How should I test a used graphics card for full functionality? Advice needed

Hello nice people. I just purchased a used GTX 1070 founders edition and I need some advice. I just couldn’t pass up a 1070 for 220 on ebay. It felt like a great deal and I was still run a GTX 670 in my desktop.
So what stress test should I run on my new used card? Is there any process you do when you purchase new hardware? Should I take off the cooler and replace the thermal paste? Should I also replace the thermal pads? Any advice would be much appreciated. I just want to make sure my purchase is operating perfectly.

I wouldn’t take it apart just yet, as that might impact your ability to return it if it is defective. I would first run it through a benchmark loop, like running the Valley benchmark continuously for several hours and monitor your temps and check for video artifacts.

Some would suggest Furmark for a stress test, but I find it to be unrealistically hard on your GPU. I’d stick with a mainstream benchmark that you can loop, like Valley, Heaven or the paid version of 3DMark.

If you find the temps getting too hot, or you lock up, then you can think about repasting. But if you disassemble it, the seller might not accept a return. Personally, I’d just return it if it doesn’t perform as expected.

3 Likes

I woul suggest that if you do decide to re-paste, do replace the pads as well, even if only the first time you do so

Good luck on your new card. I got a similar deal on the same card.

One word of advice. The founders edition blower fan gets really loud! I slowed my fan down so it runs a little hot but still well under max allowable temps. It’s been rock solid reliable and I didn’t redo the paste.

First and foremost thank you for your advice. I will run it through the Valley benchmark continuously for several hours as imrazor has suggested before I try and pull the card apart. I have ordered new thermal pads as trooperish has suggested just in case I decide to pull it apart after testing. I would defiantly adjust the fan speed as positron suggested but I have a 1u super micro server running 24/7 in my computer room. I could run the fan full speed and not even hear it over that lovely server room sound. Thank you again for the support and feedback. As soon as I get some time I will run the tests and update you all.

im guessing the card was used as a mining card before you bought it.
Also i’m guessing you’re going to use it for gaming.
Id suggest firing up steam and try it out, and if it works, enjoy it, with some silky smooth 1440p gaming.
maybe run a 3d mark or two on your gpu, but most likely it wont find anything a game wont find, but if anything you’d get your epeen on, which is important when buying new hardware.
Burnin testing really is most for enterprise where downtime can be damd expensive, and therefor you want to find defective hardware before putting it into production.
When a card is used for intensive computing over a prolonged time, it doesn’t usually mean that it becomes defective, but rather the wear and tear of it is higher then usual, and thus it’s life span shortens.
And more stress testing will only shorten the life span more, but furmark does put your card through some trials, but again if you’re “just” gaming, “abit” redundant.

UpDate
I ran the benchmarks for 30 minutes and it was stable and temperatures were great. Had an amazing time playing fallout with everything cranked up. Thank you for all your advice the card seems perfect. I am a Happy Boy.

u1

Glad you got it sorted out. If you do decide to go ahead and re-paste, you might want to gather up some benchmark results and temps to set a baseline and see how much difference the new paste job makes.

Just wanna drop a line in here.

30 minutes isn’t enough, imo.

My general benchmarks for stuff I build and sell is 4 hours, maybe 2 hours minimum. My reason is to check for prolonged thermal temperature control. I’ve found that issues really tend to show after the 2-3 hour mark, and its where I’ve found a lemon or two in my time. Artifacts can appear later rather than sooner when testing.

I played 5 hours straight at 4k res and it ran perfectly. the CAM software said 100% utilization the hole time. I don’t need to run a stress test after something like that right ?