What can cause artifacting without overheating?

[NewbieOneKenobi]

280x (Sapph Dual-X) with i5-6600, Asrock K6, approved RAM. Several hours into a long gaming session CPU temps 40-ish, 50 is rare. System several degrees less (40 tops). GPU in low 70s (don't think it's even capable of hitting 75 no matter what I do to it and how long), even after some time under 100% load. GPU is not OCed beyond default 1020 MHz detected by the driver. Mem is not overclocked; I keep it down to 1350 from 1500 in fact. Still got some corruption in fancy light, weather or similar effects, or even texture corruption, some time into the game, as if some kind of fatigue that isn't exactly connected to the heat. In theory, VRAM cooling shouldn't be an issue. Saph's Dual-X cooler covers the whole card, so mem shouldn't be overheating without OC. Still, something's causing it. Haven't seen it before, when playing less demanding games and hitting similar temps. What else could be the problem?

Meanwhile I could add a 12cm floor fan under it just to see if keeping the temps a couple degrees down won't solve the problem. Not sure if blowing upward in the opposite direction to the card's own exhaust flow will do much good, though, with or without roof exhaust added, which I could also do. Could also put a 92mm exhaust through PCI slot holes, maybe this would do better? Case is Phanteks Enthoo pro, stock fans (20cm front intake, 14cm rear exhaust). Airflow may be poor-ish as the PC sits under the desk right in the corner of the room, so the rear of the case is close to the wall. But I still don't think temps are the issue.

 

[trents]

Buggy video card driver or buggy coding in the game?

What about temps for other non GPU components such as CPU and VRM?

 

[Tír na nÓg]

^Happened to me with VRM's overheating (was on a nVidia card though)

 

[NewbieOneKenobi]

Buggy video card driver or buggy coding in the game?

Hmm… Can't recall updating the drivers lately, but in any case looks like I'm on AMD's latest. What's strange is that even in the same game this particular type of artifacts hasn't really occurred before. Perhaps it might be a problem with the current location then. What's different about it is that it has a lot of weather and light effects: snow, wind, crystals, reflections etc. Some of it might be done through AA, from what I've heard, but I'm not sure. I've seen AA corruption before on cards or at least systems that couldn't handle the load.

What about temps for other non GPU components such as CPU and VRM?

CPU 30-ish, up to 40 in general, sometimes tops out at 50. MB/sys usually a couple degrees hotter than CPU but still not really above 40, and never close to 50. Can't see a specific VRM read in HWInfo, GPU-z or mobo utility, will have to keep looking.

^Happened to me with VRM's overheating (was on a nVidia card though)

I'll try undervolting it perhaps.

EDIT: The problem is getting worse and worse, and the artifacting can now be quite massive while still below 70C. Worried about the card, it didn't act this way before.

EDIT2: Looks like it does reach 70C, in places in the game that hadn't previously been a challenge. This is despite DC-ing the memory and undervolting the card a bit, as well as rolling down to previous drivers. So perhaps the card is pushing for early retirement. Alternatively I could remove the fan shroud and put some of my own stronger fans on the heatsink.

EDIT3: Upping power limit full way to +20% seems to have reduced the artifacts, if not temps. I'm rather sure fans work overtime compared to before, too.

This reminds me of a past card that occasionally had such issues despite never really getting too warm because of its beefy aftermarket cooling.

Visually, I'm pretty sure something's wrong with lights that wasn't before. It's not just the usual case of flicker, it really is lights being a little off.

Update: Upped power limit to the max, downclocked mem further and GPU slightly, set up a more aggressive fan profile. No problems so far, despite temperatures being roughly equal. I'll see where this takes me, but I fear I'll have to replace the card anyway.

Update again: Looks like my soundcard may be involved. If the temps are comfortably 55 or under on the GPU and 35 or under on the CPU, and there are some crackling sounds when alt-tabbing between the game and the desktop in addition to visual corruption, then I can't really think the soundcard isn't involved.

 

[Suppressor1137]

Sounds similar to my old gtx 460. Temps completely fine on sensors, but artifacts were harsh when the card experienced load. Turns out the vram was not being adequately cooled, and a reseat of the heatsink and added mini-heatsinks to various bits and bobs that had no protection fixed it.

Basically what I did. He even uses the same Thermal Compound that I did, the Gelid Solutions GC-Extreme haha. Just make sure that the heatsinks fit under the main heatsink before applying them.

 

[NewbieOneKenobi]

Sounds similar to my old gtx 460. Temps completely fine on sensors, but artifacts were harsh when the card experienced load.

Had similar symptoms on another 280X (Gigabyte Windforce), one I bought used from a guy who turned out to have been a front for his friend's business, selling defective stuff during the last period. I had it tested by a more capable person and the feedback I got was just the same thing I'd thought — power circuitry perhaps, and/or trouble with thermal sensors, can't tell. The card sometimes work fine but never achieved the full extent of the power it should have had, and it even managed to bench lower than my old HD4850 (theoretically not impossible with older DirectX versions depending on what each respective card's architecture was optimized for, but DirectX is not that old). I fear my current 280X (Saph Dual-X) may be suffering from the same condition. Incidentally, this makes me worry about the prospect of replacing it with a 480 too. Polaris is not Tahiti, but stil.

Turns out the vram was not being adequately cooled,

Hmmm… The 4850 sometimes experienced artifacting despite low temps on the GPU, and it had no direct VRAM cooling because the heatsinks had fallen off.

and a reseat of the heatsink and added mini-heatsinks to various bits and bobs that had no protection fixed it.

Hmmm… I should probably start from reseating the heatsink anyway. I have time today, so why not. Just worry about doing irreparable damage to a screw or two in the process, and the cooler may be hard to remove, but yeah, since the card just might be on its way out anyway…

Basically what I did. He even uses the same Thermal Compound that I did, the Gelid Solutions GC-Extreme haha.

Same as mine too, actually.

Just make sure that the heatsinks fit under the main heatsink before applying them.

Lol, yeah. Don't want to either bend the PCB or end up using tape or a thick layer of compound to bridge the difference.

 

[Suppressor1137]

You shouldn't have to worry about damaging the screws. I'd be more afraid of a static discharge. Use one of these when working on any electronic component; Its super cheap, and significantly lowers the chance of a catastrophic static discharge:

https://www.newegg.com/Product/Prod...JqCFqpQkpOBTI28tZwZWQaAmuj8P8HAQ&gclsrc=aw.ds

 

[NewbieOneKenobi]

News: It looks like my most recent problems were possibly connected with MSI Afterburner, which I'd used to deal with the initial problems. With Afterburner, I'd have artifacts at even 55C with fans roaring; without, it's closer to 65 with no issues most of the time, except rare problems. OCing the card oddly seems to have reduced the problems too.