Was going to respond to this earlier, but wanted to avoid offending you and sounding like a broken record. So here's how I'll respond:
70C, you're right, it'll be mostly harmless. However, 80C is where silicon starts to get "iffy" with is semi-conductor properties. Electron migration begins to be a problem as well as heat induced electron pathway degradation. Saying that 80C is acceptable depends greatly on what "acceptable" means to someone personally.
Me? I think 75C is barely acceptable. I strive to make sure the temps on my system, except for VRMs, stay below 70C, ideally 60C as a ceiling. So this card with it huge heatsink and triple fans will maintain a temperature threshold that I would consider acceptable.
Meh. Electron migration isn't an issue unless you're pushing clocks and voltages very high - much higher than modern boost algorithms will do, and much higher than the firmware/driver limitations of these cards will allow. I've also never, ever heard of a GPU's silicon failing from running in the 80s. Other faults can occur if QC or production quality is too low - solder joints breaking from thermal expansion/contraction cycles; PCBs warping; that kind of thing. But seriously, show me any hard evidence of reasonably modern high end hardware failing from running in the 80s, please? I'd love to see it, but until I do, I'll trust the tJMax specs. If anything, the recent addition of hotspot temperature measurements has shown us that the temperatures we've been looking at for years are 10-20 degrees lower than what parts of the die are running at, and things have still worked fine. AMD specs their GPUs for up to 110 degrees hotspot temperature, and I don't see that as an issue. Intel specs their CPUs for 100 degrees, but allows for increasing this to 115 - at the user's risk, of course. The point being, 80 degrees does not cause meaningful electromigration in a stock or moderately overclocked chip over its useful life time. Unless you can provide significant data to prove otherwise (and, crucially, this data must show actual silicon failures rather than undiagnosed "my GPU died" failures), this is a safe assumption to make.
And, again, we aren't looking at those temperature ranges in the first place. I showed you a mid-range dual fan cooler that can keep a more power hungry card well below 80.
it's probably a triple fan card because they have too many left overs from other RDNA2 cards nobody wants.
Given that RDNA2 cards are just as out of stock as Ampere cards, that's quite unlikely.
)
