Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

Strange RX 580 behaviour

Agosto
 Share

Hi! I'm trying to undervolt my Sapphire RX 580 Pulse but I'm noticing something strange. The card will show random artifacts in FS Extreme stress test (a rather good test for artifacts) at stock clocks (1366MHz) and 1100mV, but everything's fine at 1340 MHz (even at 1080mV, with only one artifact seen in FS Ultra)
 

The point is that core voltage doesn't seem to stay still. At 1100mV on P7, it usually hovers between 1037 and 1087mV with peaks at 1100mV, while at 1080mV P7 it stays around 1030-1075mV with 1081mV peaks. Is it some kind of vdrop or it's a normal behaviour for Polaris? Can I force a static P6/P7 voltage rather than having it oscillate?


Tests used: Firestrike Extreme, TimeSpy, Unigine Heaven and Valley.
Monitoring software used: HWInfo, Afterburner, GPU-Z
All the oc/uv is done through Wattman, +30% PL, gpu temps always below 68-69°C and various memory "voltage" settings tried (1000mV or 1050mV don't make a difference). Shouldn't the "memory voltage" in Wattman be a voltage floor for the gpu core and the memory controller? Setting it to 1050mV doesn't prevent the chip to go below that floor.


Other issues are: 
1) random fullscreen artifacts on hardware accelerated loads (satellite Maps is a very good way to induce them). Workaround found: raise P1 and P2 voltages to keep the minimum voltage during load at about the same idle one (800mV). Again, it looks like there's some voltage drop during load. Also looks like it only happens in Freesync is turned on, but I'm not 100% sure
1.1) idle voltage is always at 800mV, while in vbios is 750mV. I think it may be because of the high res screen (UHD) triggering higher idle voltage, like on multimonitor setups. Again, not sure.

2) I get memory errors, at least on my 1100 mV undervolt, both at stock clocks and 1340MHz, no matter the "memory voltage" settings in Wattman is. Based on the settings they range from a few per FS stress test runs (like 20-40) to 2000-3000 after a couple hours of gaming (only tested Dark Souls 3 and Quake Champions, actually, don't know how other games behave). Interestingly, on FS stress test I get more errors in 1080p (normal firestrike) rather than UHD (Ultra). I don't know what they actually are, if corrected ECC errors or something that could impact performance; only thing I may have noticed is that they tend to show after the GPU temperature has stayed enough time at the max value (68-69°), indicating that there might be some temperature correlation; however it's only 8Gbps hynix GDDR5 and also in contact with the cooler baseplate, it should be cool enough.

3) probably for the same reason as above, I can't manage to OC my vram. As soon as I get above 2000MHz, memory errors get higher and higher. 2100 crashes after some stressing, 2200+ means almost instant crash or red screen. Again memory controller voltage makes no difference, as far as I can tell, and I don't want to go above 1050mV for that one.

The system is on my profile. Windows 10 is on a rather old release (1709) and about 3yo installation, but that's not very dirty. AMD Drivers used: currently 18.10.2, but it was exactly the same with 18.3.1 drivers. I can't RMA, if anyone is thinking about it, since I bought it used.


Thank you in advance and sorry for all these questions, but I can't find anything that can answer them by searching on the forum/google/reddit etc.

 

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

Well I suppose no one can help me

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

Are you getting artifacts when your card is running completely stock?

 

Memory corruption issues can come up when undervolting sometimes, and it might end up being a case of you losing the silicon lottery spectacularly. If there are artifacts when you're running completely stock, I'd think about replacing the card when you can, as it might be something on the board starting to die.

Link to comment
Share on other sites

Link to post
Share on other sites

On 11/15/2018 at 3:09 PM, jkelley1123 said:

Are you getting artifacts when your card is running completely stock?

 

Memory corruption issues can come up when undervolting sometimes, and it might end up being a case of you losing the silicon lottery spectacularly. If there are artifacts when you're running completely stock, I'd think about replacing the card when you can, as it might be something on the board starting to die.

Thank you for you answer and sorry if I'm writing late!

As I wrote on point 1, I get fullscreen artifacts if I'm using stuff like light/moderate hardware accelerated workloads, like Maps on satellite mode for example, and looks like it only happens with freesync active. However, I get no artifacting on games at stock, didn't check memory errors though. With the undervolt and the low p-states "fix", I get neither artifacts nor memory errors on Maps, which is the most reliable way to induce that stuff.

No artifacts even at 1340MHz and 1100mV, however I got some strange light issues on Quake champions at 1366MHz core clock; in some matches some parts of the map had lighting issues and there were very strong coloured light "patches" (like a very reflective surface, or even a diffused light) in very defined points. It may have been the game, but they looked like lighting errors caused by the video card

One of the thing I don't get is the variable voltage even on load. Shouldn't it be more stable around the selected voltage in wattman? 70mV drop seems kinda high to me.

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

Well, just after saying this I had this brief issue on the forum... with Twitch opened in another tab. I've already seen someone posting this issue though, especially related to Twitch. It's a screenshot, but the artifacts were intermittent.
 

wat artefatti.png

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

6 minutes ago, Agost said:

Well, just after saying this I had this brief issue on the forum... with Twitch opened in another tab. I've already seen someone posting this issue though, especially related to Twitch. It's a screenshot, but the artifacts were intermittent.
 

wat artefatti.png

 

Do you get artifacts with GPU Core at 1366 MHz, and Voltage on ... Auto?

I am thinking it is running above 1100 mV at times (e.g. 1150 mV) for 1366 MHz to be 100% stable.

 

Yes, VDROOP is normal ... as with any electronic device.

Once a load is applied, the voltage will take a small dip, hence, why Load-Line Calibration exists.

It is not so much of a problem with GPUs, but more critical with CPUs, and CPU overclocking -- why a lot of mid to high end motherboards have LLC settings in the UEFI.

Intel Z390 Rig ( *NEW* Primary )

Intel X99 Rig (Officially Decommissioned, Dead CPU returned to Intel)

  • i7-8086K @ 5.1 GHz (still tweaking) -- i7-6800K
  • Gigabyte Z390 Aorus Master -- ASUS X99 Deluxe
  • Sapphire NITRO+ RX 5700 XT Special Edition -- 2x Sapphire NITRO R9-Fury in Crossfire
  • 32GB G.Skill TridentZ DDR4-3000 CL14 (16GB TridentZ RGB + 16GB Red/Black TridentZ)
  • SanDisk 480 GB SSD + 1TB Samsung 860 EVO + 1TB Intel 660p NVMe
  • EVGA SuperNOVA 850W P2 + Red/White CableMod Cables
  • Phanteks Enthoo Luxe Tempered Glass Edition
  • Ekwb Custom loop
  • Logitech G502 Proteus Spectrum + Corsair K70 (Red LED, anodized black, Cheery MX Browns)

AMD Ryzen Rig

  • AMD R7-5800X
  • Gigabyte B550 Aorus Pro AC
  • 32GB (16GB X 2) Crucial Ballistix RGB DDR4-3600
  • Gigabyte Vision RTX 3060 Ti OC
  • EKwb D-RGB 360mm AIO
  • Intel 660p NVMe 1TB + Crucial MX500 1TB + WD Black 1TB HDD
  • EVGA P2 850W + White CableMod cables
  • Lian-Li LanCool II Mesh - White

Intel Z97 Rig (Decomissioned)

  • Intel i5-4690K 4.8 GHz
  • ASUS ROG Maximus VII Hero Z97
  • Sapphire Vapor-X HD 7950 EVGA GTX 1070 SC Black Edition ACX 3.0
  • 20 GB (8GB X 2 + 4GB X 1) Corsair Vengeance DDR3 1600 MHz
  • Corsair A50 air cooler  NZXT X61
  • Crucial MX500 1TB SSD + SanDisk Ultra II 240GB SSD + WD Caviar Black 1TB HDD + Kingston V300 120GB SSD [non-gimped version]
  • Antec New TruePower 550W EVGA G2 650W + White CableMod cables
  • Cooler Master HAF 912 White NZXT S340 Elite w/ white LED stips

AMD 990FX Rig (Decommissioned)

  • FX-8350 @ 4.8 / 4.9 GHz (given up on the 5.0 / 5.1 GHz attempt)
  • ASUS ROG Crosshair V Formula 990FX
  • 12 GB (4 GB X 3) G.Skill RipJawsX DDR3 @ 1866 MHz
  • Sapphire Vapor-X HD 7970 + Sapphire Dual-X HD 7970 in Crossfire  Sapphire NITRO R9-Fury in Crossfire *NONE*
  • Thermaltake Frio w/ Cooler Master JetFlo's in push-pull
  • Samsung 850 EVO 500GB SSD + Kingston V300 120GB SSD + WD Caviar Black 1TB HDD
  • Corsair TX850 (ver.1)
  • Cooler Master HAF 932

 

<> Electrical Engineer , B.Eng <>

<> Electronics & Computer Engineering Technologist (Diploma + Advanced Diploma) <>

<> Electronics Engineering Technician for the Canadian Department of National Defence <>

Link to comment
Share on other sites

Link to post
Share on other sites

5 minutes ago, -rascal- said:

 

Do you get artifacts with GPU Core at 1366 MHz, and Voltage on ... Auto?

I am thinking it is running above 1100 mV at times (e.g. 1150 mV) for 1366 MHz to be 100% stable.

 

Yes, VDROOP is normal ... as with any electronic device.

Once a load is applied, the voltage will take a small dip, hence, why Load-Line Calibration exists.

It is not so much of a problem with GPUs, but more critical with CPUs, and CPU overclocking -- why a lot of mid to high end motherboards have LLC settings in the UEFI.

I'm currently at 1340MHz and 1.1V, no artifacts in game... but it just happened with Twitch active in another tab (seen from another sapphire 580 owner too)

The point is: why can it keep higher voltages at stock while dropping so much while undervolting? Is there a way to make it stay at 1.1V on every load? The power limit is already maxed out at 30%, don't know if I can push it to 50% in some way (but GPU-Z and Afterburner tell me it's never going over 140W on the core with instant peaks of 160W...)

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

8 minutes ago, Agost said:

I'm currently at 1340MHz and 1.1V, no artifacts in game... but it just happened with Twitch active in another tab (seen from another sapphire 580 owner too)

The point is: why can it keep higher voltages at stock while dropping so much while undervolting? Is there a way to make it stay at 1.1V on every load? The power limit is already maxed out at 30%, don't know if I can push it to 50% in some way (but GPU-Z and Afterburner tell me it's never going over 140W on the core with instant peaks of 160W...)

 

Okay, is the GPU stable with NO undervolting, NO overclocking/underclocking?

That would be... Step #0.

If it is not stable at BONE stock settings from the factory, then that is a problem.

 

Power Limit only allows the GPU to pull more power from the PSU.

It does not affect the voltage -- they are, to an extent, independent from one another.

 

When you undervolt, you could affect how some of the transistors and/or electrical components on the PCB function.

You might have some components that  gets biased (triggered on) at ... 1090 mV.

If it drops slightly below, say VDROOP  happens, and it goes down to 1065 mV for 1/16 th of a second, that component is no longer being biased ... when it SHOULD be. If that component is a critical part, your GPU is now unstable.

Intel Z390 Rig ( *NEW* Primary )

Intel X99 Rig (Officially Decommissioned, Dead CPU returned to Intel)

  • i7-8086K @ 5.1 GHz (still tweaking) -- i7-6800K
  • Gigabyte Z390 Aorus Master -- ASUS X99 Deluxe
  • Sapphire NITRO+ RX 5700 XT Special Edition -- 2x Sapphire NITRO R9-Fury in Crossfire
  • 32GB G.Skill TridentZ DDR4-3000 CL14 (16GB TridentZ RGB + 16GB Red/Black TridentZ)
  • SanDisk 480 GB SSD + 1TB Samsung 860 EVO + 1TB Intel 660p NVMe
  • EVGA SuperNOVA 850W P2 + Red/White CableMod Cables
  • Phanteks Enthoo Luxe Tempered Glass Edition
  • Ekwb Custom loop
  • Logitech G502 Proteus Spectrum + Corsair K70 (Red LED, anodized black, Cheery MX Browns)

AMD Ryzen Rig

  • AMD R7-5800X
  • Gigabyte B550 Aorus Pro AC
  • 32GB (16GB X 2) Crucial Ballistix RGB DDR4-3600
  • Gigabyte Vision RTX 3060 Ti OC
  • EKwb D-RGB 360mm AIO
  • Intel 660p NVMe 1TB + Crucial MX500 1TB + WD Black 1TB HDD
  • EVGA P2 850W + White CableMod cables
  • Lian-Li LanCool II Mesh - White

Intel Z97 Rig (Decomissioned)

  • Intel i5-4690K 4.8 GHz
  • ASUS ROG Maximus VII Hero Z97
  • Sapphire Vapor-X HD 7950 EVGA GTX 1070 SC Black Edition ACX 3.0
  • 20 GB (8GB X 2 + 4GB X 1) Corsair Vengeance DDR3 1600 MHz
  • Corsair A50 air cooler  NZXT X61
  • Crucial MX500 1TB SSD + SanDisk Ultra II 240GB SSD + WD Caviar Black 1TB HDD + Kingston V300 120GB SSD [non-gimped version]
  • Antec New TruePower 550W EVGA G2 650W + White CableMod cables
  • Cooler Master HAF 912 White NZXT S340 Elite w/ white LED stips

AMD 990FX Rig (Decommissioned)

  • FX-8350 @ 4.8 / 4.9 GHz (given up on the 5.0 / 5.1 GHz attempt)
  • ASUS ROG Crosshair V Formula 990FX
  • 12 GB (4 GB X 3) G.Skill RipJawsX DDR3 @ 1866 MHz
  • Sapphire Vapor-X HD 7970 + Sapphire Dual-X HD 7970 in Crossfire  Sapphire NITRO R9-Fury in Crossfire *NONE*
  • Thermaltake Frio w/ Cooler Master JetFlo's in push-pull
  • Samsung 850 EVO 500GB SSD + Kingston V300 120GB SSD + WD Caviar Black 1TB HDD
  • Corsair TX850 (ver.1)
  • Cooler Master HAF 932

 

<> Electrical Engineer , B.Eng <>

<> Electronics & Computer Engineering Technologist (Diploma + Advanced Diploma) <>

<> Electronics Engineering Technician for the Canadian Department of National Defence <>

Link to comment
Share on other sites

Link to post
Share on other sites

5 minutes ago, -rascal- said:

 

Okay, is the GPU stable with NO undervolting, NO overclocking/underclocking?

That would be... Step #0.

If it is not stable at BONE stock settings from the factory, then that is a problem.

 

Power Limit only allows the GPU to pull more power from the PSU.

It does not affect the voltage -- they are, to an extent, independent from one another.

 

When you undervolt, you could affect how some of the transistors and/or electrical components on the PCB function.

You might have some components that  gets biased (triggered on) at ... 1090 mV.

If it drops slightly below, say VDROOP  happens, and it goes down to 1065 mV for 1/16 th of a second, that component is no longer being biased ... when it SHOULD be. If that component is a critical part, your GPU is now unstable.

Well, the card IS stable at least on my tests. I only get some artifacts sometimes and HWInfo shows memory errors, but there's nothing on the display when they happen. However I'd like to have a more stable voltage in order to maybe get a little more frequency at the same voltage settings, since I'm pretty sure I can go higher with stable 1.1V or something like that (it's even quite stable at 1080mV, but the drops get seriously low, like 1018 mV)

I thought it could be the power limit in case the voltage drop was be related to some kind of power throttling. Moreover I know that "memory voltage" in Wattman is actually a floor for vcore and can't go lower than that, but the drops are still lower (e.g. if i set 1100mV core and 1050mV memory, it should never go below 1050 but still drops to about 1030)

Fun fact: I've never actually kept it at stock voltage since the undervolt makes it go at constant clock under load, while auto voltages yield more like fluctuating 1250-1350MHz

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

On 11/16/2018 at 10:12 PM, -rascal- said:

 

Okay, is the GPU stable with NO undervolting, NO overclocking/underclocking?

That would be... Step #0.

If it is not stable at BONE stock settings from the factory, then that is a problem.

Well, I've just tried the maps test and strangely I had no artifacts at stock, I'll have to test it a bit more. May also be an issue solved via drivers.

However, the voltage peak was 1.19V just under maps, I'm a bit uncomfortable with that. I may try different settings these days, perhaps starting again from scratch for my undervolt

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

Another update, I suppose. Memory errors also happen at stock settings, at about the same rate that with my mild undervolt. I guess it's a very unlucky card

amd gib us navi pls

On a mote of dust, suspended in a sunbeam

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×