Nvidia really has great marketers and market research/analytics.
They really nailed the price on these cards and looking at the performance graphs, there looks to be a great uplift gen over gen* (not as big as Pascal to Turing when including raytracing and definitely much less when including tensor cores DLSS) but they came correct with the pricing.
A $499 2080Ti equivalent card is unreal. ~2X 2080 performance at 2080 Super price is great too. But they are kind if blinding us (the tech people) with these types of comparisons.
These cards use a lot of power.
Look at the amount of shader cores they had to put on the card for these performance increases, look at their stated compute difference in TFs and ML for lower real world performance. The ampere design doesn’t seem...efficient and it isn’t bringing anything new to the table.
Oh, and the reference cards are large.
But, because its cheap, Nvidia knows that people will eat it up.
Personally, Im most impressed when I see actual architecture improvements vs sheer brute force/brute power usage.
*Why is Nvidia marketing this card as the biggest leap ever when Pascal running an RTX game vs an equivalent Turing card running an RTX game can see over 100% improvement. Technically isn't that the biggest leap?