F5F Stay Refreshed Power Users Overclocking Are you making a mistake with undervolting, or is it simply bad luck with the silicon?

Are you making a mistake with undervolting, or is it simply bad luck with the silicon?

Are you making a mistake with undervolting, or is it simply bad luck with the silicon?

Pages (2): 1 2 Next
B
BRushton
Member
61
12-05-2021, 09:34 PM
#1
Hello, I bought an MSI Suprim X RTX 3080 Ti last month. It tends to get very hot, reaching temperatures around 80-82°C. After trying undervolting, it now uses about 350-400 watts. I’m new to this process with MSI afterburner and have watched several tutorials, but I’m unsure if I’m doing things correctly.

My first step was lowering the core clock to -400MHz, then adjusting it to around 1850MHz at 865Mv. It works, but the power consumption remains high. If I follow what others have done with the RTX 3090, they managed stable performance at 1900MHz and 837Mv, even reaching 1920MHz at 843Mv.

I’m wondering if I might be facing difficulties or if there’s something wrong. My clock and voltage settings seem inconsistent during games, but benchmarks show stability without crashes.

The games I tried included The Witcher 3 @ 3840x2160 on ultra, and Metro Exodus @ 3840x2160 also on ultra with DLSS enabled and ray tracing on ultra.

My current setup is:
- Ryzen 9 3900X @ 4.2GHz (1.175Mv)
- MSI Suprim X RTX 3080 Ti
- Asus Rog Strix B550-F
- Corsair Vengeance RGB 8x4 @ 3600MHz
- Corsair H150i Elite Capellix as top intake
- Lian-li O11D XL
- 7x Corsair QL120 fans (1 exhaust rear, 3 intake bottom, 3 exhaust side)
- Corsair RM1000X

For the power settings, I’m using a custom fan curve, but it still draws 350-400W and causes my PC and room to overheat.
B
BRushton
12-05-2021, 09:34 PM #1

Hello, I bought an MSI Suprim X RTX 3080 Ti last month. It tends to get very hot, reaching temperatures around 80-82°C. After trying undervolting, it now uses about 350-400 watts. I’m new to this process with MSI afterburner and have watched several tutorials, but I’m unsure if I’m doing things correctly.

My first step was lowering the core clock to -400MHz, then adjusting it to around 1850MHz at 865Mv. It works, but the power consumption remains high. If I follow what others have done with the RTX 3090, they managed stable performance at 1900MHz and 837Mv, even reaching 1920MHz at 843Mv.

I’m wondering if I might be facing difficulties or if there’s something wrong. My clock and voltage settings seem inconsistent during games, but benchmarks show stability without crashes.

The games I tried included The Witcher 3 @ 3840x2160 on ultra, and Metro Exodus @ 3840x2160 also on ultra with DLSS enabled and ray tracing on ultra.

My current setup is:
- Ryzen 9 3900X @ 4.2GHz (1.175Mv)
- MSI Suprim X RTX 3080 Ti
- Asus Rog Strix B550-F
- Corsair Vengeance RGB 8x4 @ 3600MHz
- Corsair H150i Elite Capellix as top intake
- Lian-li O11D XL
- 7x Corsair QL120 fans (1 exhaust rear, 3 intake bottom, 3 exhaust side)
- Corsair RM1000X

For the power settings, I’m using a custom fan curve, but it still draws 350-400W and causes my PC and room to overheat.

G
Gr33nFlar3
Member
158
12-07-2021, 08:05 AM
#2
the overall cooling configuration isn't properly tuned. this might be leading to temperatures higher than expected for the GPU.
every cool air intake is being drawn straight from the side exhaust before it can help circulate the case and cool parts.
your AIO is continuously pushing warmed air back into the system.
adjust the radiator fans to exhaust through the radiator.
and turn the side fans to draw in air.
G
Gr33nFlar3
12-07-2021, 08:05 AM #2

the overall cooling configuration isn't properly tuned. this might be leading to temperatures higher than expected for the GPU.
every cool air intake is being drawn straight from the side exhaust before it can help circulate the case and cool parts.
your AIO is continuously pushing warmed air back into the system.
adjust the radiator fans to exhaust through the radiator.
and turn the side fans to draw in air.

C
211
12-08-2021, 07:49 AM
#3
the overall cooling configuration isn't properly tuned. this might be leading to temperatures higher than expected for the GPU.
every cool air intake is being drawn straight from the side exhaust before it can help circulate the case and cool parts.
your AIO is continuously pushing warmed air back into the system.
adjust your radiator fans to exhaust through the radiator.
and switch the side fans to draw air in.
C
coolbencool890
12-08-2021, 07:49 AM #3

the overall cooling configuration isn't properly tuned. this might be leading to temperatures higher than expected for the GPU.
every cool air intake is being drawn straight from the side exhaust before it can help circulate the case and cool parts.
your AIO is continuously pushing warmed air back into the system.
adjust your radiator fans to exhaust through the radiator.
and switch the side fans to draw air in.

P
pikkilo
Member
74
12-08-2021, 09:48 AM
#4
That's the configuration I used before. The radiator was exhaust and the coolant temperature hit 57ºC after about ten minutes of gaming (before learning how to undervolt). I think moving the radiator to the side would help, but the issue remains—I don't want my GPU using 350-400 watts.
P
pikkilo
12-08-2021, 09:48 AM #4

That's the configuration I used before. The radiator was exhaust and the coolant temperature hit 57ºC after about ten minutes of gaming (before learning how to undervolt). I think moving the radiator to the side would help, but the issue remains—I don't want my GPU using 350-400 watts.

K
KR4TOS
Member
193
12-09-2021, 05:50 AM
#5
Did you explore the MSI Afterburner power limit slider?
K
KR4TOS
12-09-2021, 05:50 AM #5

Did you explore the MSI Afterburner power limit slider?

I
IPuckFenguins
Senior Member
380
12-09-2021, 08:38 AM
#6
Yes, i adjusted it to 80% but the clock remains around 1750-1775mhz, isn't that affecting performance?
I
IPuckFenguins
12-09-2021, 08:38 AM #6

Yes, i adjusted it to 80% but the clock remains around 1750-1775mhz, isn't that affecting performance?

_
_Coosme
Junior Member
2
12-15-2021, 05:29 AM
#7
This sounds quite cool. It might be interesting to explore the possibility of silicon lotteries. I was surprised by how much lower I could set the voltage on a 6700xt, but there are limits where it just doesn't perform well. Lower temperatures seem to help—around 28°C idle and 50°C during games. It's not as smooth as a waterblocked GPU, but it's still decent. Perhaps experimenting a bit more will reveal what the system can really do.

Same situation here; good tips on checking radiators would be helpful too. My setup has 7 LL120s, 3 exhaust fans at the top, 3 at the bottom, and one intake fan. I recently added a fan at the back expecting better cooling, but it didn't show up in my data. It might be related to cooling stages that move hot air away from other parts, offering some benefits that are hard to quantify.
_
_Coosme
12-15-2021, 05:29 AM #7

This sounds quite cool. It might be interesting to explore the possibility of silicon lotteries. I was surprised by how much lower I could set the voltage on a 6700xt, but there are limits where it just doesn't perform well. Lower temperatures seem to help—around 28°C idle and 50°C during games. It's not as smooth as a waterblocked GPU, but it's still decent. Perhaps experimenting a bit more will reveal what the system can really do.

Same situation here; good tips on checking radiators would be helpful too. My setup has 7 LL120s, 3 exhaust fans at the top, 3 at the bottom, and one intake fan. I recently added a fan at the back expecting better cooling, but it didn't show up in my data. It might be related to cooling stages that move hot air away from other parts, offering some benefits that are hard to quantify.

P
168
12-15-2021, 10:43 AM
#8
if the main drawback of running at default power is increased temperatures compared to lowering them, focus on resolving your problem.
consider using graphic-heavy programs; stress tests, etc.
disable the case side panel and use a high airflow static room fan directed at the GPU.
if temperatures drop slightly, it suggests the airflow configuration is the issue.
P
Pixelplayer145
12-15-2021, 10:43 AM #8

if the main drawback of running at default power is increased temperatures compared to lowering them, focus on resolving your problem.
consider using graphic-heavy programs; stress tests, etc.
disable the case side panel and use a high airflow static room fan directed at the GPU.
if temperatures drop slightly, it suggests the airflow configuration is the issue.

B
benjimaub
Junior Member
8
12-15-2021, 11:38 AM
#9
Well, I don't think I have enough room in my space to carry out that change, but I appreciate the advice. I'm certain the airflow is the problem since the QL fans are running more than moving air through the case. My idea would be to place the rad on the side (with the QL fans as intake) and position the ML rad fans at the top for exhaust. If I follow that, the airflow should look like this: 3x QL120 1600 RPM Intake bottom, 3x QL120 1600 RPM Intake side + rad, and 3x ML120 1600-2500 RPM Exhaust top with one QL120 1600 RPM exhaust rear.
B
benjimaub
12-15-2021, 11:38 AM #9

Well, I don't think I have enough room in my space to carry out that change, but I appreciate the advice. I'm certain the airflow is the problem since the QL fans are running more than moving air through the case. My idea would be to place the rad on the side (with the QL fans as intake) and position the ML rad fans at the top for exhaust. If I follow that, the airflow should look like this: 3x QL120 1600 RPM Intake bottom, 3x QL120 1600 RPM Intake side + rad, and 3x ML120 1600-2500 RPM Exhaust top with one QL120 1600 RPM exhaust rear.

E
eduardodd08
Posting Freak
852
12-29-2021, 06:10 PM
#10
it's always more efficient from a temperature standpoint to use any radiators for cooling.
avoid letting the air warmed by the CPU be sent back into the case, which can affect other parts.
E
eduardodd08
12-29-2021, 06:10 PM #10

it's always more efficient from a temperature standpoint to use any radiators for cooling.
avoid letting the air warmed by the CPU be sent back into the case, which can affect other parts.

Pages (2): 1 2 Next