Pretty apt seeing what happened with this interview and all xD
People should take singular developer opinions with grains of salt anyway. It's not as though just being a "developer" means you can't release bad or unoptimized work. Hell, i've released games on google play, that technically makes ME a developer.
I think the weight of this comes from the fact that he is with Crytek, and they have historically been making hardware-pushing engines, and he's a rendering engineer so he's right in the thick of the hardware. So the opinion of a developer from a company like this will inherently hold more weight than usual, even someone responsible for AAA games.
It could be a PR stunt, or it could be true. Impossible to tell, but so far he seems to mostly check out and he's dumped alot of info.
I mean there's only so far he can stretch the truth, yeah? He's basically saying the PS5 is the superior console, which is going to be VERY easily disproven in a few months time, especially as nobody has been operating on the idea that XSX's 12 TFLOPS could possibly not be consistent to such a degree as this. Or even that XSX could actually be weaker than the PS5 in any real capacity.
People are going crazy at him suggesting the PS5 is the better console, but TBH his accounts don't seem to be all that farfetched. Outside of just plain deniers, we've been having people speak on PS5's design merits for a long time.
The biggest thing this article did was echo some of the POSSIBLE design defects of the XSX, which have mostly dodged conversation due to them just having much better on-paper numbers.
Well, ironically the interview itself is kind of a moot point now since it's been redacted, but a few things I want to touch on real quick.
1: He wasn't saying the PS5 was the superior console, he was saying it was his preferred system to work on due to ease of development. For some ease of development does lend itself to being superior, but for others it may not. We've seen enough examples of both throughout console history to know this.
2: Not every game on XSX is going to need 12.147 TF of performance, same as how not every PS5 game will need 10.275 TF of performance. However the main difference between the two systems is that when a game needs 100% performance from XSX the system will always deliver that and the cooling will kick up into gear to compensate for the extra heat being drawn. OTOH, if a PS5 game needs 100% performance from the system it will need to be careful in how long it consecutively taps at 100% or the power budget will be stressed and the system will clock down the processors (going by Cerny's theoretical amount, by 2%, though that is open to speculation and interpretation).
3: The majority of the people that've been speaking of PS5's design merits since Road to PS5 are Sony 1st-party, like the ones Jason Schreir talked to. Which would make since because that is their chief platform of operations. Same would be the case with XSX if you had a bunch of MS devs coming out praising its design merits. And you generally won't have 3rd-parties saying the things the guy in this interview did because of NDAs which basically prevent these kind of snafus from happening in the first place (and possibly earning repercussions from a platform holder along the way).
4: Dunno about other people, but I've touched on some of XSX's design quirks for a while now, and I've been reading around other places seeing people who know their tech touch on those as well. Even there, though, there are aspects where some of them amy not be looking at things from the right perspective, or are exaggerating certain possible design quirks. That isn't to say they don't exist, though.
But like I said, it'd appear all the discussion in this thread was generated by an interview that's now been redacted, and a dude who happens to be a
really big fan of PS (nothing wrong with that). But they mixed business with pleasure, and not in a way fit for Pornhub, so that was a mistake.
Hypervisors is not performance free.
This doc has a lot of tests with memory, I/O, processing, etc.
It has comparison with Virtual Machines (hypervisior), Container-based virtualization and Physical Machines too for better understand of the overhead add with each option.
In terms of performance Container-based virtualization is a bit better than hypervisor (Virtual Machine) and can reach average of 4% overhead with some operations reaching 50% overhead.
PDF | The current virtualization solution in the Cloud widely relies on hypervisor-based technologies. Along with the recent popularity of Docker, the... | Find, read and cite all the research you need on ResearchGate
www.researchgate.net
That's true, they aren't performance-free. Nothing is; everything that requires performance has a cost attached to it.
But also worth keeping in mind is that paper is speaking of generic examples; it's harder to know what proprietary implementations will bring, what changes they make to aspects of the concepts etc. Usually those can perform better than theoretical generic examples.
Also FWIW the implementation of hypervisors in XBO, like so many things with that system, were hampered by the multimedia focus of the console design, rather than being focused as a gaming-first device. So there was bloat and imperfections (in regards to a gaming-first implementation) to its use of hypervisors that could've been greatly improved if the vision was tighter.
