Next Gen Tech (NGT): STORAGE

[Pemalite]

Bofferbrauer2 said:

Pemalite said:

I would hope for an octo-core CPU in 2020 at a minimum.
Ryzen will be old and outdated by then, you would be looking at Ryzen 3/4 by then on the PC.

Flops doesn't tell us how powerful a GPU is.

Ryzen 2 is slated for 2019 (Pinnacle ridge is just a Speedbump for 2018, not Ryzen 2). Ryzen 3 may be a possibility, but there's no chance there will be a Ryzen 4 by that point already.

Pinnacle Ridge is likely to be more than just a simple speed bump... It might involve a respin as it may be targeting 12nm, which will bring with it errata fixes, clock rate increases and lower power consumption.

That is "Zen+.". Aka. Ryzen 2.

Ryzen 3 will drop the year after in 2019.
Ryzen 4 in 2020.
Note: That the chips name doesn't necessarily have to correspond with a new core architecture.

If there is a delay or a blow out, my statement still stands as we will be "looking at Ryzen 3/4" and not necessarily "have" Ryzen 3/4.
AMD will be iterating on a yearly cadence going forwards.

Intrinsic said:

And they are what will determine what kinda storage solution we will use.

Except it doesn't.

Because Sata can be used for SSD's and Mechanical Hard Drives, regardless if it is Sata 1 or Sata 3.2.

If it has M.2 they might just use it as a cache drive.

Intrinsic said:

I am also aware of SATA express or SATA3.2, but you and I both know thats not going to end up being in the new consoles. There are very few consumer devices that even uses it today and they have been around for a good while now.

I don't have issue of what uses what today. I just have issue with your original claim, that Sata is "up-to 500MB/s" when that is blatantly false.
Here is a heap of Ryzen boards supporting Sata Express.
https://www.gigabyte.com/mb/am4/Model

I think going forward tech will be ditching propriety standards and instead have everything dangling off PCI-E lanes in the end.

Intrinsic said:

We all know that SATA 3 has a max theoretical bandwith of 600MB/s. I used 500MB/s cause in reality no one is going to be hitting that theoretical max on average. Even that 500MB/s isn't achieveable in real everyday use unless you are running some sort of test thing.

One thing for certain is that we can't stick to the what we are currently using now. Especially if next gen we have up to 20-32GB of Ram in consoles.

SSDs over sata 3 may seem reasonable, but thats still talking about a peak of about 550MB/s. And thats assuming that peak is hit often. But if they are goiing with SSDs, then they very well could use an M.2 interface. It gives room fro growth. The consoles won't have to ship with the NvME variant of those drives, but with the sata variant, which are significantly cheaper. Then the users could always upgrade their drives if they want better performance to the NvME variant.

The SATA speed is almost irrellevent if they still use mechanical disks though, which is the likely outcome due to pricing.

SvennoJ said:
What's holding current consoles back from making more use out of SSDs?

I wonder why the gains are generally pretty small compared to HDD. The 5400 rpm drive in the consoles probably don't do more than 80 MB/s. an SSD should be at least 3 times faster even on Sata 2. Load times are faster, yet not that much faster.

Not all games load things the same way.
Some games will opt for an extremely large chunky load. Some would prefer to do a smaller load and stream in assets over time.
Some games will take a mixed approach.

Some games use an abundance of compression on it's assets which actually loads into RAM quickly, but takes longer to decompress thanks to the CPU.

Which is why there is such large discrepencies in performance in regards to SSD vs mechanical load times on consoles.

All up to the developer essentially.