So after doing a bit of research on how Navi GPU (RX 5700/5700 XT)  perform, it look like Navi are doing great or good at least from AMD told us so far (we expect less or more slightly on real benchmark very soon)  , and RX 5000 will be far efficient on performance compared to GCN based GPU. So what does this means for the console comparison sake, it means we will get less compute performance (Teraflop number)  but better real world performance on Next Gen Console.

For example Nvidia RTX 2070 with 8 to 9 teraflop are perform better or equal than Vega 64 that run at 12,6 teraflop, but now RTX 2070 were beaten slightly by  RX 5700 or 5700 xt with almost the same compute performance. This means all this new GPU perform better under less teraflop than GCN based GPU like Vega/Polaris etc . 

PS5 are  rumored to be using Navi 10 which is RX 5700/ 5700 XT or slightly better. So probably we will see PS5 run at 2070 level or 1080 slightly better on GPU alone. So expect PS5 or Xbox Scarlet will have 36 to 40 CU or possibly 44 CU if it's using (7 nm EUV) on GPU compute unit and ended up 9 to 10 teraflop, far less than the rumored 12 or even 14 teraflop on pastebin or reddit leak.

But don't forget the early PS5  rumored are based on 2017 to 2018  early leak, so probably the Sony engineer  were using overclock Vega 64 or crossfire for the prototype and  targeting at GCN teraflop number measurement level . So if the comparison between Vega and Navi are true, it means the targeted rumor (it said PS5 will have 12 to 14 teraflop) are actually hit the mark. PS5 will end up at 14 to 15 teraflop if it's using GCN teraflop measurement, but we will see as less teraflop number on Navi between 9 to 10 teraflop RDNA architecture. 

My claim actually in line with the latest leak/rumor,  that the PS5 prototype APU chip supposedly called Gonzalo (according to Tum Apisak and Digital foundry), are performed closely to RTX 2080 on 3D Mark Firestrike benchmark https://www.digitaltrends.com/gaming/playstation-5-benchmark-leak/?amp

https://youtu.be/p2bm5kXGiVg

The score is hidden.
I only see the overall score.
PS4 - Rough score 5000
Gonzalo - score 20000 up

— APISAK (@TUM_APISAK) June 25, 2019

The score number is the total number of combination of CPU and GPU score, it doesn't say about the GPU score, but some people at Reset Era said that even with even with  Intel i9900K and RTX 2070 the total number will likely to hit at 19000 with these benchmark, while PS5 APU is 20000. So if these score are true and PS5 using Ryzen 3700 8 core (Ryzen 3600 score number on these benchmark are better than Intel i9900K ) than PS5 will sit around  GTX 1080/ RTX2070 or slightly more and even closer to RTX 2080. 

So, I am happy if the number is 8, 9 to 10 teraflop,  even though we will get less teraflop but in reality we will get more in real performance than the early rumored suggest.

So tell me what do think about this performance for PS5 and Scarlet, are these performance within on your expectation, or you still insist that even on Navi PS5/Scarlet will still be able to achieve 14 teraflop?  Are you satisfied, or tell me your prediction performance number.

Theres more to real world performance than just the number of Tflops.
These cards have higher IPC (instructions pr clock) than their older gen cards tech.

5700 XT should be abit faster than a 2070 right? (not in all games, but overall)
Im not sure consoles will get that level of performance, but if we do.... holy hell...

Be honest guys, even those PC master race people, how many people here have a Geforce 1080 or better?
plus with consoles and close to metal codeing, you get more bang for the buck,.... PS5  & XB2 are gonna be monsters.

PS5/Scarlet will probably have similar power to an RX 5700 and a Ryzen 7 3700X. Probably slightly worse especially on cpu side. But yes as you point out the RTX2070 has better performance in gaming than the Vega64 while having almost half the TFlops, so while "8 tflops" might seem pretty underwhelming, it's misleading, actual performance will be anything but. But due to this you definitely might not see Teraflops used in marketing as it undersells the change.

RTX 2070 - 7.465 TFLOPS
Vega 64 - 12.58 TFLOPS

Yet RTX 2070 performs better in games.

New consoles will perform similarly to a GTX 1080.

Actually the CPU side is what sell these console,  the CPU benchmark is crazy good , it's even beat i9900K according to some forum and gaming site. Also the benchmark is actually slightly  better than GTX 1080, because GTX 1080 sit around 18000 to 19000 score . And the final performance will double or more than of GTX 1080 due to low level API, optimization, and no overhead because all the spec are gaming device focus. 

Yes, that's true PS5/Xbox Scarlet will be a beast , but the real evolution is on the CPU side , the GPU will be an OK jump but not that bad , because we already had mid gen refresh it doesn't actually look super impressive, but nonetheless it's impressive. Combine that with low level API, optimization, combination of all spec for single purpose only (gaming) these spec is crazy good , i bet the price will increase this time around 499 USD but of course it's inline with the inflation of price and still considered affordable if you compared to the price of single RTX 2070 GPU.   

Well I have 3 thoughts on performance

1) I am fine with the power of the Switch. Perhaps it is because I grew up playing ps1/N64, but any game that is HD with a stable frame rate is good enough for me. The ps4 Pro looks beyond amazing.  Pretty much all games are leaps and bounds better than the stuff I played growing up. 

2) Graphics mean nothing without art style. The "prettiest" game I have ever played is Wind Waker HD. I'll take style over graphics all day long.

3) Games matter. I don't care how good something looks, what matters is how good the game is.  And the quality of a game is 99% gameplay, controls and level design.  Dark Souls 3 isn't the prettiest game on the ps4, but it is the best.

Yea, Tflops will look unimpressive on paper (while actual processing power will be very good if Gonzalo really is one of the console APUs), so I think they might instead market the consoles with transistor count, as that is pretty sure to increase ~4x (and is a bit better processing power indicator anyways, but not a flawless one either).

Yup that's true, if we compare it on paper  RX Polaris with 36 CU is around 5 million something,  while RX Navi with 36 CU are more than 10 millions. The problem is the casual people still does not bother and simply don't have clue, but then again PS5 will focus on hardcore gamer thus "the hardcore " gamer probably can easily understand and research it by them self,  because hardcore gamer tend to stick on gaming forum and gaming news often. 

Well the good thing about power it also gives developers a lot more flexibility to offer features, assuming they leverage it properly.

Im with you in regard to visual fidelity not being the most important aspect. But more power can also mean 60 fps can become more common, potentially more splitscreen games with stable performance, bigger and more dynamic worlds, better load times, etc.

Hence, I like the push for power but Im often critical of how its used. Halo:MCC is arguably my favorite release this gen simply because it made numerous old games I love better. The power was simply leveraged for higher resolutions, 60 fps, better performance for split screen, etc.

Then you have something like Halo 5 which is a good product, but released with no split screen, jarring visual quirks and a disappointing resolution. In comparison Halo 4 in MCC looks like the more polished game.

So in a nutshell, more power is great but it often feels wasted.

Most likely scenario is 2x XBO X, hopefully with additional hardware for RT.
Still not very good considering much bigger 1080p to 4K UHD jump than 720p to 1080p this gen, so hopefully devs will not chase native 4K too much, at least later in gen.