[MSI VEGA56 Airboost] GPU Clock Speed problem, please help

[Potato90]

Hi everyone,

 

I have MSI vega 56 airboost for nearly 1 year and it work well for me. Yesterday I decided to repaste the thermal paste (Noctua NT-H1) to the graphic card. However after reassembly and put into the computer, start the program, GPU clock is alway maximum at 800MHz - 900MHz when fullload, despite I increased the power limit to 50%, GPU power is only maximum 100W.

Temperature are totally fine, all are bellow 50°C

Everything I reset to default, fresh window installed, up to date driver installed but nothing change. Also repaste again but no effect.

 

What is wrong with my graphic card ? Any idea to fix this?

 

My system:

Mobo: Gigabyte X370M

CPU: AMD Ryzen 5 2600

RAM: 16 GB Patriot Viper DDR4 3000

GPU: MSI Vega 56 Airboost

PSU: Corsair Vengance 650M 80plus silver

[Jurrunio]

no such problem before repaste? Did you have enough paste to cover up the entire package (core and HBM)? Did you make sure all the thermal pads are covering places they did before?

[Potato90]

4 minutes ago, Jurrunio said:

no such problem before repaste? Did you have enough paste to cover up the entire package (core and HBM)? Did you make sure all the thermal pads are covering places they did before?

Yes, before this I do not have problem. I think I had enough paste to cover up all the package (core and HBM) because the temperature is totally fine. GPU hotpot temperature also is just 50°C. When I reduced the thermal paste, temperature ramp up to 60, hbm around 65 and hotspot is around 80 but the GPU clock is just 800MHz - 900MHz. Afterthat, I repasted again with more thermal paste, also on the heatsink, temperature back to 50°C but the GPU core clock not change. This is strange and I don't why.

The thermal pad for VRM and Choke I kept from reference card and also correct position.

[Jurrunio]

1 minute ago, Potato90 said:

Yes, before this I do not have problem. I think I had enough paste to cover up all the package (core and HBM) because the temperature is totally fine. GPU hotpot temperature also is just 50°C. When I reduced the thermal paste, temperature ramp up to 60, hbm around 65 and hotspot is around 80 but the GPU clock is just 800MHz - 900MHz. Afterthat, I repasted again with more thermal paste, also on the heatsink, temperature back to 50°C but the GPU core clock not change. This is strange and I don't why.

The thermal pad for VRM and Choke I kept from reference card and also correct position.

what kind of full load is that? Furmark?

[Potato90]

1 minute ago, Jurrunio said:

what kind of full load is that? Furmark?

I tested with uniengine valley, superposition 1080p extreme and also Shadow of Tomb Raider without Vsync. All of them clock GPU is just run around 900MHz. HBM clock is fine, I overclocked to 900MHz and then have effected in game and benchmarks.

Superposition 1080p extreme normally I get 4200 point with power limit +50%

Now I get just 2300 @@.

[Jurrunio]

6 minutes ago, Potato90 said:

clock is fine, I overclocked to 900MHz and then have effected in game and benchmarks.

so overclock it more to its usual speed?

 

I guess the vbios just corrupted itself, you can try flick the bios switch and get onto its slave bios. If that solves the problem, your masterr bios has bugged out.

[Potato90]

4 minutes ago, Jurrunio said:

so overclock it more to its usual speed?

 

I guess the vbios just corrupted itself, you can try flick the bios switch and get onto its slave bios. If that solves the problem, your masterr bios has bugged out.

I tried to set the state 7 (1622MHz) in Wattman is Minimum State, GPU clock just jump around 900MHz.

 

I just test the 2nd bios, nothing change, GPU clock is the same, around 900MHz, GPU power draw is around 100W.

I just think maybe I screw too tight when re-assembly but I saw the core, there is no crash or broken.

[Jurrunio]

Just now, Potato90 said:

there is no crash or broken.

too late at that point

 

1 minute ago, Potato90 said:

I tried to set the state 7 (1622MHz) in Wattman is Minimum State, GPU clock just jump around 900MHz.

does it say its throttling reason? Maybe your system power plan is set to somthing meant to save power? (I know Ryzen Master has one but dont know if Radeons have their equivalent).

 

Also make sure Radeon Chill is disabled.

[Potato90]

27 minutes ago, Jurrunio said:

does it say its throttling reason? Maybe your system power plan is set to somthing meant to save power? (I know Ryzen Master has one but dont know if Radeons have their equivalent).

 

Also make sure Radeon Chill is disabled.

I checked all of them. Everything is correct setting with high performance power plan. I will reassembly it once again and look for some scratch or brocken on the PCB board. Hope that this is not hw issue.

Is that okay if the thermal paste go around the small capacitors and ICs around GPU core, or I need to clean all of them?

[Jurrunio]

36 minutes ago, Potato90 said:

Is that okay if the thermal paste go around the small capacitors and ICs around GPU core, or I need to clean all of them

usual pastes arent electrically conductive, so no you dont have to clean them. It is recommended though because clean is better to the eyes.