TheF34RChannel
New member
The 2080 is also physically a larger chip than the 1080Ti by 529mm to 471mm squared. The RTX 1080 is NOT a midrange chip from the way I see it. It is a large die, it has many of the advanced features of the flagship GPUs, and costs more than or around the same as the previous flagship. It also has a high TDP.
From the way I see it, Nvidia don't intend on reducing prices. The 'only' thing they've done is, swap the 1080Ti with the RTX 2080 and added Tensor cores and RT abilities. The clock speeds are similar, the prices are the same, the die size is similar, the TDP isn't far off. In other words, Pascal performance per dollar is going to stick around. It's not like previous architectures where you get more performance for less money (970 was as fast as a 780Ti but for less money, 770 was the same chip as the 680 but cheaper, etc). Now, what Nvidia have done is, keep the same performance per dollar, thus shifting everything up a notch, and added features you won't benefit from for a few years. The 2080 quite literally replaces the 1080Ti. Performance per dollar doesn't change. It's almost like Turing doesn't exist. It's just adding another layer of sugary fat to the cake that was already there. It's not a new cake that's better than the previous one. They've just added another layer of flavour.
Someone over at the AnandTech forum also made a valid point by counting production cost; Turing cannot be cheap to produce.