RTX 2070 are these artifacts?

[Mazer]

Hello I have a RTX 2070 that I been using for a few years things been fine it does run a little hot after I repasted the card the hotspot does get like 20c higher than the core temp. The only thing I worry is it's not holding up good anymore and I have two games that when really looking at decals (blood, wet surfaces) I get these weird flickering details that I don't see in any other games and by that, I mean older like half life 2 or bioshock or even metro exodus so I could rule out DLSS games, I also don't usually have any upscaling on however it does show them more pronounced as the wet puddles in robocop for instance, have all these weird glowing lights all over them it did make me worry a lot that something wasn't right, I did update my driver and game and it helped a little bit but I still get strange glowing decals for some reason. I added two videos to showcase them ( and yes I know hotkeys I was trying to record quickly lol) I want to know are these bad in any way is my 2070 dying or is it something else?

Edit: I'm so sleep deprived it's a 2070 not 2080...

 

[Lew Zealand]

I get similar artifacts on Nvidia GPUs when I OC my VRAM too much. I also have a single game which shows consistent shading (could be decal?) errors similar to what you show but it's just the one game and it's consistent. Maybe the shaders need to be recompiled for that one game or two in your case? This is something I haven't done but I believe you can dump the shader cache in the Nvidia Control Panel. I'd wait until someone more familiar with this issue stops in though.

 

[Carlyle2020hs]

Artifacts as in "my ram is at its end"?
What i see is a software based glitch that melds into the scene.
Only showing if specific code is triggered.

ram artifacting rather resembles big blocks like this in the same color.

 

[Mazer]

Artifacts as in "my ram is at its end"?
What i see is a software based glitch that melds into the scene.
Only showing if specific code is triggered.

ram artifacting rather resembles big blocks like this in the same color.
View attachment 342775

Right I'm thinking my vram is bad.

 

[Waldorf]

when i experienced issues on a gtx 580, i swapped it out for an Arctic cooler,
that included cooling the vram.
it took another +2y of gaming to finally go out.

not sure what options there are for your card, and if price is worth it.
tried contacting mfg?
you never know, they might have some on the shelf, and are willing to "help out".

usually only costs the time to contact them..

edit: using any ocing on gpu/vram?

 

[RJARRRPCGP]

Looks like core clock too high on the second one. Especially that, and if you start getting an error pop-up about Direct3D failing to initialize, even when Direct3D was initialized already, in the middle of a session. Or it crashes with no-signal message from the monitor, randomly and frequently! The ones I mentioned, are classic GPU core instability symptoms. Especially blocks like shown above, but they don't have to be one color. (raster mess, like a bad DTV signal, is classic of GPU core instability) (Very often sprinkles of squares with unstable GPU core!) (And it can hard lock right after the squares pop up)

You should check with Superposition 1080p Extreme. If it hard locks, the GPU core is most likely done for, if at stock!

 

[Mazer]

I did try to undervolt once but wasn't successful and instead left it at stock 1215 core, 1750 memory.

@Waldorf I tried to contact gigabyte never heard back from them

@RJARRRPCGP I never get any errors or black screen when I play any games like that. I did switch between windows mid game so it might of been off for a second.

 

[RJARRRPCGP]

I did try to undervolt once but wasn't successful and instead left it at stock 1215 core, 1750 memory.

@Waldorf I tried to contact gigabyte never heard back from them

@RJARRRPCGP I never get any errors or black screen when I play any games like that. I did switch between windows mid game so it might of been off for a second.

You got a DTV-like-raster-mess, so I suspect unstable GPU core!

 

[Mazer]

I did superposition benchmark like you said the core kept 1875mhz @ 211w nothing locked up or crashed and it looked fine.

 

[DaemonForce]

It looks like a very specific tessellation issue I get in VR outside a very specific voltage parameter and above a specific temperature. Your memory is getting cooked and your core is probably just outside of the window of voltage needed to remain stable. What happens when you raise the fan speed?

 

[Mazer]

It looks like a very specific tessellation issue I get in VR outside a very specific voltage parameter and above a specific temperature. Your memory is getting cooked and your core is probably just outside of the window of voltage needed to remain stable. What happens when you raise the fan speed?

Nothing I have it set to constant 65rpm ( I never hear it since I'm far away from my case) the hotspot temp is the only thing I see over 80c when the core is at 74c and having no memory sensor to tell the memory temperature doesn't help.

Here is another factor: I can't overclock either the core or memory above 50+mhz maybe it's degraded from heat and voltage over the years? Is it game related? I have no idea.

 

[DaemonForce]

That's not good. Do you see GPU memory errors popping up in HWiNFO64?

 

[Mazer]

That's not good. Do you see GPU memory errors popping up in HWiNFO64?

Where in HWiNFO64 would I find that out this? I ran a OCCT vram test and it passed for 2 hours+

 

[DaemonForce]

Exactly where you would expect it to be. GPU category, last item.

 

[Mazer]

Exactly where you would expect it to be. GPU category, last item.
View attachment 342794

How do you enable that?

 

[DaemonForce]

Layout, restore original order? You can also add it as a tray icon. I know everyone hides a bunch of sensors because there's a lot of them and it clutters the screen.

 

[Mazer]

I don't have anything else in the GPU column after PCIE link speed.

 

[Mazer]

Well I just wanted to post an update here

I changed out the 2070 for a 4060ti and then a 6600xt for testing to see if anything goes away and to my surprise, nothing different with any of those flickering shadows in robocop or the water particles in other games I seen it in, horizon forbidden west I saw the same pixilation in the waterfall in the beginning of the game, no matter DLSS or FSR was enabled, so I reinstalled the games even used different SSDs. Nothing changed same old flickering, I don't know how to get rid off that or what else I can do.

All I know is that something is cursed in my machine.. haha.

 

[Carlyle2020hs]

Please reset you monitor settings and test again.

 

[Vya Domus]

Not all cards expose memory temperatures, if it doesn't show up in hwinfo then it doesn't exist.

It's hard to tell but this just looks like something specific to those games, some of it just looks like specular aliasing, very common in games. You said you repasted and the temps are worse, so something is not right, it's possible not all of the die makes proper contact hence the artificating, I recommend repasting again, possibly using ptm7950, some of these gigabyte coolers are really shitty.

 

[Deleted member 239770]

Is it because they are all pirated?

 

[Mazer]

Is it because they are all pirated?

No, I buy most of my games on GOG or steam if theres a sale.

 

[Deleted member 239770]

No, I buy most of my games on GOG or steam if theres a sale.

It looks to me like it could be driver related.

 

[Beginner Macro Device]

All I know is that something is cursed in my machine

Could easily be your display cable. Get a 100% working one instead, maybe check a different monitor.

 

[kiddagoat]

That looks like a driver/shader issue. When your GPU Core or VRAM is starting to bite the dust, you will see pink or green blocks. That's indicating a hardware issue. It's been a minute since I have encountered this issue, but any of my Nvidia, AMD/ATI cards that have died, started with pink or green Blocks. If I recall correctly, the green ones are core faults, and the Pink are memory related.

I would turn off any overclocking and if you have the means to, lock the GPU clocks to their specified defaults.