RX480 Gaming X target temp issue?

[koniu82]

I have issues with target temperature of RX480 Gaming X 8GB.

It begins two weeks ago. When card is reaching 64C card fans start running faster and faster trying to keep 64C temperature. After minute or two i was ending with fans spinning over 2400 rpm. I set  my own fan setting in afterburner to keep them low(1200-1500rpm) . But i find new problem. When card reach 64C it is trying to keep that temp again . Card cant control fans becouse of Afterurner so it start throttling. Clocks jumping between 1200-1303MH. Throttling at 63C???

Currently i am using own setting for card in Afterburner. 1235MHz, fans at 1200rpm and Card holding clocks flat and temp is stable at 64C. I was forced to do that because throttling making my games stutter.

Earlier everything was ok. Target temp 73. Fans run at that temp at ~1300-1400rmp when playing Witcher 3 and clock was rock solid 1303MHz. An it was during hot Septembers days.

I try to change target temp in WattMan, nothing change, new drivers - different driver, DDU and new drivers, drivers with and without WattMan installed, fresh Windows 10 installation. Mobo bios reset, no effect, Card always trying to keep temp at 64C target temp.

PC spec.
i5-4590
rx 480 gaming X 8GB
MSI B85-G43 (bios 3.10)
Corsair RM550X
Windows 10 Anniversary Update with all updates
All other drivers up to date

Temps of hardware are ok. CPU under stress is ~57C, Mobo temps ~50C, GPU as i mentioned 64C

GPU-Z screenshot showing card temp stable at 64C and still You can see throttling during WItcher 3. 

 

[iRileyx]

What does afterburner have your max temp set to?
 

Also in wattman do you have a frame target set?

[Matt_98]

check the power limit

[koniu82]

Wattman is currently uninstalled i have only driver (16.10.3) installed. Earlier Wattman was showing stock setting -  73C as target and 90C as max temperature. Changing it in Watmann is not working. No effect at all.

 

In AF max temperature is 90C. But card newer go above 64C. Few times when i saw 65,66C i was only for sec and GPU is starting throttling and returning to 64C

 

3 hours ago, Matt_98 said:

check the power limit

No effect at all, card is behaving the same with Power limit set to 0 or +50%

 

[iRileyx]

16.10.3 is the most recent driver for wattman so that shouldn't be the issue.

Trying to think what else you could check, my rx480 is the XFX GTR XXX and even at 70C+ it doesn't start to throttle, weird.

[koniu82]

Just guessing, fallowing power limit trail.

Faulty PSU?, maybe damaged 6+2 PCI-E cable so it is working like 6PIN connector???

 

 

It lookrealy strange.

  Whenis start using GPU(game/bench) card is holding 1303MHZ, until temp reach 64C clock is stable. At moment GPU have  64C throttling begins

[YongKang]

Can you set your temperature limit higher? It seems like your target temps is 64 degrees and your card is desperate to keep it at that temperature.

[koniu82]

1 minute ago, YongKang said:

Can you set your temperature limit higher? It seems like your target temps is 64 degrees and your card is desperate to keep it at that temperature.

No. Afterburner is already at 90C, and in WattMan there is no reaction. I can set it at max passible number 90C but card still using 64C as target

[LukaH]

maybe try an older driver if this has started happening recently?

[koniu82]

7 minutes ago, LukaH said:

maybe try an older driver if this has started happening recently?

It was ~2 weeks ago. I was trying probably all drivers since 16,7.x, nothing help

 

Only hardware i change was PSU, maybe week earlier. But how PSU can be source of issues like that.

Also it is good quality PSU Corsiar RM550X.

[LukaH]

what did you have before?

[koniu82]

Earlier i had 4 Year old Corsair CX600M. But it was noisy, needed fan swap, and also it has coil whine

 

I must be some kind of power/temperature limit.

 

We all know that Polaris have 7 Tiers of Frequency/voltage.

For my card Tier 6 is 1236MHz and 1081mV,  tier 7 is 1303MHz/1168mV,

So far i am stable at 1235Mhz, 1000mV, also stable at 1236MHz, but i only need to set card at 1237MHz and throttle  begins. It is harder to look at voltage but i am also almost sure that as long as voltage is below 1081mV there is no throttle . If at least one of frequency/voltage/temp 1237MHz/1081mV/64C is reached GPU start throttle. It is only theory.

[ivan134]

1st of all, don't use WattMan for anything ever. The software is useless right now.

Can you show us a pic of your fan curve? I would also increase your Power Limit up to 50%.

[koniu82]

24 minutes ago, ivan134 said:

1st of all, don't use WattMan for anything ever. The software is useless right now.

Can you show us a pic of your fan curve? I would also increase your Power Limit up to 50%.

I don`t using WattMan. From AMD only installed  video and sound driver. I am using MSI Afterburner

Changing Powel Limit to +50% have no impact on throttling

 

[ivan134]

3 hours ago, koniu82 said:

I don`t using WattMan. From AMD only installed  video and sound driver. I am using MSI Afterburner

Changing Powel Limit to +50% have no impact on throttling

 

That's a really bad fan curve. Your fan speed is way too low at high temps. Make it looks something similar to this. 

 

 

[YongKang]

4 hours ago, koniu82 said:

I don`t using WattMan. From AMD only installed  video and sound driver. I am using MSI Afterburner

Changing Powel Limit to +50% have no impact on throttling

 

To be honest, that fan curve is bad.  I suggest letting the fan gradually ramp up. You better set 20-30 % fan speed for under 40. 40-60 at 30-40%. Higher temps at 50 or 60%. That should be enough.

Looking at your fan curve, you're letting way too much heat to be accumulated between 0-60. Of course you gonna have a jet engine on your hand. 

[koniu82]

I was already testing it. I have test it with fans at 40%, 60%, 80%

No effect at all. With default clocks and voltage settings card was not holding boost clock. Temperature was  64 or lower and card still throttling. Maybe with higher fans speeds  throttling was smaller but it still was happening.

 

Just to be sure i will retest it today and give some GPU-Z screenshots.

I am also planing today to test card with second PCI-E slot in my mobo. Also i am trying to test  my card in different PC but who will know how hard is to find someone with traditional PC. Everyone i know have laptops

 

Also I plan to use SATA-8Pin adapter as power source to check if PSU PCI-E cable connector is not problem but because my PSU have single 12V rail i don`t see how It can cause problem on PCI-E cable and work on adapter. Already tested different PCI-E cable, no effect

 

Also new 16.11.1 driver is out

[koniu82]

Just to find good sides of my problem, With current setting i am using to prevent throttling 1235MHz, 990mV card is very cold. GPU only power draw in now ~100W compering to 140W when clock was 1303MHz

Also i cant notice major performance drop. Time Spy is  -100 points (~2,5%), Witcher 3 is 2-3 fps (3-5%) and most of time game still holding 60fps and 50+ in CPU heavy places

[koniu82]

Just i case someone want to look this is  my card bios

https://drive.google.com/file/d/0B6FgFbWBLaUoLWZEV08wYnJreDg/view?usp=sharing

 

[koniu82]

Look at this.  This is my test at almost default settings 1303MHz, Power limit +50, Target temp 90C, Voltage -48mV. Fan at 40% (1500rpm)

 

I will talk about 4 different sections on graf. For borders i have use red line

 

1. Test beginning. Temperature below 60C, GPU clock stable 1303MHz.

2. GPU hit ~61C territory. Clock start  throttle. 10-80MHz drop. Tem,temperature never go above 63C

3. I have reset undervoltage back to default. Heavy throttle begins at moment i hit Apply. At end of sector 3 i return with voltage to -48mV throttling is much smaller.

4. i set fans at 60% (1900rpm) ver loud. GPU still throttling even if temperature is 58C

 

I ahve no idea what is happening.

[ivan134]

45 minutes ago, koniu82 said:

Look at this.  This is my test at almost default settings 1303MHz, Power limit +50, Target temp 90C, Voltage -48mV. Fan at 40% (1500rpm)

 

I will talk about 4 different sections on graf. For borders i have use red line

 

1. Test beginning. Temperature below 60C, GPU clock stable 1303MHz.

2. GPU hit ~61C territory. Clock start  throttle. 10-80MHz drop. Tem,temperature never go above 63C

3. I have reset undervoltage back to default. Heavy throttle begins at moment i hit Apply. At end of sector 3 i return with voltage to -48mV throttling is much smaller.

4. i set fans at 60% (1900rpm) ver loud. GPU still throttling even if temperature is 58C

 

I ahve no idea what is happening.

I think you should RMA your card. It seems there is something severely wrong with it

 

[koniu82]

Probably will do it next week.

I will do last test to be sure it is card fault. Probably this weekend i will be able to test it in totally different PC. If this is card flout it should behave the same in new rig

If card will star working properly i will have to look what is wrong with other parts in my syste,. But all sights telling me that it is card.

[Deli]

Although I doubt it will have any effect. What happens if you "unlink" the temp limit and the power limit?

[koniu82]

Unlink on, off no change how card is working

I am also sure that power and temp sliders are working.. Moving them down start huge frequency drop or in case power card sometimes is crashing

[koniu82]

Just start RMA process will inform if i learn something new

I hope it will be smooth and i will have new card fast