Next Generation Hardware Speculation with a Technical Spin [pre E3 2019]

Status
Not open for further replies.
The thing that bothers me about discless Xbox is bc, my entire 360 library is physical.
They have a console for you, the one with the disk drive.
It's a shame nothing came from the physical trade in scheme. Maybe next gen but I reckon it's been scrapped.
The GPU isn't the part of 1X that's holding it back. This has been my point all along. Sure, Ryzen is a big step up from Jaguar, and that extra CPU power can speed up your simulation, AI and other stuff, but it's hard to market framerate. If Microsoft tries to market a less than 4k, less than 6TF GPU that produces 1X quality visuals at higher frame rates as a next generation machine, they are going to have their sit in the corner holding one swollen eye while they see Sony spending their lunch money on twinkies with the other.
It will only be 1X visuals when in BC mode for current and previous gens, next gen it won't be. Maybe I wasn't very clear when I wrote that before.

1X was using its 6TF gpu to hit higher resolutions than Lockhart would be for next gen, not including all the other gpu features it will include.

Think I need to understand what your expecting from next gen, say ps5/Anaconda?
Is it just/mainly native 4k and maybe higher framerates?
 
Nintendo who until very recently were taking down YouTube videos of their first party games, can't be bothered to shut down emulation projects and sites linking to imagines of their own games
I was sure I heard recently they was aggressively going after websites.
 
I was sure I heard recently they was aggressively going after websites.

Then they are doing a really poor job, I've been using the same emulator resource sites for 15+ years.
 
"Additional CPU." That one really could be Cell. Should be a few mm² at 7 nm, wouldn't need to be programmed by anyone other than Sony, and could do PS3 BC.

My thinking here is how to get a Cell in there for PS3 BC. Though it is tiny, it still needs to be justified. Just as PS1's hardware did useful stuff for PS2, PS3's CPU could be beneficial for PS5 as a file processor. Considering the ARM in PS4 is too crap to enable background downloading, a tiny Cell much handle all that gubbins much better. And the value to the system, "plays all your old games," would be substantial in marketing terms.

IRC, I remember reading certain documents and seeing certain videos of RT being handled quite well in non-gaming scenarios. If the Cell is repurposed with updated logic, maybe it will help the GPU with such task, and audio as well.

Edit: see Cell RT being discussed already.
 
Last edited:
They have a console for you, the one with the disk drive.
It's a shame nothing came from the physical trade in scheme. Maybe next gen but I reckon it's been scrapped.

It will only be 1X visuals when in BC mode for current and previous gens, next gen it won't be. Maybe I wasn't very clear when I wrote that before.

1X was using its 6TF gpu to hit higher resolutions than Lockhart would be for next gen, not including all the other gpu features it will include.

Think I need to understand what your expecting from next gen, say ps5/Anaconda?
Is it just/mainly native 4k and maybe higher framerates?
What I'm saying is that if you have a less than 6TF GPU that's efficiency brings it up to the level of Xbox One X, with the same amount of memory, you can't expect a generational shift in graphics quality. If the argument is that it doesn't have to be 4K, then what is stopping One X from achieving the same graphical fidelity, perhaps at reduced frame rate, at the same resolution? The slower CPU in X would slow down the simulation but the quality of the image would be determined by the GPU and the amount of memory it has for assets. Which would be the same.

What I'm expecting out of next generation is the same we got his generation. More pixels, more realistic simulation, bigger worlds, and less genres.
 
Both are pushing the envelope, that’s why they’re using zen2 and not zen+ & Navi not Vega.
Cant wait to see Navi & zen2 start to roll on the pc.....exciting times ahead and a big indicator for ps5.
 
What I'm saying is that if you have a less than 6TF GPU that's efficiency brings it up to the level of Xbox One X, with the same amount of memory, you can't expect a generational shift in graphics quality. If the argument is that it doesn't have to be 4K, then what is stopping One X from achieving the same graphical fidelity, perhaps at reduced frame rate, at the same resolution? The slower CPU in X would slow down the simulation but the quality of the image would be determined by the GPU and the amount of memory it has for assets. Which would be the same.
Your basing everything of the TF number which even for a GPU isn't the correct thing to do.
You can when running in BC mode as your comparing like for like operation.
When running games coded for next gen you can't.
There's things that we know/expect navi to be able to do which Polaris can't, which I did list some before and they will make a big graphical difference.
The benefit of using TF number is to see where the next gen sits in comparison to each other. All using base navi arch.
I could've left of the TF numbers and said Anaconda is 2.x * Lockhart rendering at over 3* the resolution. Then maybe it wouldn't be such a concern.

You also mention same amount of memory, the difference is that you can stream and replace more and higher quality textures. The actual use of the memory is different.
That's why I also don't think 24GB in Anaconda/PS5 is a good use of the budget.
What I'm expecting out of next generation is the same we got his generation. More pixels, more realistic simulation, bigger worlds, and less genres.
You will get all that from Lockhart but at a lower resolution, you can't get that from 1X even at lower resolution as the rest of the system is not capable of it even at lower framerates.
 
Last edited:
Your basing everything of the TF number which even for a GPU isn't the correct thing to do.
Sort of. Less than 6TF is sort of entry level in the PC space right now. And we are talking about AMD parts here. It isn't like they have some insane track record of efficiency where they magically pull out higher performance per flop every time they release a new generation of graphics card. In fact, I'd say their performance has been much more aligned per flop than nVidia, who have been achieving a little more per flop each generation. So yeah, I'm basing a lot of this on the TF number, but we are talking about a company with a known history. There is a consistency here that says if AMD releases a GPU with 5.5-6 TF, it's going to perform like a R9 290X, RX 480, or RX580. Which, sort of all perform the same at most resolutions, even though they were released between 2013 and 2017, and the 290X is GCN2 while the 580 is GCN4. Gamers Nexus did an article revisting the 290X earlier this year.

You can when running in BC mode as your comparing like for like operation.
When running games coded for next gen you can't.
There's things that we know/expect navi to be able to do which Polaris can't, which I did list some before and they will make a big graphical difference.
We don't know that because we don't know much about Navi, except that it's still based on GCN. What we know about GCN is that it has performed pretty consistently per FLOP. Now, if there's some specail ray tracing hardware, well that's a different story. Actually, thinking about the RTX launch where ray tracing halved or quartered the frame rate and some people claimed to barely see a difference... Maybe the story is different but worse. I guess I'm not sure what special features you know that Navi has that are going to be such a game changer.

The benefit of using TF number is to see where the next gen sits in comparison to each other. All using base navi arch.
I could've left of the TF numbers and said Anaconda is 2.x * Lockhart rendering at over 3* the resolution. Then maybe it wouldn't be such a concern.
Well if you read my posts you'll note that I said a less than 6TF Xbox vs a 10TF PS5 would be terrible.

You also mention same amount of memory, the difference is that you can stream and replace more and higher quality textures. The actual use of the memory is different.
That's why I also don't think 24GB in Anaconda/PS5 is a good use of the budget.

You will get all that from Lockhart but at a lower resolution, you can't get that from 1X even at lower resolution as the rest of the system is not capable of it even at lower framerates.
Why can't 1X stream textures? Yeah, the drive's slower but if we are talking about lower frame rates, you have more time. Also, why 12GBs? Part of the reason why One X has 12GB is to achieve the bandwidth MS needed. The 256 bit bus on Xbox one is really a collection of smaller bus that connect directly to clusters of memory. They increased the memory by 50% to increase the bus width by 50% while making the switch the GDDR5. 12GB is kind of an odd number, and I'd expect next gen to either feature a 256 bit bus with 16GB at higher frequencies, or a 512bit bus and 16GB with a smaller frequency bump.

If we are assuming that the graphics chip is going to be 15% more efficient than X so lets just drop the FLOPS to match, why not do the same with the memory? Is it's going to be more efficient, why not just drop back down to 8 and let the special sauce bring you back up to par?
 
https://www.myce.com/news/chinese-manufacturer-announces-open-channel-ssds-84846/


And open channel SSD no problem with basic operations, they are done by the controller

In fact Sony may invest much more on fast SSD solution than MS does. Because MS has almost no exclusive first-party games. All of them are cross-platform to PC. In other words, when developing first-party games MS has to consider the lowest common denominator of PC storage, that is, 500MB/s SSD or even 200MB/s mechanical HDD.They can't design a game that needs 2~3GB/s streaming rate.

In fact MS only needs the SSD solution which can decrease some load time, not the ultra-fast SSD .
 
In fact Sony may invest much more on fast SSD solution than MS does. Because MS has almost no exclusive first-party games. All of them are cross-platform to PC. In other words, when developing first-party games MS has to consider the lowest common denominator of PC storage, that is, 500MB/s SSD or even 200MB/s mechanical HDD.They can't design a game that needs 2~3GB/s streaming rate.

In fact MS only needs the SSD solution which can decrease some load time, not the ultra-fast SSD .
this is least logical because that means all third party developers can't take advantage of it either meaning only Sony's exclusives would have to be developed with this and be the only games to take advantage of it.

Doesn't seem like a big win for such a niche group of the entire library of games.
 
Next-Gen consoles will be around to slightly under GTX 2070 performance levels.
 
Next-Gen consoles will be around to slightly under GTX 2070 performance levels.

Did they announce the configuration (Core count / frequency) of that 5000 series card being used in that demo? Because the RTX 2070 is about 7.5TF in performance.
 
Did they announce the configuration (Core count / frequency) of that 5000 series card being used in that demo? Because the RTX 2070 is about 7.5TF in performance.

Nope. Only said that they'd have more details on Navi at E3 on June 10th (the day after Microsoft's Press Conference).
 
this is least logical because that means all third party developers can't take advantage of it either meaning only Sony's exclusives would have to be developed with this and be the only games to take advantage of it.

Doesn't seem like a big win for such a niche group of the entire library of games.

However, who will design a game that needs several GB/s of streaming rate since most PC games don't have such kind of storage?

The answer is that only Sony' exclusive will design this kind of games. Maybe this is just what Sony needs because PS5 exclusives can have more advanced game design and make much difference than cross-platform games, so that Sony' exclusives especially Sony first-party games can become greater system sellers.

Other multi-platform games still have benefits such as ultra fast loading and less pop-in.

Navi is an all new architecture RDNA, not GCN.

7nm transistors are 2.56x power efficient than 16nm. If Navi really has 1.5x power efficiency than previous GCN. Can we expect next-gen consoles with 1.5 x 2.56 = 3.84 x performance of xbox one x?
 
7nm transistors are 2.56x power efficient than 16nm. If Navi really has 1.5x power efficiency than previous GCN. Can we expect next-gen consoles with 1.5 x 2.56 = 3.84 x performance of xbox one x?

So you're really expect to be around 23 TFlops?

That's not how things work.
 
7nm transistors are 2.56x power efficient than 16nm. If Navi really has 1.5x power efficiency than previous GCN. Can we expect next-gen consoles with 1.5 x 2.56 = 3.84 x performance of xbox one x?

Marketing wise, hopefully Sony and MS next-generations systems can hit double digit TF performance. Regardless of how efficient the RDNA architecture is or it being closer to Nvidia's RTX architecture performance... seeing a 2020 console with ~8TF performance compared to a 3-4yr old 6TF Xbox-X would look awful from a pure technical marketing approach. I'm still rooting for 11-12TF team.... :yep2:
 
Navi is an all new architecture RDNA, not GCN.

Going by the LLVM changes for GFX10, if someone were to stumble unknowingly into a Navi ISA doc, I think it would be very recognizable as a GCN or GCN derivative.
In the Navi thread, there is some debate as to what "GCN" can be defined as before deciding what can be defined as being "not-GCN".
There are elements to the architecture disclosed so far that would actually be more familiar to Sea Islands or the current gen consoles than someone looking at Fiji and beyond.

There are some signs of significant changes, such as the mostly undescribed cache hierarchy that might pass some threshold of being new-enough to be considered an distinct evolutionary successor. Whether that's a successor of a single unbroken line of GCN generations or an amalgamation of features from different console and client/compute products is not easy to determine from the outside.

Since the GCN architecture as a marketing concept has been rather haphazard on what counts as a feature of an overall architectural model versus implementation detail, something arbitrarily put together by AMD could be more arbitrarily declared to be different as well.
 
RDNA or whatever they want to call it is definitely GCN based as seen in the LLVM changes. AMD is simply forking GCN into two branches one for "gaming" (RDNA) & one for Compute (currently Vega & then its successor).
 
Status
Not open for further replies.
Back
Top