AMD Desktop screen tearing

[iviv]

Running an R9 290, asus MH279 2560x1440 144hz freesync monitor connected over miniDP as my main monitor and an HannsG 1920x1200 60hz via DVI, windows 10 x64, catalyst 15.9.1 beta.

Since running this setup I have had some horizontal tearing on the Asus monitor, always in the lower quarter of the screen, but it has been infrequent enough that I have just ignored it. This has always been in the desktop rather than in a fullscreen game, can't recall if it happens in any windowed games, and I can't find anything specific which causes it, though it happens more with mouse movement, and rarely happens if the mouse isn't moving.

After reading this thread about freesync, I checked in catalyst and saw that I had only been running my desktop at 60hz, so I tried setting it up to 144hz in catalyst and the tearing became a constant flickering, as seen in this quick vid I shot:
https://www.youtube.com/watch?v=RzH-oIESd4E

So, any ideas what's causing this, and any known solutions?

 

[shankly1985]

First why you running Mini DVI? Freesync only works with DisplayPort cable.

Edit
Just seen video that's glitching not screen tearing.. Do you not have a DisplayPort cable? one should have come with the Asus monitor.

 

[iviv]

Fixed the post, it is indeed mini displayport. Too used to writing DVI instead of DP

 

[shankly1985]

Fixed the post, it is indeed mini displayport. Too used to writing DVI instead of DP

Ok do you have a link to your monitor am having a hard time finding it.

 

[drunkenmaster]

Are you running something like afterburner and have you turned down voltage for desktop settings or anything? 99% certain you'll find that is not enough voltage for the memory clock. I can literally induce that from having memory speed to high and voltage to low for it.

In general at completely stock settings what should happen is the card has (numbers vary by card/bios) lets say 0.9v idle and 1.15v load with 300/150 clocks idle and 1000/1000 clocks load. Often with dual screens(I have same setup except 120hz second screen rather than 60hz) the memory won't downclock as low. So the likelyhood is the card is downclocking to voltage required for the 150Mhz idle memory clocks but running at much higher memory speed.

First thing would be to install afterburner, get it setup to control voltage/overclocking then monitor voltage and see what it's doing at idle. Try lower memory clocks to see if that fixes it and then try raising the voltage a little. I tend to for simplicity save one profile for desktop, lowest voltage/clocks that maintain everything working fine for anything 2d, then 2-3 profiles for gaming at various overclocks/voltages as some games run hotter or cooler, some will run at higher clocks stably, others won't.

 

[shankly1985]

In the manual for Asus MG279Q ROG if this is your monitor. 144hz is saying DisplayPort but not mDP.. Is mDP and DP the same thing guys?

@OP can you try a normal DisplayPort Cable.

WQHD Mode
2560x1440 60Hz 88.787kHz
2560x1440(Only for DP) 120Hz 182.817kHz
2560x1440(Only for DP) 144Hz 221.97kHz

 

[iviv]

Ok do you have a link to your monitor am having a hard time finding it.

Asus and HannsG.

Not installed Afterburner on this windows install, never tried to overclock the graphics card or anything like that. Installed it, I'll report back tomorrow on voltages and so on (can't reboot now)

I don't have a normal DP cable, but I'll look into getting one if needed, would probably be good to have as a backup, but as far as I'm aware, there's no difference between DP and miniDP, just a different sized connector.
I'm using the DP cable that came with the monitor, for what that's worth.

Cheers for all the help so far guys!

 

[drunkenmaster]

It really shouldn't do it without screwing around with voltages yourself as I do but is very easily induced. if you play around with voltages, particularly lower voltages, you won't do any damage.

I actually can get higher performance out of xfired 290's by running a good deal less voltage. It brings temps down on stock cooled reference cards by some 10-15c which allows then to not throttle and even set a decent overclock with the reduced voltage. So instead of something like 1.23v under load and throttling to 950Mhz a lot of the time I run around -70mv giving around 1.15v at 1050Mhz. It's a bit picky, some games like a bit more voltage or crash or a little lower overclock, others let me go even higher. Dropping voltage at both desktop and in game drop power usage a bit but not enough to save major money, it's really temps/sustained clock speeds being the reason I use it.

But when I play around with profiles when I have the voltage too low for a given memory clock speed I get that glitching every time. If you actually set memory too high you get glitching like that all over the screen to the point you can't even see the mouse icon or do anything useful anymore... opps.

So 1000-1300Mhz type clocks with -70mv will give that level of glitching, -100mv with those clocks gives insane levels of screen artifacts, 500-800Mhz and -100mv is completely stable with no glitching at all. Dropping memory clocks a little for desktop or increasing voltage a very small amount 'should' get rid of all such glitches.

Hopefully anyway, it's also possible that the card is simply faulty though it's less likely that than just a marginally over aggressive voltage drop at desktop. A faulty card would likely do this under gaming but as it doesn't, when voltages will be up, it's 98% likely too low voltage.

 

[iviv]

It really shouldn't do it without screwing around with voltages yourself as I do but is very easily induced. if you play around with voltages, particularly lower voltages, you won't do any damage.

I actually can get higher performance out of xfired 290's by running a good deal less voltage. It brings temps down on stock cooled reference cards by some 10-15c which allows then to not throttle and even set a decent overclock with the reduced voltage. So instead of something like 1.23v under load and throttling to 950Mhz a lot of the time I run around -70mv giving around 1.15v at 1050Mhz. It's a bit picky, some games like a bit more voltage or crash or a little lower overclock, others let me go even higher. Dropping voltage at both desktop and in game drop power usage a bit but not enough to save major money, it's really temps/sustained clock speeds being the reason I use it.

But when I play around with profiles when I have the voltage too low for a given memory clock speed I get that glitching every time. If you actually set memory too high you get glitching like that all over the screen to the point you can't even see the mouse icon or do anything useful anymore... opps.

So 1000-1300Mhz type clocks with -70mv will give that level of glitching, -100mv with those clocks gives insane levels of screen artifacts, 500-800Mhz and -100mv is completely stable with no glitching at all. Dropping memory clocks a little for desktop or increasing voltage a very small amount 'should' get rid of all such glitches.

Hopefully anyway, it's also possible that the card is simply faulty though it's less likely that than just a marginally over aggressive voltage drop at desktop. A faulty card would likely do this under gaming but as it doesn't, when voltages will be up, it's 98% likely too low voltage.

Thanks again for the reply, much appreciated. Though reading through the AMD driver thread I found this post: http://forums.overclockers.co.uk/showpost.php?p=28778740&postcount=8327 which links to a massive thread on the AMD forum full of people with what seems to be the same problem, seems like a driver issue with Win 10 Someone mentioned they fixed it by setting the 2D clocks to match the 3d clocks so I'll try that tomorrow and see what happens, besides my electric bill shooting up. At least it's winter and the extra heat will be put to use!

 

[iviv]

It really shouldn't do it without screwing around with voltages yourself as I do but is very easily induced. if you play around with voltages, particularly lower voltages, you won't do any damage.

I actually can get higher performance out of xfired 290's by running a good deal less voltage. It brings temps down on stock cooled reference cards by some 10-15c which allows then to not throttle and even set a decent overclock with the reduced voltage. So instead of something like 1.23v under load and throttling to 950Mhz a lot of the time I run around -70mv giving around 1.15v at 1050Mhz. It's a bit picky, some games like a bit more voltage or crash or a little lower overclock, others let me go even higher. Dropping voltage at both desktop and in game drop power usage a bit but not enough to save major money, it's really temps/sustained clock speeds being the reason I use it.

But when I play around with profiles when I have the voltage too low for a given memory clock speed I get that glitching every time. If you actually set memory too high you get glitching like that all over the screen to the point you can't even see the mouse icon or do anything useful anymore... opps.

So 1000-1300Mhz type clocks with -70mv will give that level of glitching, -100mv with those clocks gives insane levels of screen artifacts, 500-800Mhz and -100mv is completely stable with no glitching at all. Dropping memory clocks a little for desktop or increasing voltage a very small amount 'should' get rid of all such glitches.

Hopefully anyway, it's also possible that the card is simply faulty though it's less likely that than just a marginally over aggressive voltage drop at desktop. A faulty card would likely do this under gaming but as it doesn't, when voltages will be up, it's 98% likely too low voltage.

http://i.imgur.com/4CD2W5Y.png

There's the readouts from Afterburner, about a quarter of the way in I switched up to 144hz, though there's no indication on any of the graphs that anything changed. Is this normal? Gaming clocks for reference.

 

[drunkenmaster]

From the looks of things under gaming it's mostly around 1.227v with 1350Mhz clocks, at idle it's mostly 0.98. Have you managed to get the voltage control option to work within MSI afterburner? I think in general it requires setting the overclocking limits option and the voltage control options in settings for a 290 though it can depend on what 290 you have.

If/when you get that working, try setting a 2d profile for when you're at desktop as aggressive in power saving as you want or don't want. You could up the voltage a tad or drop memory clocks. For desktop I put gpu clocks to lowest, memory clocks to 800Mhz and lowest voltage which works for me but may not work for you. Different cards, different memory can all have their own requirements.

Just to see if it is a voltage/clocks issue I'd say try memory anywhere from 800-1000Mhz and see if that works. Also try normal mem clocks and +25-50mv, if both get rid of the glitches then I'd say that is definitely the cause. If neither did then it could well be a faulty card.