|
Guys, me, being a power electronics engineer who specialises in power circuitry can safely say there is no precise openly shared data on every component on that PCB. It's commercial knowledge thats well guarded and typically datasheets have simplified models that don't thoroughly explain cause and effects etc. So it's hard to estimate the lifetime of each and every part on that PCB even if we were to meticulously search for their datasheets. What I can safely say is that the way NVIDIA would manage this by:
a) having loads of engineers simulating PCBs on 3D softwares to figure out temps on each component and
b) to specify a lifetime to each manufacturer and set some legal clauses to chargeback to the parts manufacturers for EARLY failures.
The second part is more significant. Nvidia can promise customers say 3 years warranty. Nvidia will then specify the parts manufacturer to have a lifetime of 3+1 years @ 85 Celc. Next, the parts manufacturer's design teams will design for 3+1+1 years @ 85 Celc to safeguard its company's ass.
Now, there's a tradeoff of lifetime vs cost. The longer its lifetime, the more expensive it'll be. A business is a business and if Nvidia provides 3 years warranty, I expect the cards to mostly last up to 4 years if ran at our example of 85C.
Running it at a lower temp will raise its lifetime as its less stressed mechanically but I don't have the ballpark figures to say what it'll be (I know it well for PSU components). We have to get an Nvidia / AMD engineer to elaborate on that.
|
Show Posts
