And there is nothing in the Console-world that can leverage RAID SSD's and offer 64-128GB/s. Absolutely NOTHING. NIENTE. NADA.

And you don't seem to like or understand that either, because it means your doesn't have the best of anything?

No. No and No.

The Playstation 5 SSD offers 5.5GB/s of Real-World sequential read performance, not sequential writes, not random reads/writes, those are the fundamental hardware facts.

That is what the SSD is actually physically capable of in the real world.
https://www.eurogamer.net/articles/digitalfoundry-2020-playstation-5-specs-and-tech-that-deliver-sonys-next-gen-vision

From there... Sony leverages compression... Kraken is actually a software compression algorithm which will increase the compression by 10% over the other compression algorithm in the system, sounds like fuck all... But 10% when your throughput is 7GB/s (5.5GB/s+1.5GB/s of compression) is 7.7GB/s.

There is also a decompressor unit in the main SYSTEM SoC and not a separate CHIP, which is also named Kraken, which decompresses that data on the fly which is basically offloading work from the CPU, the PC can do this to... Because the PC isn't limited by only 8x CPU cores @ 3.5ghz.

You will need to provide a citation and/or appropriate accompanying benchmark to substantiate that 20GB/s claim.

If you had bothered to listen to Microsoft, their approach is similar to Sony's, they offloaded compression/decompression to a separate unit on the SoC.

But even when you leverage all this compression, the PC can implement an SSD setup that still beats both approaches with just raw throughput alone... Not to mention additional compression on top if it if you wanted.

Sure. Keep in mind that the 2060 Super can match the RTX 2070 though and in some cases even beat it by a tiny margin.
https://www.anandtech.com/bench/product/2517?vs=2513

And Sony has provided a "best case scenario" with it's clocks, there is the potential that a developer will prioritize CPU tasks at the expense of the GPU, you can't peg both chunks of the silicon and maintain the full 3.5Ghz+2.23Ghz.

There is the potential we could see the GPU limited to something like 2ghz or about 9.2 Teraflops for what it's worth.

I would disagree with that of course.

The thing with compression is that not all datasets are compressible to the same ratio.

Consoles and PC games have been leveraging various forms of compression for a third of a century now, it's not a new or novel thing, we are actually rather good at it by this point.

The priority levels are there to prioritize certain pieces of data for decompression and memory transfers, some forms of data are more important than others, it's a fantastic feature, but not really relevant when the PC has more Ram and can just dump everything into it.

RDNA 2 should bring with it a plethora of efficiency gains as well, the extent of which is a big unknown, but AMD is pegging a performance/efficiency increase of 50%.
Salt and all that until we see benchmarks though.

The Xbox One X has the GPU edge no doubt, it has the bandwidth, it has the functional units and will take the fight to the RTX 2080.
The Playstation 5 is a step down from the Xbox and falls into the Geforce RTX 2060 Super/RTX 2070 territory which is a step down from the RTX 2080.

This is just the raw performance from the details we have currently, obviously we don't know all the details on the next gen hardware to actually make a 100% accurate comparison.

You can do it on the PC, it requires CPU or GPU cycles to pull off, where-as the console use specific co-processors to offload the task.
The PC also has compression/decompression standards as well.

For example: https://en.wikipedia.org/wiki/DriveSpace#Overview

There are going to be instances where the Playstation 5 cannot maintain it's full clockrate due to TDP limitations, anyone who has used and extensively tested how the power relationship between the CPU and GPU portions of the APU will understand what I mean.

https://www.gsmarena.com/samsung_galaxy_s20_ultra_5g-10040.php

The Samsung Galaxy S20 Ultra has 16GB of Ram. My "old" Galaxy Note 10+ has 12GB.

And there are lots of reasons... Caching of the phones "SSD", Multi-tasking, Reducing memory transfers can save on power... And more.

Mid-Range phones these days are around 8GB and starting to trend towards 12GB at the moment.

See above.

We all know what comes after that the XSXsupersexy

If Lockhart is still a thing, I find that to be very unlikely.

1. The Xbox One X has the GPU edge no doubt, it has the bandwidth, it has the functional units and will take the fight to the RTX 2080.
   The Playstation 5 is a step down from the Xbox and falls into the Geforce RTX 2060 Super/RTX 2070 territory which is a step down from the RTX 2080.
   This is just the raw performance from the details we have currently, obviously we don't know all the details on the next gen hardware to actually make a 100% accurate comparison.
2. You can do it on the PC, it requires CPU or GPU cycles to pull off, where-as the console use specific co-processors to offload the task.The PC also has compression/decompression standards as well. For example: https://en.wikipedia.org/wiki/DriveSpace#Overview
3. There are going to be instances where the Playstation 5 cannot maintain it's full clockrate due to TDP limitations, anyone who has used and extensively tested how the power relationship between the CPU and GPU portions of the APU will understand what I mean.

Why is it silly? Series X will be fully backwards compatible with all the previous Xbox consoles. I just expect the same from Playstation 5. And it's a feature I'm really looking forward because I'm tired of having to set up 15, 20 or 25 year old systems every time I want to replay a classic (and I tend to that a lot: I love replaying old games). Needing only one console connected in the room in order to play any game I want is something that really appeals to me. Especially in the case of Sony, because unlike Xbox, there are tons of PS1 and PS2 games I love with all my heart and I replay every now and then.

So, that feature alone would make me buy a PS5 instantly. If it doesn't have that, I see no reason to get a PS5. Simply because since the PS3 days, pretty much all the series I care about that were on Playstation became multiplatform. And Sony changed drammatically the kind of games they make themselves. Nowadays they are mostly western, adult, mature-story driven titles that I personally just don't enjoy at all. Sure, there is the occasional Gravity Rush, Astro Bot and Ratchet and Clank... but that is not enough for me. And since I feel mostly the same way about Microsoft... backwards compatibility is a decisive factor for me.

1. If the XBX is comparable to the 2080... there is no way that the PS5 is comparable to the 2060... there's like a 15% performance difference between the 2080 super and the 2070 super. So that main one with the difference between the XSX and the PS5. You are right though, we do not know all the details and it would be silly making these kinda assessments now.
   
   
2. Yes, I know you can obviously do compression stuff on a PC too. Using the GPU/CPU like the current-gen consoles are doing right now. But the next-gen consoles and particularly the PS5 has specific silicon exclusively for that task. And not just talking about compression/decompression here, I am saying there are other components built into the PS5s APU specifically to make this whole instant access/data throughput/management unique to the PS5.
   
   
3. I am also aware of that, but that applied to every single processor out there doesn't it?

 With Lockhart it's one more SKU than Sony so by following last gens upgrade path,  a PS5 Pro matched by a XSXsupersexy still means just the one extra SKU , that way Xbox maintains it SKU sandwich and the focus shown by MS on being the power leader means not matching the PS5 Pro would be seen as a backdown.

The XSX is their premium model which is being followed by their base model in Lockhart. This situation is reversed for the PS5. If the PS5 Pro aims to double its TF count, similar to the PS4 Pro, MS at the bare minimum would need to surpass that, probably double the XSX, if they wanted another premium option. That would make the gap between Lockhart and the XSX Super astronomical. Is scalability really feasible with a low end of 4/6TF to a high end of 24TF? Its much easier with 4/6TF & 12 TF. MS can't just drop Lockhart despite all the bottlenecks it would present to the highest end model. That doesn't even consider what MS might cut back in order to hit a favorable, mass market price.

Ultimately, MS priorities and goals revolve around subscription models, & they would prioritize that over being the power leader 4 - 5 years down the line. Lockhart at $299, maybe even $199, seems like a more favorable strategy for them, instead of releasing another high end Xbox model. This option goes down the drain if Xbox Series X is their base of course, but according to MS insiders, Lockhart is still in the works.

If you rewatch the presentation you'll see Mark Cerny saying something about it hitting 20Gb/s depending on the situation.

And yes I know not all compression is the same. Thus why the only firm data is the 5.5Gb/s both 8-9 and 20 are depending on scenarios, 8-9 the most likely during most of time and 20 a theoretical limit.

About the RAID and other PC superiorities, which even Digital Foundry is putting the solution in PS5 was better than what was in the PC at the time, does most current gen AAA games on PC load in 1s with keeping the status of like 4 games saved at once for quick resume?

I think the other comment was saying that today, an equivalent to the SSD in PS5 doesn't exist. And the link you gave was from end of Dec 2019, so I got the impression that you were also talking about nowadays.
But in the case that they don't exist today, that should be true at the time Cerny made his Wired statement as well, so I'll focus on today.

I'm not knowledgeable on the subject at all. But I was listening to Digital Foundry's analysis, and they did say that "SSD's capable of these sustained transfer rates don't really exist in the consumer space today."
And then one of them said that today, "the drives are catching up to the performance level of Sony's internal solution."

https://youtu.be/4higSVRZlkA?t=1343

He mirrors what Cerny said about those drives needing to be even faster than Sony's 5.5 GB, because of the 6 levels of priority on the PS5's SSD.
Because PS5's I/O would need to step in and enforce those extra levels of priority difference on drives that don't support it.

I don't know how much faster they need to be though. Maybe RAID would function in a similar manner on PC, but I don't know how that accounts for the difference in technology.
By the way, the Kioxia SSD you linked to has 4.2 GB write speed. If the SSD is supposed to function as RAM then is write speed important as well?

I suppose Cerny does need to prove how that technology compares to industry standard SSD speed. Though Digital Foundry weren't skeptical, so I guess it seems plausible to them. I wouldn't know.
And yeah, the physical size of the drive I didn't mention because of anything you said, but just to emphasize that it may be tricky to find a compatible drive in more ways than one.

1. If the XBX is comparable to the 2080... there is no way that the PS5 is comparable to the 2060... there's like a 15% performance difference between the 2080 super and the 2070 super. So that main one with the difference between the XSX and the PS5. You are right though, we do not know all the details and it would be silly making these kinda assessments now.

1. Yes, I know you can obviously do compression stuff on a PC too. Using the GPU/CPU like the current-gen consoles are doing right now. But the next-gen consoles and particularly the PS5 has specific silicon exclusively for that task. And not just talking about compression/decompression here, I am saying there are other components built into the PS5s APU specifically to make this whole instant access/data throughput/management unique to the PS5.

2.) Cerny said that both would usually run at the max speeds.