• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Question Nitro Rx480 8GB when put under full load, freezes the system

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
I have this Rx480 which i bought used from someone locally has this weird thing going on, when run with a benchmark software or a heavy game (I ran gta v), it freezes the system.
I tried flashing the latest available v-bios from techpower-up "Sapphire.RX480.8192.160603_1.rom"
even flashed bios from another working card, but nothing. So far, i have reached nowhere, but ya, if the load is like 50-60ish% it works fine. Help me if you can. I want to revive this card so bad.
Also the fan has a whiny sound.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,617
2,486
136
That's not very good. There's a possiblity of a hardware fault. If you were close I'd try diagnosing it.

Let's try checking the temps first. Does it run hot? Does it crash instantly as you put 3D load to it?
 

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
That's not very good. There's a possiblity of a hardware fault. If you were close I'd try diagnosing it.

Let's try checking the temps first. Does it run hot? Does it crash instantly as you put 3D load to it?
no it doesn't run hot, max goes to around 60ish . When stressed with 3d load, the fans ramp up to 100% and the system freezes after 2-4sec of the load.
 

Leeea

Golden Member
Apr 3, 2020
1,145
1,345
96
no it doesn't run hot, max goes to around 60ish . When stressed with 3d load, the fans ramp up to 100% and the system freezes after 2-4sec of the load.
The seems to indicate power fault.

Maybe issue with power supply*?


*2 to 4 seconds is the key part of this. Just enough time for a capacitor to get discharged by the load and drop below spec, causing the gpu to freeze up. Is there some other load you can test the PSU with?

Guessing it was run into the ground by mining.
In this case this seems unlikely. VRM failure typically would be instantaneous, it would not wait a few seconds. Full amperage is applied in less then a millisecond.

The fans and thermal pads are clearly fine as the card remains cool.

While it is possible a hotspot on the GPU is causing this to trigger a silicon fault, that is rather rare. It appears something is causing the GPU to go out of spec and crash. Odds are it is something in the power system, the question is where?



but ya, if the load is like 50-60ish% it works fine.
In the Radeon driver for the 580, you can turn down the power limit to -50%. A crappy band aid, but if you have a power issue it could fix your problem. The downside of this is the card only has 50% power to work with, so performance is lost.
 
Last edited:
  • Like
Reactions: CuriousMike

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
The seems to indicate power fault.

Maybe issue with power supply*?


*2 to 4 seconds is the key part of this. Just enough time for a capacitor to get discharged by the load and drop below spec, causing the gpu to freeze up. Is there some other load you can test the PSU with?
I have tried changing psu's thrice, earlier with a HX-1200, then Deepcool DQ850M and now a cheapo(but reliable) one.
I test either the other system with this psu or i can try benchmarking the card on 850watt deep cool one again.
In the Radeon driver for the 580, you can turn down the power limit to -50%. A crappy band aid, but if you have a power issue it could fix your problem. The downside of this is the card only has 50% power to work with, so performance is lost.
Getting half performance is probably better then not getting any? xd
Mmaybe you meant 480*
 
Last edited:
  • Like
Reactions: Leeea

Leeea

Golden Member
Apr 3, 2020
1,145
1,345
96
I have tried changing psu's thrice, earlier with a HX-1200, then Deepcool DQ850M and now a cheapo(but reliable) one.
I test either the other system with this psu or i can try benchmarking the card on 850watt deep cool one again.
hmm, if you have tried 3 separate PSUs seems unlikely.

This is going to be a bummer, but I suggest turning down the power limit in the driver and see if that makes your problem go away.
 
  • Like
Reactions: Quanta Girl

jpiniero

Lifer
Oct 1, 2010
10,130
2,379
136
In this case this seems unlikely. VRM failure typically would be instantaneous, it would not wait a few seconds. Full amperage is applied in less then a millisecond.
Silicon does degrade. It could be the chip, could be the memory. You just normally don't see it but if it's been running at 100% 24/7 for 5 years it could be at the point where it can't be stable at full load anymore. That's likely why it was sold.
 

Leeea

Golden Member
Apr 3, 2020
1,145
1,345
96
Silicon does degrade. It could be the chip, could be the memory. You just normally don't see it but if it's been running at 100% 24/7 for 5 years it could be at the point where it can't be stable at full load anymore. That's likely why it was sold.
Miners typically do not run rx400 or 500 series at 100%. Typically they upclock the memory and downclock the GPU to reduce the power usage. This extends the life by quite a lot. With the rx500 series it is the hash rate efficiency that can be achieved using this technique that has kept them competitive and popular as mining cards.

These cards also have a legendary reputation for reliability when abused.
 
Last edited:
  • Like
Reactions: Quanta Girl

CuriousMike

Diamond Member
Feb 22, 2001
3,010
479
136
Do you have a completely different system you can test the card in? ( Rule out anything fishy in your system.)

My first guess was power supply - but you've definitely ruled that out.

The power limit suggestion is good - I would just take the power limit down 5% at a time and find where it stabilizes. You might not need to pull it down much.
 
  • Like
Reactions: Quanta Girl

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
hmm, if you have tried 3 separate PSUs seems unlikely.

This is going to be a bummer, but I suggest turning down the power limit in the driver and see if that makes your problem go away.
I did first to max i.e. -50%, works but stutters with loading frames, tried -40, works but the same issue, -30 crashes while in operation. But yea, overall its an improvement over the previously mentioned state.
 

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
Silicon does degrade. It could be the chip, could be the memory. You just normally don't see it but if it's been running at 100% 24/7 for 5 years it could be at the point where it can't be stable at full load anymore. That's likely why it was sold.
Probably
Miners typically do not run rx400 or 500 series at 100%. Typically they upclock the memory and downclock the GPU to reduce the power usage. This extends the life by quite a lot. With the rx500 series it is the hash rate efficiency that can be achieved using this technique that has kept them competitive and popular as mining cards.

These cards also have a legendary reputation for reliability when abused.
"Legendary" xd

Do you have a completely different system you can test the card in? ( Rule out anything fishy in your system.)

My first guess was power supply - but you've definitely ruled that out.

The power limit suggestion is good - I would just take the power limit down 5% at a time and find where it stabilizes. You might not need to pull it down much.
Well, the one its currently on, is a seperate rig with nothing janky and power limit down to -35 seems to be working fine but still gives stuttering frames
 

Leeea

Golden Member
Apr 3, 2020
1,145
1,345
96
I did first to max i.e. -50%, works but stutters with loading frames, tried -40, works but the same issue, -30 crashes while in operation. But yea, overall its an improvement over the previously mentioned state.
Verify enhanced sync is turned off
Anti lag is turned off
Test vsync on or off
Verify driver is up to date
Is the monitor attached with HDMI ( if so perhaps toggle hdcp off )

Which game? Cpu? How much ram?
 

solidsnake1298

Senior member
Aug 7, 2009
205
113
116
I know you said that it doesn't get excessively hot, but I am fairly certain that there is a delay of a couple seconds between when the temperature is reported in software and when that reported temp actually happened. Are you reading 60*C when you apply a heavy load just before it crashes? Or does it run at that temp under less demanding loads?

I don't know about your particular card, but my card's fans don't spin up to 100% until it is over 90*C. I would be curious what would happen if you removed the heat sink, cleaned the thermal paste, and applied fresh paste. I wonder if the thermal paste is dried/cracked.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,617
2,486
136
I did first to max i.e. -50%, works but stutters with loading frames, tried -40, works but the same issue, -30 crashes while in operation. But yea, overall its an improvement over the previously mentioned state.
Try underclocking. Actually you'd combine it with power limits.

1. Underclocking the core
2. Underclocking the memory
3. Underclocking the core + lowering power limit
4. Underclocking the memory + lowering the power limit
5. #3+#4
 

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
I know you said that it doesn't get excessively hot, but I am fairly certain that there is a delay of a couple seconds between when the temperature is reported in software and when that reported temp actually happened. Are you reading 60*C when you apply a heavy load just before it crashes? Or does it run at that temp under less demanding loads?

I don't know about your particular card, but my card's fans don't spin up to 100% until it is over 90*C. I would be curious what would happen if you removed the heat sink, cleaned the thermal paste, and applied fresh paste. I wonder if the thermal paste is dried/cracked.
Well, i did thoroughly clean it from inside and also replaced the thermal paste, but I'm concerned if the dusty thermal pads over the vram or the vrm are causing issues but I cannot see high temp spikes at all.
Try underclocking. Actually you'd combine it with power limits.

1. Underclocking the core
2. Underclocking the memory
3. Underclocking the core + lowering power limit
4. Underclocking the memory + lowering the power limit
5. #3+#4
i did all those seperately but nothing worked untill the power limit was down by atleast 40%. moreover the stutter didnt get better with no matter what things i did.
 

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
Verify enhanced sync is turned off
Anti lag is turned off
Test vsync on or off
Verify driver is up to date
Is the monitor attached with HDMI ( if so perhaps toggle hdcp off )

Which game? Cpu? How much ram?
i did ensure all those were to your recommendation, but the stuttering didnt go away.
Monitor is attatched with an hdmi to vga.
Furmark, Gta V, Hitman, Rocket League, ... all faced the same issue when underclocked.
I5 4460, 8gb gskill ripjaws 1600mhz
Could having a single stick of ram be an issue??
 

Leeea

Golden Member
Apr 3, 2020
1,145
1,345
96
Could having a single stick of ram be an issue??
not sure, but seems unlikely

four cores, should be good enough for what your using it for

Monitor is attatched with an hdmi to vga.
curious. This is rare.

Could you elaborate what this is? Are you using some sort of HDMI to VGA adaptor?
(aka something like this: https://www.amazon.com/Moread-Gold-Plated-Projector-Chromebook-Raspberry/dp/B00SW9JI9A/ref=sr_1_1_sspa )

Recommend you disable HDCP if you can, HDCP is known to cause stuttering with non-standard display setups.

Verify your monitor refresh rate is greater or equal to 60hz

What sort of FPS are you getting?
 
Last edited:
  • Like
Reactions: Quanta Girl

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
not sure, but seems unlikely


four cores, should be good enough for what your using it for


curious. This is rare.

Could you elaborate what this is? Are you using some sort of HDMI to VGA adaptor?
(aka something like this: https://www.amazon.com/Moread-Gold-Plated-Projector-Chromebook-Raspberry/dp/B00SW9JI9A/ref=sr_1_1_sspa )

Recommend you disable HDCP if you can, HDCP is known to cause stuttering with non-standard display setups.

Verify your monitor refresh rate is greater or equal to 60hz

What sort of FPS are you getting?
It is just the one you have attatched the amazon link for but with an aux port too. And, sorry i meant hdmi to vga on the gpu and not on the monitor. Monitors refresh rate is 60hz. Fps is anywhere in the range of 45 to 60. If you want, i can share teamveiwer links if you wish to take a look yourself.
Also, i did disable HDCP and it does seem to help in games where i dont see this weird stuttering, instead i am experiencing overall less frames.
This is what I have been facing
 
Last edited:

Leeea

Golden Member
Apr 3, 2020
1,145
1,345
96
Do you have a different monitor you can test it with?

or perhaps plug it into a TV to test it with?

I would like to rule out the converter, unplug the converter from the computer entirely when you do this. These are known to be yanky even if they are not the monitor being used. To elaborate, the converter you are using is an active converter that has a computer inside it that converts hdmi to vga. This introduces a small amount of lag and parasites power from the GPU. This can cause all sorts of problems / issues.



The rx480 your using can output VGA natively on unused pins of the DVI connector. It does not need a converter, just an adapter. However, it is best to verify this is the problem with an alternate monitor/TV before purchasing anything. aka this: https://www.amazon.com/DVI-I-Adapter-Benfei-Female-Plated/dp/B07G77CBBB/ref=sr_1_1_sspa
 
Last edited:
  • Like
Reactions: Quanta Girl

Stuka87

Diamond Member
Dec 10, 2010
5,488
1,277
136
Out of curiosity, where is video memory usage at when it crashes? Is it only high GPU usage, or is it also high memory and low GPU?

Also, has there ever been a different GPU in this machine?
 
  • Like
Reactions: Quanta Girl

Quanta Girl

Junior Member
Jun 22, 2021
13
1
36
Do you have a different monitor you can test it with?

or perhaps plug it into a TV to test it with?

I would like to rule out the converter, unplug the converter from the computer entirely when you do this. These are known to be yanky even if they are not the monitor being used. To elaborate, the converter you are using is an active converter that has a computer inside it that converts hdmi to vga. This introduces a small amount of lag and parasites power from the GPU. This can cause all sorts of problems / issues.



The rx480 your using can output VGA natively on unused pins of the DVI connector. It does not need a converter, just an adapter. However, it is best to verify this is the problem with an alternate monitor/TV before purchasing anything. aka this: https://www.amazon.com/DVI-I-Adapter-Benfei-Female-Plated/dp/B07G77CBBB/ref=sr_1_1_sspa
I have tested it with my tv using hdmi and also i have tried using the internal graphics with the gpu in the pcie still, the same result. The converter was added later, ie, after testing on tv and then on internal gfx output.


Out of curiosity, where is video memory usage at when it crashes? Is it only high GPU usage, or is it also high memory and low GPU?

Also, has there ever been a different GPU in this machine?
I couldn't enable mem usage in afterburner, but acc to task manager it was under 5gb, its 100%gpu and 5ish gigs memory usage.
And yes there has been a 1060 in it before, also for testing. it didnt face any such errors.



Moreover, an update; if i underclock the power limit to around -40%, and then underclock the core and mem, then run anything with vsync on, as long as it get 60fps or more, it has no such weird stutter but if it dips below, it does.
Also, I tried it for mining on nicehash, if mem and core and power limit are all underclocked, to max, max and -39% respectively it manages 10mhz.
 

ASK THE COMMUNITY