Is my new GPU dying?

I’m pretty sure my new GPU is faulty but I’d like confirmation of my diagnosis.

First of all, my system specs are as follows-

CPU: AMD Ryzen 7 1700x
Motherboard: Gigabyte AX370 Gaming 5
RAM: 2x 16GB Corsair Vengence
PSU: Corsair AX760 (760 watts, 80+ Platinum)
Old GPU: Asus GTX 670 DirectCU II TOP
New GPU: MSI 1660 Super Gaming X

I should note that I’ve never had any issues with any of the hardware listed above before I installed the new GPU. Aside from the new GPU, everything has passed burn-in testing, the RAM has went through about 30 passes with MemTest and the PSU’s voltage and power good signal were all within spec the last time I tested them.

I installed the new GPU yesterday afternoon, along side my 670, and everything worked just fine for the rest of the day. I updated the GPU drivers on both Debian and Windows to the latest stable releases but I don’t think the driver is the issue. Not least because I played a couple of games without issues (albeit games that even the 670 could handle @ 60hz). I also did an hour or two of modelling using Sketchup, along with various web browsing and trying to download CoD: Warzone. Everything was completely as would be expected.

The one issue I had is I didn’t have a decent Displayport cable for the new 1660s, so I’ve been using HDMI on my 1440p (primary) monitor, which looks like hammered shit, and a crappy Displayport cable from the 670 to my secondary monitor.

Earlier today, whilst I had Sketchup, Firefox (with a few hundred tabs) and the Battle.Net client open and downloading, my system froze, black screened and power cycled. When the system froze, I could still move the mouse cursor around but non of the open applications responded to interaction. I didn’t have time to try anything else before the system restarted itself. During boot up, green lines appeared all over the screen (the first picture), even in the BIOS, and the BIOS, the GRUB menu and the Windows login screen all looked garbled and had the same green lines everywhere. As the lines appeared in the BIOS, I believe that that probably indicates that the driver isn’t the issue.

At this point, I think I may have checked that the HDMI cable was connected proplerly but I’m not 100% certain I did.

Either way, one of the next steps was to try booting with just the 1660s installed. I rebooted and (after an "oh shit, the GPU is completely dead!.. nope, just forgot the PCI power cable moment) I got the same green artifacting.

I rebooted again, this time with the 1660s in the second PCIe x16 slot… same deal. I’m pretty sure that this was when I took the second picture, as seen just below.

Finally I risked it and tried using the potentially crappy Displayport cable I have. It did appear to fix the issue, I logged into Windows, open the same three apps backup and then even more severe artifacting appeared on the screen. I swear to Wendall the artifacts looked like frickin’ Space Invaders! I wasn’t quick enough to get a picture though.

It was about that time I powered the system down via the power button, removed the 1660s, reinstalled the 670, booted back up and about half an hour later of typing this post, it’s been completey as it was before the 1660s came along. I still have way too many tabs open, Sketchup is still sluggish on the 670, and the Warzone download is still a pain in the arse.

So… is there any reason to believe that the 1660s isn’t faulty or dying?

While I’m not an expert on hardware, I would definitely come to the assumption that the GPU is, in technical terms, dead.

It’s for sure in warranty, so I’d contact the Manufacturer and get a replacement.

3 Likes

Basically the only variable is the GPU, the old one works, the new one doesn’t.

Case closed.

2 Likes

I guess I’m just kind of paranoid I’ve missed some obvious, innocuous cause.

Perhaps it’s just pining for the fjords…

I decided to err on the side of caution though and I’ve contacted the retailer and asked for an RMA, which is the standard process in the UK.

2 Likes

100% you have a faulty GPU, looks like bad RAM on it or the LVDS interface has failed. RMA it, you should have no issues.

1 Like

Before I conclude that I’m an idiot… is it possbile that the cause of the problems could have been a slightly loose PCIe power cable?

Why, did unplugging and plugging it in again fix it?

If so, sweep it with tests and see if anything looks wrong or goes wrong.

At first I thought it did.

I’ connected the new GPU using a cable that came with my PSU but hadn’t been used before installing the 1660 Super. Whilst testing the theory that the cable or the PCIe / Peripheral inputs on the PSU unit itself might be faulty, I noticed that the little clip on the 6+2pin PCIe cable that hold the additional 2pins in place, was bent. Meaning that the large clip on top of the cable, that clips onto the GPU, could clip into place without the 6pin section being fully inserted.

Normally I would notice the cable not being fully inserted as I usually check for a gap between the cable and the socket on the GPU by trying to run my finger nail between the two (which is why I noticed the cable wasn’t fully inserted this time).

I made sure to fully insert the cable and then I performed some stress tests with FurMark… probably an hours worth in all, between using the system normally whilst diagnosing my now dead firewall appliance.

The system performed normally for two days, which lead me to conclude that even though it would have meant inserting the cable with the bent clip, and the old reliable cable, wrong around 10 times or so, it would seam to indicate that it was user error.

Only that’s all out of the window again as the system has started power cycling today. So I tested my PSU with my ThermalTake Dr Power II and all the voltages and the PGS are within spec. I’ve now removed the card, plugged my display cables back into my old GPU and everything is back to normal (touch wood) and to the hardware configuration that I’ve never had any issues with.

If it’s not the GPU, then I’ve got no idea what else I can feasibly test.

It sure does some like it is the GPU then. can you get it RMA’d still?

This topic was automatically closed 273 days after the last reply. New replies are no longer allowed.