gpu always clocks at 300mhz

[Julian321]

Hi everyone,

I have a problem with Palit RTX 2070 Super.
If want to play a game, the game starts to bucking.
Then I run a Firestrike test and I can see that the gpu always clocks at 300mhz.
The HW Monitor shows that the gpu only used 0 Watt.
I have aready download a new driver.
I dont'n know what I should do.

Thank's for Help

My Hardware:
- Palit RTX 2070 Super
- B550 Elite V2 Mainboard
- Ryzen 5 5600g
- Thermaltake 630w power supply
- Corsair 32gb Ram 3200mhz

 

[eidairaman1]

I see it was at 1605 from the value, get gpu z and click the arrow next to the uefi checkbox and run the load test on it.

 

[Julian321]

There is a GPU Z Test

 

[DarkDreams]

What are the temperatures (GPU and hot spot) during load. Hard to tell, but from the GPU-Z shot it looks pretty hot, might be thermal throttling. Also when you scroll down in GPU-Z there should be limit reasons. Those could be helpful too.

 

[Julian321]

The temperature is between 70 and 80 degrees

 

[Marcus L]

It's at 300mhz now with no/6% load though you can see the clocks were higher when you had more of a load on it though fluctuating up and down a lot, what's the hot spot temp when running the gpuz test, what is your case ventilation like? Is the gpu clean and free from dust? May benefit from a clean and repaste

 

[Julian321]

Yes, the case ist new and hast a good airflow.
The GPU ist clean I have also change the thermal passt.

 

[Tatty_Two]

Your graphics card appears to be only taking power from your PCI-E slot which I am guessing is why it is not boosting beyond it's base clock and dropping down to those low speeds under load if it is only getting a max of 75w , what is your specific PSU model and how old is it?

 

[Aken Bosch]

Make the GPU-Z window bigger like this in order to see all of the sensors readings:

 

[eidairaman1]

There is a GPU Z TestView attachment 371149

Go on the graphics card tab and do the run, with hwinfo64 open

Your graphics card appears to be only taking power from your PCI-E slot which I am guessing is why it is not boosting beyond it's base clock and dropping down to those low speeds under load if it is only getting a max of 75w , what is your specific PSU model and how old is it?

Can that gpu even run on a 630w unit to begin with

 

[Tatty_Two]

Go on the graphics card tab and do the run, with hwinfo64 open


Can that gpu even run on a 630w unit to begin with

Well his card is upto 220w and taking into consideration his other specs then he should be OK depending on age and model of the PSU but I am getting the feeling it is a power supply issue looking at what he has provided to date.

 

[Marcus L]

Go on the graphics card tab and do the run, with hwinfo64 open


Can that gpu even run on a 630w unit to begin with

Well his card is upto 220w and taking into consideration his other specs then he should be OK depending on age and model of the PSU but I am getting the feeling it is a power supply issue looking at what he has provided to date.

My son's PC has similar spec, 2070S and Ryzen 5500 on a cheap EVGA 600w 80plus white PSU and is more than enough, he's been running that combo well over a year without issue, in theory OP's PC shouldn't use more than 400w at a push, though you could have a point about possible PSU/power issue as 630w is not a common wattage you will find on PSU's today and that PSU could be 10+ years old at this point by doing a quick google for TT 630w PSU lol

 

[Aken Bosch]

My son's PC has similar spec, 2070S and Ryzen 5500 on a cheap EVGA 600w 80plus white PSU and is more than enough, he's been running that combo well over a year without issue, in theory OP's PC shouldn't use more than 400w at a push, though you could have a point about possible PSU/power issue as 630w is not a common wattage you will find on PSU's today and that PSU could be 10+ years old at this point by doing a quick google for TT 630w PSU lol

Most likely this one https://www.emag.ro/sursa-thermaltake-smart-se-630w-sps-630m/pd/EFK0NBBBM/

My money's on borked PSU and possibly even the graphics card because of that.

 

[Julian321]

Yes, I have the thremaltake smart se and I have a other gpu (RX6800 XT).
The other gpu is working with this power supply.
So I think the problem is the gpu.

 

[Aken Bosch]

every graphics card has some (huge) power spikes that may trigger your PSU, just saying. Get a reliable 800W+ PSU and check how's it going. Depending on where you live, it's easy or not... (I mean getting an 850+ PSU, testing, get it back in warranty because reasons...).

 

[eidairaman1]

Yes, I have the thremaltake smart se and I have a other gpu (RX6800 XT).
The other gpu is working with this power supply.
So I think the problem is the gpu.

Well since other is working but this 1 is not, it is possible there is a component that failed that keeps that gpu in a limp mode, probably just easier to replace it.

 

[:D:D]

Your graphics card appears to be only taking power from your PCI-E slot which I am guessing is why it is not boosting beyond it's base clock and dropping down to those low speeds under load if it is only getting a max of 75w , what is your specific PSU model and how old is it?

HWiNFO shows PCIe also at 0W. Does seem IMO to be a power sense problem.

 

[agent_x007]

There is a GPU Z TestView attachment 371149

Can be just an issue with signals/shunt resistors, but may also mean a failed controller (since it can't figure out what is the power draw of both "GPU Chip" and "Board").
Which may suck for repair (as controller has internal settings, and those are set by GPU manufacturer)

 

[eidairaman1]

6 yo card by arch, not economical to repair, go get a new unit

 

[Julian321]

Es kann sich lediglich um ein Problem mit Signalen/Shunt-Widerständen handeln, es kann aber auch ein Hinweis auf einen Controllerfehler sein (da der Controller nicht ermitteln kann, wie hoch die Leistungsaufnahme von „GPU-Chip“ und „Platine“ ist).
Das kann bei der Reparatur schwierig sein (da der Controller interne Einstellungen hat und diese vom GPU-Hersteller festgelegt werden)

Yes, the Chip in the RtX 2070super ist a holtek chip. I had already bought it with the correct Firmware and soldered it but unfortunately it didn't get any better

 

[Solaris17]

Weren’t 300mhz locks generally fuse related? I know many in the old days replaced them or bridged them. 300mhz locks used to always be limp mode indicators. Maybe I’m too out of touch, but if it is no amount of firmware will fix it. Go find the fuse and jump it.

Edit: really racking my memory here but it also happened when thermistors were broken. Does anything read the temp of the sun? Maybe it reads the surface of Pluto?

 

[Julian321]

Unfortunately, I don't know exactly where to look, I'm honestly still a beginner.

 

[Marcus L]

Unfortunately, I don't know exactly where to look, I'm honestly still a beginner.

I believe these highlighted in red are the fuses, wait for someone else to confirm, unfortunately never bridged fuses myself in the past so I can't help you further, I know a few people on here are familiar with GPU repairs @Solaris17 can possibly elaborate further

 

[Solaris17]

I believe these highlighted in red are the fuses, wait for someone else to confirm, unfortunately never bridged fuses myself in the past so I can't help you further, I know a few people on here are familiar with GPU repairs @Solaris17 can possibly elaborate further

View attachment 371324

yeah those are them; though the exact location on the card in question, and which of the several that likely exist is the one thats bad? No idea. But if you have a multi meter you should be able to measure them.

 

[StViolenceDay]

I believe these highlighted in red are the fuses, wait for someone else to confirm, unfortunately never bridged fuses myself in the past so I can't help you further, I know a few people on here are familiar with GPU repairs @Solaris17 can possibly elaborate further

View attachment 371324

That are NOT fuses, those are 0.005Ohm shunt resistors used for current sencing. Unlike fuses that are fast burning on 20-30Amps current cuased by a short circuit - those are not designed to burn immediately, so while they are sometimes damaged during a short-circuit - that typically quite late when many other things already burned.

Visually there is no fuses in OPs GPU.
And while the OPs problem is related to current sencing - it seems that it is 0W on all of them, the problem is not directly relaed to those sencing resistors, something more common

For example the element marked in red near the Current-sencing IC U503 look strangely white -

White elements on GPU boards are very rare, so I think thats a damaged resistor - compare it with a black one on a similar board https://www.techpowerup.com/review/...0-super-gamerock-premium/images/back_full.jpg

It may be damaged-by-itself, especially if thats version without a backplate, or damaged du e to a shortcircuit inside current sencing IC.