Switch's GPU also has a variable clock rate. As for the PS4 and Xbox One, yes, they were both designed around this same ancient design philosophy even though it was pretty much irradicated from the then current PC architectures for both CPU and GPU.
Indeed. And I suspect if I go through my post history, I'd probably find a post I made along a similar vein. I think on the topic whether compute as a marketing metric versus compute as an actual metric are probably 2 different topics. I'm going to go with DF leaning towards discussing the marketing aspect, as it's clear compute matters. As pointed out, XSX is at a deficit nearly everywhere else, so the fact that it continues to put up equal or better resolution at times, is probably indicative of the compute difference putting in the work there to blanket it out. Compute most certainly matters, and if it didn't matter, people wouldn't be discussing the importance of a mid-gen refresh model having 2x more compute. But as a marketing topic, its clear that compute alone isn't the only metric that matters, we just assume the rest of the pipeline is sufficient in size to support that size of compute, which has been a decent barometer for some time now.
Yes. I am sorry for confusion I should be more clear.
It was a big part of making checkerboard rendering viable from what I understand. Which definitely got a fair bit of usage in games. I think I remember Far Cry 5 used it for something as well. But really, that's also one of those things that we likely just aren't gonna hear developers talking about too much, either. That's kind of getting into the nitty gritty of performance optimization and gamers just generally dont care about that kind of thing.
I think 360's hardware tessellator might perhaps a better example.
One of Sony mistakes not to buy Ready at Dawn. Order 1886 was clearly rush with repeating boss fights and strange abrupt ending. Sequel on ps5 could be awsome.
I'd say DX12 has overall been a very solid positive, but it was a slow start and despite all the advantages it offers in terms of new features and performance potential, it comes with some rather unsurprising downsides in terms of extra potential complications.
From the graphically impressive games I count Ryse, Killzone Shadowfall, The Order 1886, AC: Unity, Alien: Isolation and Battlefront 2015 among those who belong to the early phase of the last console generation.
AC: Unity has many strengths but unlike Ryse and The Order 1886 it has some clearly visible weaknesses like objects suddenly appearing. The Order 1886 is perfection. However, they could have placed the cover objects a bit more cleverly. In The Division they fit more natural into the landscape.
Although Ryse and The Order 1886 are ancient they could be released today and they wouldn't stand out negatively graphically. For comparison just look at what the games looked like 10 years before Ryse. Ryse and The Order 1886 would even look better than many current games. The graphical leap from a flickering, artifact rich and often implausible-looking top tier PlayStation 3 game like The Last of Us to a polished and filmic looking game like The Order 1886 was gigantic for the few months difference.
The second phase started for me in 2016 with games like The Division, Quantum Break and Uncharted 4.
The last phase is RDR2, The Last of Us 2 and Detroit Become Human.
You often can't make out any big technical leaps among the top titles. The hardware was quickly mastered. In return, the projects became more and more ambitious over the years.
However, there are also exceptions. Naughy Dog was able to improve the graphics of The Last of Us 2 a lot compared to Uncharted 4.
EDIT
From my point of view DigitalFoundry could have also named Quantic Dream around the 6:40 minute mark as one of the early studios which worked with physially based materials. For a PlayStation 3 game Beyond Two Souls looked phantasitic in places with smaller areas.
