I'm not a big fan of burn/stress tests on any PC hardware. You never know what's going to happen.
For video cards that I fix I usually run at least 3 loops of Unigine Valley, or for low-end cards I do the same with Heaven in DX9 mode 720p windowed. That's enough to see if something is wrong. Or simply - play a modern game for a 2-3 hour stretch.
For your card - Unigine Superposition or TimeSpy will do.
With stress-tests, like furmark, intense OpenCL workloads or crypto, you may occasionally see weird stuff happening to GPU, rarely - sparks and explosions.
One of the late examples is my very own ASUS GTX950. I found out the hard way that when GPU hits 80C, the fan starts spinning at 100% and does not know how to come back to normal even after complete cool down and system restart. Had to tweak it with afterburner for a few minutes to get the fan speed back to normal auto.
Another example from last year: GTX660 (I think it was an MSI Gaming, or something like that). The VRM design on this card was cutting close to it's limits (can't remember for sure, but I think it was something like 35-40A per phase on x4 GPU phases, while using some off-brand chinese FETs), when the previous owner had decided to overclock the card w/ modded BIOS. First run of Furmark, and the card went kaboom, taking out 2 phases. The PCB was intact, so I simply replaced all of the mosfets to higher-rated equivalents. The card was fine after that, and even the VRM temperatures dropped.