System Lockup when GPU hits full clockspeed

[Grimm Spector]

Hey, I've got a system that was working previously. I replaced the stock cooler, and the new cooler doesn't have the right fan power connector, so I have it connected to main power so it's just always on. The temperature on the GPU (R9 280X Visiontek) is very low, idling under 30 degrees, and never passes 40. However whenever I run anything that pushes the clockspeed up to its max the screen goes black and the whole system locks up. I have to power it down manually. Any ideas if it's perhaps the fan not being connected, or what the issues could be? Super frustrating, thanks!

 

[tential]

What power supply do you have?
What are your full system specs.

The more info the better.
Thanks,

 

[Grimm Spector]

I don't remember the model of the power supply but it's a 700 watt supply Corsair I'm pretty sure. I wasn't having any power issues before the switch.

It's an i5-2600k, Visiontek 3GB Radeon 280X, 16 GB DDR3, Gigabyte H61M-S2PV board, not sure what else might be helpful here. It's running a pair of crappy conventional SATA hard drives and one SSD. Nothing else should be drawing power, removed excess USB devices while trying to figure out the problem.

 

[ZGR]

How are the VRM temps?

 

[dlerious]

Did you check to make sure the cooler was seated properly? Did you replace the paste (TIM) as well?

 

[Grimm Spector]

How are the VRM temps?

I don't have anything that detects those, the GPU has a core temp, never goes above 40, which makes me pretty confident of the cooler being seated properly. Motherboard also only shows temperature for the CPU core, not for any other bits, but the temperature of that doesn't get above 50 or so unless I am running it hard for awhile. Unfortunately Gigabytes mobo util EasyTune doesn't actually run, it just dies when I try to run it, so I can't see if it has more information about the VRM or the PSU.

Did you check to make sure the cooler was seated properly? Did you replace the paste (TIM) as well?

And yes the paste was removed and replaced, cooler is seated firmly.

I'm very curious if it's possible that the card is killing itself because it is detecting that it's fan plug is not connected to anything, and therefore thinks the fan is dead and fails out.

 

[Paratus]

I’m leaning towards the PSU being bad. Any chance you can measure the voltage on your 12V rail, either though software or with a multimeter?

If you have access to another similarly powerful GPU you could swap them and see if it still hangs. If it doesn’t it’s the GPU. If it does it’s likely the PSU.

 

[Guru]

Its clearly the new fan since its not connected to the GPU. Therefore the GPU can't regulate the fan speed and it crashes. Its simple as that. you didn't do due diligence before buying a new cooling solution and now you are paying the price.

If you have the old cooler(assuming it works and you replaced it just to have better cooling) then put that back in.

I'm assuming the connector on the GPU is with 3 pins and the new cooler has 4 pins? If its the other way around you could still plug the 3 pins in the 4 pins on the GPU, just make sure the right side end is placed to the most right side of the connector on the GPU as well, usually the last pin is the sensors, while the others are for the power for the fans.

Also are you checking the temperature during playing games? Idle temperature is always going to be low no matter what, you could literally have no cooler and idle temperature would still be low, so you need to check it when the GPU is being fully used in games!

 

[daveybrat]

Can you post a link to the new cpu cooler you purchased? If the system was working fine before replacing that, then we have to focus on the cooler.

 

[Grimm Spector]

Can you post a link to the new cpu cooler you purchased? If the system was working fine before replacing that, then we have to focus on the cooler.

It's just a Kraken G10 with a Corsair H60 in it.

Is it possible it's the VRMs overheating? My GPU doesn't seem to report the temperatures of those at all, and the board didn't appear to have a heatsink on them of it's own, just the fan above them, which is still there, but always on with main power.

Edit: Note, the cooler was installed on the GPU, not the CPU. CPU had not been touched during any of this.

 

[Grimm Spector]

Its clearly the new fan since its not connected to the GPU. Therefore the GPU can't regulate the fan speed and it crashes. Its simple as that. you didn't do due diligence before buying a new cooling solution and now you are paying the price.

If you have the old cooler(assuming it works and you replaced it just to have better cooling) then put that back in.

I'm assuming the connector on the GPU is with 3 pins and the new cooler has 4 pins? If its the other way around you could still plug the 3 pins in the 4 pins on the GPU, just make sure the right side end is placed to the most right side of the connector on the GPU as well, usually the last pin is the sensors, while the others are for the power for the fans.

Also are you checking the temperature during playing games? Idle temperature is always going to be low no matter what, you could literally have no cooler and idle temperature would still be low, so you need to check it when the GPU is being fully used in games!

To give some more depth here, and some new information I was able to pull with another utility, the GPU with the old cooler was idle around 56 degrees and got far too hot during play. Artifacting started showing up, so I stopped using it until I got a new cooling solution so I wouldn't continue likely damaging the card.

When running a game with it at a somewhat low load it goes from idling between 30 and 40 to about 52 so far, but I don't know how much higher it will go since it crashes out.

I'm glad it's likely the fan connection, I have an adapter on the way to go from the mini GPU 4 pin to the TX3 on the new fan.

A proper adapter should as far as I know bridge the sensing pin so it just sees the fan as always at 100%.

The new data shows that the card is getting a very stable 12.16 volts, fluctuating down as low as 12.16 and as high as 12.18, right up until it shuts off. VRM temperatures are low, never exceeding 43 degrees so far.

 

[Grimm Spector]

So an adapter didn't work to extend the fan out and let me run it; that said with the stock fan attached the system will run a game or benchmark for a minute or two, but then it does the same thing as before, locking up with a black screen on the main monitor and the system freezing. No voltage issues, no heat issues on any sensor. I cannot figure out what is wrong here.

Someone must have another idea? PSU swap didn't help either. And I can't see why having the stock fan attached but not cooling the card in any way, just running nearby makes any difference to how long the card goes before dying.

 

[thecoolnessrune]

What do you mean that the adapter "didn't work"? There's no reason why a mini-PWM fan connector to PWM fan shouldn't work, as it's a passive device, unless you got something that was in itself faulty, or there's an issue with the fan?

 

[Krteq]

@Grimm Spector: Any OC on that card? I had the same issue with R9 290 and it was caused by insufficient voltage set for MEMIO. It was not happening once I reset clocks to factory defaults and issue was gone when I adjusted auxiliary voltage via MSI Afterburner.

 

[Grimm Spector]

What do you mean that the adapter "didn't work"? There's no reason why a mini-PWM fan connector to PWM fan shouldn't work, as it's a passive device, unless you got something that was in itself faulty, or there's an issue with the fan?

Didn't work as in it did not resolve the problem. With the adapter attached the card functioned as though the fan wasn't attached at all. With the fan wedged in there which won't work with the case closed, it works for a few minutes, and then fails the same way. Which seems even more unusual.

@Grimm Spector: Any OC on that card? I had the same issue with R9 290 and it was caused by insufficient voltage set for MEMIO. It was not happening once I reset clocks to factory defaults and issue was gone when I adjusted auxiliary voltage via MSI Afterburner.

No overclocking, it's stock. Otherwise this would've been one of my thoughts.

 

[Guru]

Can you post pics of how you connected the fan with and without the adapter? We can't help you anymore if its all so vague.

You also revealed in later posts that your card already had issues with the old cooler, info that should have been available from the start. It seems to me from everything I've gathered, all your bits and pieces you had a custom 280x, at some point it started overheating and causing issues(probably thermal paste wasted out), and would cause black screen when playing games, you replaced the cooler thinking it was the cooler, but now you have the same issues.

It seems like the card is somewhat faulty, have you tried undervolting the card and see if you still have issues?

 

[Grimm Spector]

Can you post pics of how you connected the fan with and without the adapter? We can't help you anymore if its all so vague.

You also revealed in later posts that your card already had issues with the old cooler, info that should have been available from the start. It seems to me from everything I've gathered, all your bits and pieces you had a custom 280x, at some point it started overheating and causing issues(probably thermal paste wasted out), and would cause black screen when playing games, you replaced the cooler thinking it was the cooler, but now you have the same issues.

It seems like the card is somewhat faulty, have you tried undervolting the card and see if you still have issues?

It is not custom, it was a Visiontek card, not modified at all initially when problems began. The thermal paste looked fine when I removed the stock cooler, but i cleaned it off and replaced it when putting on the water cooler.

I'm not sure how images will help, however I can post some. Currently I have the card out of the system and am using a crappy older card to let me at least use the system in the meantime.

I have not tried undervolting the card, it's been suggested to me that can be as harmful as over voltage, how far should I drop the voltage?