Consistent means a high confidence result, something like 90% chance that the 7900xt beats the 4070 ti in a randomly chosen game, that's how consistent is interpreted
Btw I looked at power efficiency more closely. This is total system power (again all hub)
4k
Dying Light 2: 18% more power
Doom: 6%
Hitman 3: 6%
1440p
Cyberpunk: +22%
Depending on the resolutions you are aiming this may not be a marginal difference. Though more data is needed to settle this conclusively
By and large we are in agreement that the cards are evenly matched from a performance perspective. My numbers and consistent statements are based on an overall average. In general there is a statistally high liklehood of a 7900 XT being faster in any randomly chosen raster game, and the 4070Ti being faster in any randomly chosen RT game. Though thre will of course be outliers/exceptions. Power consumption again needs to be taken at an average, or tpyical gaming, idle load etc but the 4070Ti wins here.
So for most gaming and use cases the 4070Ti will consume less power but not a large enough amount (IMHO) to sway
my (or indeed most peoples ) decision.
The 4070Ti will be faster in RT in general and considering both need DLSS/FSR at 4K, this is moot IMHO
The 7900 XT will be faster in raster for the majority of cases and this IMHO is the most important metric at 4K.
So while I prefer the 7900 XT based on my needs, others may prefer the 4070Ti. So as you have correctly asserted yourself, these are evenly matched GPUs and pricing for both is dire.