write 90C.

My computer on standby (basic apps like chrome, steam, discord) runs at 48C.
When playing games it can go up to 90C.

OK. 90C is too high. Gpu should be thermal throttling. You gotta get that down.
Hotspot can't be much better, as it is normal for it to be 10-20C hotter.

Chrome (Eating up ~6GB of RAM, but I'll leave it on as it's a likely app to keep running when gaming)

Browsers have the gpu acceleration option, which will screw with some games. Would be a good idea to disable it if it is enabled and the browser is running in the background.

Wallpaper Engine (this seems to have been taking 20% GPU)

Good move. That app is unusually resource hungry; some users have reported lower performance when running this.

Google Spreadsheet of HWInfo of Heroes of the Storm run.

In Nvidia Control Panel, Manage 3D Settings, is power management on Optimal(it's the default setting)? I see gpu core clock is all over the place - that tends to happen in games that don't need much gpu muscle, and with the Optimal power plan, it's going to clock down to save more power.
Set it to Max Performance and see if it helps.

This might be a false alarm, but there was a very brief 'Performance Limit - Thermal' warning at column OE, Line 156. When I scroll across to view gpu thermals, there's nothing particularly alarming there.
Nothing really stands out at the ~10 and 15min marks either.

 

[dizzyjay]

Did you try with non ultrawide resolutions?

Today while trying this out with Heroes of the Storm again. Same monitor (Samsung G9 49", ultrawide)
I tried non-ultrawide resolutions, 1600x900, 1920x1080 and a mix of others, they all gave some level of lag (mind you, it was difficult to play as I couldn't zoom out anymore lol)

OK. 90C is too high. Gpu should be thermal throttling. You gotta get that down.
Hotspot can't be much better, as it is normal for it to be 10-20C hotter.

Right, this is where I setup the initial throttle of 82C.

Browsers have the gpu acceleration option, which will screw with some games. Would be a good idea to disable it if it is enabled and the browser is running in the background.

I've disabled GPU Acceleration now and from running in the background if closed.

In Nvidia Control Panel, Manage 3D Settings, is power management on Optimal(it's the default setting)? I see gpu core clock is all over the place - that tends to happen in games that don't need much gpu muscle, and with the Optimal power plan, it's going to clock down to save more power.
Set it to Max Performance and see if it helps.

I'm looking at NVIDIA Control Panel -> Manage 3D settings -> "Global Settings" -> "Power Management Mode" is set to "Normal" (the other option is Prefer Maximum Performance)
I will set this to "Prefer Maximum Performance". I don't see any other "Power Management" settings, so I believe that's the one you're saying.

This might be a false alarm, but there was a very brief 'Performance Limit - Thermal' warning at column OE, Line 156. When I scroll across to view gpu thermals, there's nothing particularly alarming there.
Nothing really stands out at the ~10 and 15min marks either.

What can be done here? The other day I was debating if I could possibly return my ASUS 3080ti with the manufacturer's warranty, but I still need to read through the stipulations.


I've ran another test, and this time with lower settings, then changing it back to higher settings mid-game.
Initial settings
Start @ 7:25PM
1600x900; Medium settings
At my first death in-game at 7:43 (not sure exact seconds here), I experienced heavy lag to the point I couldn't move and just died during a fight while (once things cleared up, it appeared I just stood there).
Results can be found here -- I've also highlighted (fill color red) the time I noticed this, too.


Lastly, I've ordered a CPU AIO LC by MSI, I will be replacing my current CPU cooling system with this so that I may move my GPU up a PCI-E slot.
Here's an image of the current placement of the GPU (Imgur link); I think this adds to the problem of generating too much heat, but I'll be testing it out tomorrow night when I obtain the replacement.

I will perform more runs with this game specifically with different settings and the aforementioned suggestions.
Also, I'll try to do this while running against bots so I don't ruin anyone else's game by lagging.


Thank you again for the feedback!

 

[Phaaze88]

I'm looking at NVIDIA Control Panel -> Manage 3D settings -> "Global Settings" -> "Power Management Mode" is set to "Normal" (the other option is Prefer Maximum Performance)
I will set this to "Prefer Maximum Performance". I don't see any other "Power Management" settings, so I believe that's the one you're saying.

Probably just my version of NVCP, or the fact that I'm on Pascal. They got rid of Adaptive Mode from the 3 options, from the looks of it; it didn't always work anyway.

What can be done here? The other day I was debating if I could possibly return my ASUS 3080ti with the manufacturer's warranty, but I still need to read through the stipulations.

I don't know, I've not seen that before.
Can you even return it at this point, since you've taken it apart and even tried a DIY AIO mod with it? They CAN deny warranty for that, if they discover any traces.

Results can be found here -- I've also highlighted (fill color red) the time I noticed this, too.

/shakes head
I'm not seeing anything separating the red zone from the rest of the chart - nothing's popping in my face like, "Yo, this is bad, something needs to be done here..."

Lastly, I've ordered a CPU AIO LC by MSI, I will be replacing my current CPU cooling system with this so that I may move my GPU up a PCI-E slot.
Here's an image of the current placement of the GPU (Imgur link); I think this adds to the problem of generating too much heat, but I'll be testing it out tomorrow night when I obtain the replacement.

1)Hopefully, you get the revised model, if you ordered a MAG or MPG. They have a problem with contaminated coolant and clogging early. MEG doesn't have that problem.
2)In the meantime, why don't you test with the U14S' heatsink turned 90 degrees, so you can install the gpu in the top slot?

 

[dizzyjay]

Can you even return it at this point, since you've taken it apart and even tried a DIY AIO mod with it? They CAN deny warranty for that, if they discover any traces.

I haven't modded anything yet, I like to research before I go ahead and take things apart; hence my troubleshooting here to see what I could do to alleviate the heating problems. The biggest hurdle with modding it was simply the tools needed, I don't have any of it to cut steel and mount a G12 with AIO.
So, the manufacturer's warranty should still be available at least. I purchased this one via Newegg, so it's well passed the alloted time for returns.

/shakes head
I'm not seeing anything separating the red zone from the rest of the chart - nothing's popping in my face like, "Yo, this is bad, something needs to be done here..."

Thank you for looking. Neither did I see anything that popped out myself.

1)Hopefully, you get the revised model, if you ordered a MAG or MPG. They have a problem with contaminated coolant and clogging early. MEG doesn't have that problem.
2)In the meantime, why don't you test with the U14S' heatsink turned 90 degrees, so you can install the gpu in the top slot?

(Links below are via Amazon)
I hope do I get a revised model. But this is the one I ordered, which seems to be a MAG.
I looked to see the difference between MAG and MEG, and it's about $100.

In terms of moving the U14S heatsink, I tried but it initially blocked the RAM installations.. since it's so bulky, I'm probably going to get rid of it in place of the MSI to have a cleaner finish . (and possibly better airflow? There'd be a big fan missing in the middle of the rig. xD)
I actually debated using the NZXT stuff while I was researching components and noticed some AIO can also block the RAM installations.

Lastly, I was looking at possibly filling the bottom PCIE slot if the above heating issues persist.

1. OPTION #1: Fan PCIE installation
   1. This seems like a fairly lower-cost solution to increase air-flow.
2. OPTION #2: Upgrade my case to a bigger mid-size that supports vertical mounting and have the GPU vertically mounted.
   1. This option seemed a little more excessive and possibly risky since my research kept leading me to unknown factors on a proper vertical mount that will support PCI-E 4.0 GPUs. Everything I've been finding has been PCI-E only slots and I can't tell whether or not I can replace the actual slot with a PCI-E 4.0 replacement. I feel like it's a fire hazard waiting to happen.

Thank you, and everyone else that has taken the time to go through this with me. Again, I'll try to post back when I have the updated parts installed and run a couple more tests.

 

[Phaaze88]

I haven't modded anything yet, I like to research before I go ahead and take things apart; hence my troubleshooting here to see what I could do to alleviate the heating problems. The biggest hurdle with modding it was simply the tools needed, I don't have any of it to cut steel and mount a G12 with AIO.

My mistake. I misread: Solution (Attempt TBD #5): Try to install NZXT Kracken G12 with an AIO LC system onto the GPU. This solution requires modding the G12 to fit the 3080 Ti.

I looked to see the difference between MAG and MEG, and it's about $100.

They have different OEMs too. Most companies don't make their own - another company does.
MAG and MPG are made by Apaltek. MEG was done by Asetek.

Lastly, I was looking at possibly filling the bottom PCIE slot if the above heating issues persist.

1. OPTION #1: Fan PCIE installation
   1. This seems like a fairly lower-cost solution to increase air-flow.
2. OPTION #2: Upgrade my case to a bigger mid-size that supports vertical mounting and have the GPU vertically mounted.
   1. This option seemed a little more excessive and possibly risky since my research kept leading me to unknown factors on a proper vertical mount that will support PCI-E 4.0 GPUs. Everything I've been finding has been PCI-E only slots and I can't tell whether or not I can replace the actual slot with a PCI-E 4.0 replacement. I feel like it's a fire hazard waiting to happen.

1)The air still needs to come from somewhere. The PCIe fan can be blocked in the same manner as the gpu's fans.
2)That card should be backwards-compatible with PCIe 3.0 risers. Compatibility with 4.0 has been a bit iffy with Ryzen systems, requiring users to manually set the link speed.
4.0 doesn't matter for performance at the moment - unless one is using a 5500XT or 6500XT - it's just going to take time for that to be useful for most. By the time that happens, 6.0 might be on the way - who knows...

3)Fire hazard is news to me. The last, and only thing I'd observed regarding fire hazards with risers is that mess with NZXT's H1.

 

[KyaraM]

Quick question. Did you try undervolting the GPU yet to reduce temps? My 3070Ti never gets above 80°C, and hitting that is super rare, too. Most of the time it's running at 60-70°C core temperature and about 9-11°C higher for hot spot, with VRAM directly in the middle. It's also overclocked at the same time.

 

[drivinfast247]

Lastly, I've ordered a CPU AIO LC by MSI, I will be replacing my current CPU cooling system with this so that I may move my GPU up a PCI-E slot.
Exere's an image of the current placement of the GPU (Imgur link); I think this adds to the problem of generating too much heat, but I'll be testing it out tomorrow night when I obtain the replacement.

I will perform more runs with this game specifically with different settings and the aforementioned suggestions.
Also, I'll try to do this while running against bots so I don't ruin anyone else's game by lagging.


Thank you again for the feedback!

Is that GPU in PCIEX_2 or PCIEX_3?

You should have clearance to use it in the top slot PCIEX_1.

 

[dizzyjay]

Quick question. Did you try undervolting the GPU yet to reduce temps? My 3070Ti never gets above 80°C, and hitting that is super rare, too. Most of the time it's running at 60-70°C core temperature and about 9-11°C higher for hot spot, with VRAM directly in the middle. It's also overclocked at the same time.

Hello, yes, I have. Undervolting / underpowering the GPU was my second attempt at resolving the issues. See below quote:

Solution (Attempt #2): Underpower GPU: I've forced the GPU to stay at 82F (27C) via MSI Afterburner.

Is that GPU in PCIEX_2 or PCIEX_3?

You should have clearance to use it in the top slot PCIEX_1.

I believe this is in the PCIEX 2 (middle slot) which is a PCIE 4.0. See: (imgur, official asus website with the imgur image in the middle)
Providing the image again, here's my current rig (imgur) where a massive U14S heatsink sits that blocks the PCI-E slot #1 closest to CPU. Previous user, Phaaze88 suggested the same, but until I recieve a new CPU AIO LC system (tomorrow afternoon hopefully), I won't be able to properly sit the GPU in PCI-E slot #1. I know I can easily remove the RAM sticks, but I also enjoy looking at my build and that possibly having a large heatsink in the center of the computer isn't the prettiest of things, so I'm just going to replace it.

1)The air still needs to come from somewhere. The PCIe fan can be blocked in the same manner as the gpu's fans.
2)That card should be backwards-compatible with PCIe 3.0 risers. Compatibility with 4.0 has been a bit iffy with Ryzen systems, requiring users to manually set the link speed.
4.0 doesn't matter for performance at the moment - unless one is using a 5500XT or 6500XT - it's just going to take time for that to be useful for most. By the time that happens, 6.0 might be on the way - who knows...

3)Fire hazard is news to me. The last, and only thing I'd observed regarding fire hazards with risers is that mess with NZXT's H1.

I saw multiple reviews / an image that spooked me of a fried GPU (amazon link). Funnily enough, they do mention NZXT H* models which I had also planned to get, but will hold back until I try the other suggestions / ideas first.

 

[KyaraM]

Hello, yes, I have. Undervolting / underpowering the GPU was my second attempt at resolving the issues. See below quote:

Nah, thatnot what I mean. You just told the card to not draw more than 80% (for example) of what it could according to its TDP. That reduces performance and while helping with temperatures isn't optimal. I ask if you set a hard limit for the voltage the card can run at.

For example, my 3070Ti normally uses up to 1.065V. However, I locked it to 0.975V while also increasing clock speed to 2025MHz top. Now, the card runs at a stable 2010MHz OC (or 1995MHz, in case of less demanding games that run into my screencap of 144FPS since running that's my monitor's refresh rate anyways), capped at the specified voltage. Stock, it goes up to 1960MHz with the same or higher power draw. It's running cooler than with an OC without undervolt (or simply a card that wasn't optimized, for that matter...) and faster as well. It also never gets to the specified TDP of 290W and stays at around 230-240W. This isn't even an optimized OC by any means. I probably can increase it, but I'm quite happy with how it is so I will leave it at that. Undervolting works without OC, of course.

 

[dizzyjay]

Nah, thatnot what I mean. You just told the card to not draw more than 80% (for example) of what it could according to its TDP. That reduces performance and while helping with temperatures isn't optimal. I ask if you set a hard limit for the voltage the card can run at.

For example, my 3070Ti normally uses up to 1.065V. However, I locked it to 0.975V while also increasing clock speed to 2025MHz top. Now, the card runs at a stable 2010MHz OC (or 1995MHz, in case of less demanding games that run into my screencap of 144FPS since running that's my monitor's refresh rate anyways), capped at the specified voltage. Stock, it goes up to 1960MHz with the same or higher power draw. It's running cooler than with an OC without undervolt (or simply a card that wasn't optimized, for that matter...) and faster as well. It also never gets to the specified TDP of 290W and stays at around 230-240W. This isn't even an optimized OC by any means. I probably can increase it, but I'm quite happy with how it is so I will leave it at that. Undervolting works without OC, of course.

I've not done this as I'm not entirely certain how to do it. Do you have a reference guide of how to do this? When I start searching online, I get mixed results. My primary concern here is the risk of bricking the GPU.

They have different OEMs too. Most companies don't make their own - another company does.
MAG and MPG are made by Apaltek. MEG was done by Asetek.

** UPDATE **

• Received AIO CPU LC
  • It's too large for the case; I realize the specs were JUST right, but there won't be enough clearance to actually put it due to case design.
• Moving around CPU heatsink
  • I looked at this again and it appears to be impossible given the layout of the motherboard.
    • Image of ASUS MOBO from official website: imgur link
    • Image of my ASUS MOBO: imgur link
    • Specifically look at the CPU mount and how close the RAM slots are, and the clearance above and to the left of it. There's something sticking out which I'm not certain I can remove. This motherboard appears to be intended for smaller heatsinks.

On that note, I have decided to go ahead and increase the size of the case with a newer case that I should receive sometime April 4-6 (hopefully sooner, but never know).
The case I purchased to provide myself a bigger clearance space for now is the Fractal Design Torrent; which I also hope will provide enough air flow to help keep the GPU cooler. I may still remove the NOCTUA U14S heatsink, but will keep it there for now and test out the temperatures, then switch it out with the AIO LC, my only concern with doing this is how impractical it is to mount and dismount a CPU repeatedly, but alas it may come down to it if I need more clearance for the GPU.
My hopes with this new case is

• Provide more clearance for the GPU to stay cool
• Allow me the ability to swap out the NOCTUA U14S heatsink and put in place a AIO LC system for the CPU
  • Which in turn should allow me to mount the GPU up a PCI-E slot closest to the CPU.
• The design of the Torrent appears to provide useful cooling for the GPU and CPU alone, so I may no longe rneed the AIO, but we will see how it goes!

My next update, barring any other suggestions here that I can try on my own will be when I receive the Torrent case.

Thank you all so far for the input provided here, it's been invaluable.

 

[KyaraM]

I've not done this as I'm not entirely certain how to do it. Do you have a reference guide of how to do this? When I start searching online, I get mixed results. My primary concern here is the risk of bricking the GPU.

I can't really see how providing less voltage would fry anything. All you do is lowering voltage, not increase it (which really can damage the card). Worst case, your system crashes. That's it.

This was the first I found about it myself:

RTX 3070 Ti Undervolting - Can we finally teach the drinkiest of Ti’s some manners? | Practice | igor´sLAB

Disclaimer: The following article is machine translated from the original German, and has not been edited or checked for errors. Thank you for understanding!

www.igorslab.de

However, I ended up doing it slightly differently by running the auto OC function of MSI Afterburner over my GPU curve, then clipping the curve at my target voltage and then tinkering a bit from there on. I would have to get home before I can explain it better. I simply wanted a stable undervolt, my goal wasn't to find the maximum undervolt possible. Still, it's 90mV lower than before even so, but I can easily dip deeper at the same frequencies I currently have, and it reduced temps and power draw so...

What I would do in your case, though, is test how high your card clocks and use that as a base for your undervolt target.

 

[dizzyjay]

Today I got my Fractal Torrent case early.

I have been testing this out for just about an hour with two games, Wartales & Heroes of the Storm.

1. As I further googled, I found that NVIDIA Experience's Overlay recently caused problems for Heroes of the Storm stuttering, it has since been reduced. In fact, additional overlays may cause problems, too which I have disabled for now.
   1. After further testing (and installation of the below), I no longer experience ANY stuttering.
2. Wartales was more of a brief test, I haven't played it fully yet as once I do, I lose track of time, so I'll save more testing for tomorrow.
   1. In the brief test, GPU temperatures stuck ~40C which is incredibly cool.

I will continue with more tests and see how things go. I'd like to thank everyone who contributed in efforts to reassess my issues.
My future tests will be with more intensive games.

Conclusion:
Everything appears to be successful. My feet are cold with the cool breeze by it now.
Consider this case closed.


If you'd like to know what I've done, see below:

• Removed parts from Fractal Meshify C and imported to the new Fractal Torrent.
• Installed AIO LC system
  • This seemed to be a bit problematic for the Fractal as it didn't quite fit.
  • The AIO LC system's manual required download and I had no PC to review this document, how unfortunate.
  • After multiple problems while installing, I removed this from the system and will be returning it.
    • Honestly, I believe if MSI just added a booklet for instructions, this would've went smoother.
• Reinstalled my Noctua Heatsink
  • I tried to sit this on it's other side, but again, it wouldn't fit.
• Fan Setup:
  • Installed Noctua 3200RPM fan as intake in the back of the PC, basically it should take in air from the PC and spit it out the back.
  • Reinstalled the original fans on the case after failed AIO LC above.
    • It appears the fans will take from outside and breathe cool air into the PC.
    • I initially left off the side glass case to see how it would perform and the breeze was nuts, this really does take cake for air cooling.
  • Fans are a bit louder than I'd like, but appears to be keeping things cool, the true test will be California's heatstreaks.
• Installed GPU
  • I managed to sit the GPU right under the Heatsink, BUT it's incredibly tight all around the CPU. (top, bottom, left & right side)
    • Top: There's some additional heatsinks for the CPU I believe on the motherboard that's protected by metal casing.
    • Bottom: The GPU
    • Left Side: Additional heatsinks for the CPU & other input modules.
    • Right Side: RAM sticks

Other notes:

• The torrent's glass is pretty fragile, it's already cracked just putting it back on.
• Cable management on the back was a little hectic, on a scale of 1-10 being difficult, it was at 8 as my motherboard did not seem to have some compatibilities.
  • On that note, I have a separate module that I connect so everything works perfectly fine, except for the top headboard (? is that what it's called where the default power buttons are? )

What I would do in your case, though, is test how high your card clocks and use that as a base for your undervolt target.

I reviewed this and tried some scenarios, but ultimately reverted everything back.
If the issue persists with the new setup, I will revisit this again. Thank you so much!

You should have clearance to use it in the top slot PCIEX_1.

After finally getting the new case, I tried once more and it worked. It's incredibly tight, but it works. Thank you.

They have different OEMs too. Most companies don't make their own - another company does.
MAG and MPG are made by Apaltek. MEG was done by Asetek.

In my future rig, I'll review what other options there are, as this was a quicker solution. Thank you for all the feedback you provided, I will utilize it for future debugging.

Did you try with non ultrawide resolutions?

Yes, and it did not originally help, but my original issue is now resolved.