Maybe the article has been corrected, but right now it says...So TU106 is used in the 2070, 2070 Super, and 2080 and the TU104 is used in the 2060, 2060 Super, and 2070. Did I get that right?
There's a line that reads, "Keep in mind that the TU104 die is a much larger chip than the TU106 and houses up to 3,072 CUDA cores. " Why is the TU104 a larger die when its meant to drive smaller graphics cards? Was this a typo?
Well, yeah, that's what Intel, AMD and nVidia have been doing for Decades. Imperfect but functioning chips get binned as lower tier products. The only change here is that the binning for TU104 is being extended one tier lower than previously. Yowser, how wierd...I think it's the use of imperfect TU104s that will come in here? I guess?
All I could think when reading this is to kind of wail "Dammit, Nvidia! Just because AMD is doing weird stuff with the 5600, don't YOU guys start with the craziness!"