Next Generation Hardware Speculation with a Technical Spin [2018]

[Syferz]

I wish, but I think the next version of the switch will be on the cheapest 16nm variation. It will allow for better battery life or smaller batteries, maybe more consistent frame rates if we’re lucky.

I think Holiday this year or maybe Spring 2019, we could see a Tegra ~X2 based Switch upgrade, it will perform better, but it won't really be the focus. X2 is just an X1 at 16nm really, yes it has some updated architecture (pascal) but X1 was the prototype to pascal. I could see a ~30% increase in performance over the current Switch, with a similar increase in battery life (~4 hours minimum in games like Zelda) This would really only help dynamic resolutions and more solid frame rates, the real focus would be shrinking the chip to produce more at a time.

This would give the handheld ~256GFLOPs and when docked 512GFLOPs. it's a noticeable improvement in performance, slightly better than launch XB1 to XB1S. Maybe increase the CPU speed to 1.5GHz, as n3DS did see a huge CPU performance increase, while keeping compatibility with previous 3DS hardware (266mhz dual core ARM 11 processor, to ~800mhz quad core ARM 11 processor, this would be a very tame but important improvement for catching current gen games, as 1.5GHz A57 Quad core, would perform better per thread than PS4, but still lack overall performance, it would still be close to the basic 6 core 1.6ghz jaguar cores if Nintendo allows all 4 cores to run at 1.5GHz.)

It would all be a precursor to a Switch 2 launching in 2021 holiday / 2022 spring. Based on much stronger hardware I listed above.

 

[Deleted member 13524]

Why?

Because during the last generation the console makers learned they should regulate their console releases according to what's available in the PC market (i.e. the state of the art). Otherwise they'll see the same large migration to PC as we saw in 2011-2013.
The longer they take to release a new and significantly more powerful console, the higher is the chance that a gamer is able to purchase a $400 PC that is more powerful than the current consoles, and gamers know they will save a buttload of money in software if they transition to PC. Software sales are the main source of income for console makers, and those were pretty bad for Sony and Microsoft during the last 2/3 years of the PS360.

Releasing the Pro/X mid-gens allowed them to buy time to extend from the traditional ~5 year hiatus between generations without lots of people migrating to the PC. Had they not released the mid-gens, the 65W Ryzen 5 2400G coupled with 2*4GB DDR4-3200 would already provide similar performance to the 2013 consoles in multiplatform titles using DX12 or Vulkan, at 1080p. And a PC built around that would cost around $400.

That said, the mid-gens aren't going to buy them another 5 years. In 2020 (abundant 7mn EUV and early 5nm?) we'll probably see ~$200 PC SoCs that perform as well as the current mid-gens. In 2020 we're looking at Raja-driven Intel iGPUs and AMD with post-Navi (post-GCN) iGPUs, which should get significantly more powerful to avoid the shortage of mid-end PC GPUs that we have now, due to crypto.

And by 2021 the console makers would be facing more than a year of mass migration to the PC that would downsize their software sales.

Cryptocurrency is seemingly more and more a useless blight. The future won't be gaming or entertainment or using your computer to make something, but sitting around waiting for it to find a key and make some money. Woohoo.

There are upsides in creating an universal and ultra-safe exchange system that is able to circumvent the traditional bank fees and dependence on banks and macro-decisions from large investors.
IMO, this issue is bigger than gaming.

However, I also agree that the general cryptomining industry (not hobby / small-scale miners like e.g. Mize) that was created within the last 5 years has become a threat to the environment, the economy and the success of cryptocurrencies in general:

- Large miners are putting their farms near hydroelectric producers to get low-cost power, and then hydroelectric producers don't have to spend money on distribution infrastructures to reach the general population (which was the point of building the government-funded plants) because it's cheaper to sell the energy to miners instead.
- At the same time, large-scale miners are able to circumvent taxes whenever they're able to purchase stuff with crypto, which poses a threat to the economy. These are also people who don't hire other people, since all the work is being done by computer chips through fully-automated scripts.
- Large-scale miners also get to mine (therefore keep if they want) most cryptocoins, which centralizes what should be decentralized.

So while the general idea of an utopian exchange system based on crypto is definitely not an useless blight, the greed-driven industry that has formed around cryptocurrency is threatening the environment, the economy and its own sustainability.
IMHO.

 

[Tkumpathenurpahl]

I believe PS5 will launch in late 2019 and have about 12TFLOPs of performance. This will mean that multiplatform titles for PS4 will dry up holiday 2021, and Switch multiplats with it.
My post is about Switch, I think Holiday 2021 or Spring 2022 is the launch of the next generation of Switch hardware, carrying a new Tegra based on Ampree (Nvidia said they believe their partnership with Nintendo will last 20 years, and Nintendo lags about a year and a half to 2 years behind in tech, with a May 2015 Tegra chip (X1) inside a March 2017 handheld. Ampree should hit production in Spring 2020 if these 2 year life cycles continue for Nvidia's Tegra chips. It will be a 7nm+ part and use LPDDR5 likely around 200GB/s.

• Process node 7nm+
• Display: 7 inch 1080p HDR screen. Switch 2.0 keeps the same form factor as the original Switch and joycons are compatible.
• GPU: 1024 Cuda cores clocked at ~1.3GHz for 2.7TFLOPs. 0 Tensor cores. (Tegra Volta is clocked at 1.3GHz on 12nm, 7nm+ should achieve double density and double performance)
• CPU: ARM64 8 core 16 thread, DynamIQ based on 4 A75 cores and 4 A55 cores. (Snapdragon 845 is using these cores) clocked at 2.4GHz. A55 cores clocked at 1.6GHz.
• RAM: 16GB LPDDR5 /w ~160GB/s bandwidth. (Volta uses LPDDR4 with 137GB/s bandwidth)
• Storage: 256GB flash storage.
• Media Storage: Flash cards like Switch (Same form factor, but a different color) currently 8GB cards are said to be about the same cost as a 66GB bluray disc. I believe in the 4 years we are talking about here, 64GB cards will be about the same as 100GB UHD Bluray discs and not require download/installs. 128GB cards should be available for a bit more and after a couple years, would become similar in price to discs finally.
• Optional Enhanced Dock: $199, 4K dock, with 12TFLOPs GPU inside.
• Price: $299 with standard dock. $479 with 4K dock bundle.
• Launch with the game after next in the Zelda series.

I believe this is the best option available to Nintendo given their nature. I don't expect it to do all these things, but I think this is the best guess I can give right now.

Other than the TF's of the dock, I agree. I would expect the optional dock to contain the Switch 2 SoC and function as a TV-only Switch 2 when isolated, an advanced hub for the TV when in the presence of multiple Switch's/Switch 2's, or a rendering aid when docked with a Switch 2.

I agree with the prospect of a Switch 1.5 coming this year or next. I'd like to see a 720p HDR OLED screen, with a slight bump to clockspeeds, and a much bigger battery. Same sized device, maybe with a slightly larger screen and less bezel.

 

[Gubbi]

The biggest problem with block chain crypto currencies is that they don't scale to real life transaction volumes.

A 1MB Bitcoin block holds 2020 transactions and a valid hash is found every ten minutes for a whopping 3 transactions per second. That's nothing.

You can increase the transaction volume by increasing the number of transactions in a block. But each valid block has to be peer validated, so there is a limit to the bandwidth one can use for this (1GB block broadcast worldwide every 10 minutes ? ).

The other way of increasing transaction volume is to lower the hashing difficulty to increase the frequency of valid block hashes. But this has the downside that you'll have many more miners producing valid hashes simultaneously (here simultaneously means before the peer network has validated your block hash), inducing forks in the chain. Each fork will be worked on until a new hash is found in one of the forks and the shorter branch quashed. With much lower time between valid hashes, individual forks will be extended simultaneously and you now throw exponentially more computational work away per transaction. We're already at insane electricity power consumption levels per transaction, so this is simply not viable.

In the end block chain crypto currencies is an attempt to crowd source monetary authority, and it isn't going well, IMO.

Cheers

 

[Syferz]

Other than the TF's of the dock, I agree. I would expect the optional dock to contain the Switch 2 SoC and function as a TV-only Switch 2 when isolated, an advanced hub for the TV when in the presence of multiple Switch's/Switch 2's, or a rendering aid when docked with a Switch 2.

I agree with the prospect of a Switch 1.5 coming this year or next. I'd like to see a 720p HDR OLED screen, with a slight bump to clockspeeds, and a much bigger battery. Same sized device, maybe with a slightly larger screen and less bezel.

The enhanced Dock targeting 4k, needs to be 4 times the graphical performance of the base unit using 1080p, generally, you want to be a bit beyond that, so 2.7TFLOPs to 12TFLOPs gives you some overhead. Also this allows you to turn off the internal GPU to add more CPU clock speed, to take care of the extra overhead that 4K demands. (thinking from 2.4GHz base to 3GHz max). In 2022, I also expect 12TFLOP GPUs to cost about $200, this mass production dock should bring in big profits, as the rest of the dock is just a housing for the GPU and a USB-C connection.

 

[Tkumpathenurpahl]

Yea where originally our predictions were widely cast, our viewpoints are starting to narrow into a specific direction.

Predictions seem to be either near a XBO-X configuration with Ryzen at $399 in 2019, or it's going to be much later around 2021+ pushing PS4 for 8 years of service before next gen. Anything else is possible sure, but from reading the the responses, most of us are banking on 2019 or 2021. 2019 we will know soon, they would need to announce around may of this year. (correction of next year, whoops, got my years mixed up)

Xbox 2 arriving in the 2020+ is pretty common thinking here as well. It would be weird to see it arrive earlier, they would just be better off removing the S line, and replacing it with the X line.

The narrowing of our expectations is interesting.

Does this all bolster the case for launching base and higher end at the same time?

--Purpose--

The current mid-gen consoles were released to combat migration to PC. In the next generation, releasing their equivalent at launch would be a means of combating both migration and a less and less favourable cost: performance ratio.

--Conservative + cheap base console--

A PS5 at 8.4TF would offer the same kind of improvement as PS4>Pro. It would set a new baseline far in excess of the PS4, would be enough of a leap to not upset PS4Pro owners, and could potentially run any 1080p30 PS4 game at 4K60, or any PS4 PSVR game at 4K120.

But with such a small gulf in performance between it and the X1X, the latter would be its cheaper competitor.

• A 4-core, 8 thread CPU would probably suffice if clocked high enough. As long as it's capable of running PS4Pro code.
• 16GB of GDDR6 would only need 8 chips and reach 576GB/s bandwidth.
• ~48GB of NVME could make up for the small increase in memory, and potentially allow for games to be played without having to be installed to the HDD - just one long loading screen at start up.
• A UHD drive and 750GB HDD should be the bare minimum as game sizes increase and digital distribution becomes more popular.
• 2-4GB of DDR4+secondary processor for the OS, home screen, and storing in suspense/running 1 app.

That's a solid console, which seems quite cheap to manufacture, and would set a decent baseline for the next generation.

--Two tiers--

I'd be happy with the above, but the market is now primed for two tiers of the same console and launching both at the same time would target two demographics:

1. Those who buy at launch and want the cheapest next-gen option available.
2. Those who buy at launch, and want the best next-gen option available.

The trouble is, we've yet to really see how a more expensive and more powerful console will sell. In recent times at least. The PS3 was a year later, more expensive, less powerful, with less available memory, and the harder platform for which to program. Other than being a year late - and memory bandwidth instead of size - the XB1 suffered the same fate.

The best comparison is the X1X. I'm not sure how it's selling in various metrics, but I understand that a greater ratio of XBoxOne purchases are the X than is true of the PS4 and the Pro.

--Pro specs--

I can imagine two configurations for a PS5Pro, and I'm basing them on a simultaneous base and Pro launch:

1) In the style of the PS4Pro.

Not much difference in price - just a couple of hundred dollars.

-- Another butterfly wing GPU bringing performance up to 16.8TF.
-- Either the same 4-core 8 thread CPU, or a 6-core 12 thread CPU. Either way, clocked higher than the base model.
-- 24GB of GDDR6 with 864GB/s bandwidth. Or the same amount of HBM3 if price permits.
-- Larger amount of NVME.
-- 1TB HDD.

2) Dual resources.

Take the butterfly wing design a step further, and apply it to the whole system. Much more expensive than the base model. Something like 700 dollars to the base models 400.

Bundle it with a collection or two of 4K60 remasters. Something that will cover a lot of bases, like Uncharted 1-4 + Lost Legacy + Gran Turismo Sport.


I'd love the latter, as it's a USP I've never seen a console possess: double the power or double the players. It'll probably be the former though.

 

[Tkumpathenurpahl]

The enhanced Dock targeting 4k, needs to be 4 times the graphical performance of the base unit using 1080p, generally, you want to be a bit beyond that, so 2.7TFLOPs to 12TFLOPs gives you some overhead. Also this allows you to turn off the internal GPU to add more CPU clock speed, to take care of the extra overhead that 4K demands. (thinking from 2.4GHz base to 3GHz max). In 2022, I also expect 12TFLOP GPUs to cost about $200, this mass production dock should bring in big profits, as the rest of the dock is just a housing for the GPU and a USB-C connection.

True, but I don't think it would be worth their while to target native 4K with the dock, I think PS4Pro/X1X levels of performance would suffice.

If we're talking a 2.7TF Switch 2, doubling that with the dock puts it pretty close to the X1X anyway, which I think is damn fine for a portable.

And if a 12TF dock would cost $200 in 2022, it follows that a 2.7TF dock should cost significantly less, which means a cheaper product with greater penetration.

The 4K dock, IMO, is better suited to the Switch 3.
-- Switch 2 releases at 2.7TF.
-- Switch 2 and its optional power dock reaches 5.4TF.
-- Switch 3 releases at 5.4TF.
-- Switch 3 and its optional power dock reaches 10.8TF.

Each time, the optional dock is an extremely cheap Switch home console.

 

[Scott_Arm]

A console that fully supports and endorses keyboard and mouse out of the box so we can get all of the great pc strategy, moba and indie games. Bigger deal than any other specs.

 

[Syferz]

The narrowing of our expectations is interesting.

Does this all bolster the case for launching base and higher end at the same time?

--Purpose--

The current mid-gen consoles were released to combat migration to PC. In the next generation, releasing their equivalent at launch would be a means of combating both migration and a less and less favourable cost: performance ratio.

--Conservative + cheap base console--

A PS5 at 8.4TF would offer the same kind of improvement as PS4>Pro. It would set a new baseline far in excess of the PS4, would be enough of a leap to not upset PS4Pro owners, and could potentially run any 1080p30 PS4 game at 4K60, or any PS4 PSVR game at 4K120.

But with such a small gulf in performance between it and the X1X, the latter would be its cheaper competitor.

• A 4-core, 8 thread CPU would probably suffice if clocked high enough. As long as it's capable of running PS4Pro code.
• 16GB of GDDR6 would only need 8 chips and reach 576GB/s bandwidth.
• ~48GB of NVME could make up for the small increase in memory, and potentially allow for games to be played without having to be installed to the HDD - just one long loading screen at start up.
• A UHD drive and 750GB HDD should be the bare minimum as game sizes increase and digital distribution becomes more popular.
• 2-4GB of DDR4+secondary processor for the OS, home screen, and storing in suspense/running 1 app.

That's a solid console, which seems quite cheap to manufacture, and would set a decent baseline for the next generation.

--Two tiers--

I'd be happy with the above, but the market is now primed for two tiers of the same console and launching both at the same time would target two demographics:

1. Those who buy at launch and want the cheapest next-gen option available.
2. Those who buy at launch, and want the best next-gen option available.

The trouble is, we've yet to really see how a more expensive and more powerful console will sell. In recent times at least. The PS3 was a year later, more expensive, less powerful, with less available memory, and the harder platform for which to program. Other than being a year late - and memory bandwidth instead of size - the XB1 suffered the same fate.

The best comparison is the X1X. I'm not sure how it's selling in various metrics, but I understand that a greater ratio of XBoxOne purchases are the X than is true of the PS4 and the Pro.

--Pro specs--

I can imagine two configurations for a PS5Pro, and I'm basing them on a simultaneous base and Pro launch:

1) In the style of the PS4Pro.

Not much difference in price - just a couple of hundred dollars.

-- Another butterfly wing GPU bringing performance up to 16.8TF.
-- Either the same 4-core 8 thread CPU, or a 6-core 12 thread CPU. Either way, clocked higher than the base model.
-- 24GB of GDDR6 with 864GB/s bandwidth. Or the same amount of HBM3 if price permits.
-- Larger amount of NVME.
-- 1TB HDD.

2) Dual resources.

Take the butterfly wing design a step further, and apply it to the whole system. Much more expensive than the base model. Something like 700 dollars to the base models 400.

Bundle it with a collection or two of 4K60 remasters. Something that will cover a lot of bases, like Uncharted 1-4 + Lost Legacy + Gran Turismo Sport.


I'd love the latter, as it's a USP I've never seen a console possess: double the power or double the players. It'll probably be the former though.

Just a couple thoughts on this.

• We are looking at 2019 according to Matt over on Neogaf/Resetera, who certainly has a solid track record.
• The PS4 (1843GFLOPs) to PS4 Pro (4.2TFLOPs) is actually a 128% increase in performance, having that happen again is about 10TFLOPs, which is currently the range we should expect. (according to Matt when presented with 10TFLOPs to 12TFLOPs, he said it wasn't unreasonable, but that 14TFLOPs were)
• PS4 Pro works best as a mid gen refresh, it allows your customers to double dip, retains people looking at PC hardware and allows you to come in at a reasonable market price, a similar upgrade to PS5, would be just under 22TFLOPs, and if we expected it 3 years after PS5 launches, that would put it holiday 2022-2023. I guess at this point, it's goal would be to target 8K.
• 16GB might actually be a little light for a PS4, I'd suggest 24GB for PS5, and 32GB for the PS5 Pro a few years later.
• Not sure 100GB bluray discs make sense for Next gen over similar priced 64GB game cards, with higher capacities being available for more. (this priced similar is on the idea that 66GB bluray discs are similarly priced as 8GB Switch game cards, but that these capacities continue to grow at this price point and may see 64GB cards be similarly priced in 2020. This would mean no installs for physical media, a big complaint over the 6th generation.

True, but I don't think it would be worth their while to target native 4K with the dock, I think PS4Pro/X1X levels of performance would suffice.

If we're talking a 2.7TF Switch 2, doubling that with the dock puts it pretty close to the X1X anyway, which I think is damn fine for a portable.

And if a 12TF dock would cost $200 in 2022, it follows that a 2.7TF dock should cost significantly less, which means a cheaper product with greater penetration.

The 4K dock, IMO, is better suited to the Switch 3.
-- Switch 2 releases at 2.7TF.
-- Switch 2 and its optional power dock reaches 5.4TF.
-- Switch 3 releases at 5.4TF.
-- Switch 3 and its optional power dock reaches 10.8TF.

Each time, the optional dock is an extremely cheap Switch home console.

They could offer a traditional console that acts as a dock, there is no reason they couldn't offer both, the benefit with a 4K dock is you don't need to have the Switch hardware in it, just the GPU and VRAM, that allows you to use the Switch's CPU without the GPU, for higher clocks, allowing much more performance at a similar price to a dock with an entire Switch inside of it. These docks sales are a non issue btw, it's more or less an accessory at this point, it doesn't need much to support it, rendering assets that exist in multiplats for other consoles/PCs at a higher resolution doesn't cost Nintendo anything, and allows them to compete with Sony and Microsoft in power, similarly to the way PS4 Pro and Xbox 1X compete for this market.

I don't dislike the idea btw, I just think that a mid gen .5 upgrade makes more sense for those higher performance Switch devices, and is why I think they will shrink X1 and use higher clocks to establish a better experience for people wanting multiports from XB1 and PS4. Again not a giant leap from the GPU, only 250-300GFLOPs undocked and double when docked, but the CPU gain would be substantial. Doing something similar with Switch 2 in 2024-2025, with say just under 4TFLOPs and 3GHz clock to smooth out experiences and help meet multiplayer PS5 games (~10TFLOPs) would be a nice move for aging hardware. I can't make any predictions about Switch "3" as that would likely come in 2027-2028, and any predictions 10 years out? I would guess 3nm+ would be on the table, you'd probably be able to push 7-8TFLOPs out of the thing as a handheld, ARM64 CPUs should be more than capable of gaming tasks by then, considering how insane A75 cores are today. RAM might look very different, so I'll just wait to see what wins out there. The device would likely compete with 25-30TFLOPs game consoles though, so it would certainly need the performance.

 

[Shifty Geezer]

Not sure 100GB bluray discs make sense for Next gen over similar priced 64GB game cards, with higher capacities being available for more. (this priced similar is on the idea that 66GB bluray discs are similarly priced as 8GB Switch game cards, but that these capacities continue to grow at this price point and may see 64GB cards be similarly priced in 2020. This would mean no installs for physical media, a big complaint over the 6th generation

There's a long discussion on alternative distribution formats. Please direct format ideas there, and read it to learn about actual differences in costs (carts aren't similarly priced!)

No, there's 130 pages of discussion elsewhere. ~We don't want another 130 pages of discussion here. Keep the conversation there. This thread is about the CPU, GPU, RAM, mostly.

 

[iroboto]

A console that fully supports and endorses keyboard and mouse out of the box so we can get all of the great pc strategy, moba and indie games. Bigger deal than any other specs.

This is probably where we need CPU though and a lot more than Jaguar

 

[east of eastside]

This is probably where we need CPU though and a lot more than Jaguar

ARM Ares next-gen Cortex A75 *successor* will be unveiled in May. Clocked at 3.0 GHz it will be 2-to-3x of Jaguar while offering far superior perf/watt and perf/mm2 metric over Ryzen, amongst a host of other advantages including the ability to produce a *true* 8 core console, thanks to Dynamic IQ, and be coupled with low power OS cores.

Ryzen is a server/desktop 'big' core. It is never going to be used in a console which is a low power, consumer embedded device. *Never*.

Disagree with me, whoever, all you want (was also told I was wrong when I said day-one NX would be ARM and cartridges..), but you've read this and can keep it in mind when it inevitably becomes true.

Microsoft and Sony don't give a damn about AMD Ryzen premium IP and need to respond to ARM-based competitive challenges from Apple, Google and others in the future.

ARM not Ryzen will be the true next-gen console core, mark my freaking words! I can post the tweets where I was 'wrong' before way before what the majority where saying.

 

[Shifty Geezer]

Disagree with me, whoever, all you want (was also told I was wrong when I said day-one NX would be ARM and cartridges..), but you've read this and can keep it in mind when it inevitably becomes true.

Like all other Nintendo handhelds? You're quite the clairvoyant...

For those of us discussing technical merits without needing to refer to our egos as proof, it's obvious that if there's a viable ARM core and the economics are good, it's an option. For such a discussion, it's important not to be negatively biased against one of the options as you are. That means getting good data for the options, and not trusting blindly to performance and price forecasts which aren't guaranteed to pan out as the companies plan and advertise. Indeed, technology often under-delivers on promises and timelines.

I guess ARM fanboys makes a nice change from SEGA/Sony/MS/Nintendo fanboys.

 

[itsmydamnation]

ARM Ares next-gen Cortex A75 *successor* will be unveiled in May. Clocked at 3.0 GHz it will be 2-to-3x of Jaguar while offering far superior perf/watt and perf/mm2 metric over Ryzen, amongst a host of other advantages including the ability to produce a *true* 8 core console, thanks to Dynamic IQ, and be coupled with low power OS cores.

Ryzen is a server/desktop 'big' core. It is never going to be used in a console which is a low power, consumer embedded device. *Never*.

Disagree with me, whoever, all you want (was also told I was wrong when I said day-one NX would be ARM and cartridges..), but you've read this and can keep it in mind when it inevitably becomes true.

Microsoft and Sony don't give a damn about AMD Ryzen premium IP and need to respond to ARM-based competitive challenges from Apple, Google and others in the future.

ARM not Ryzen will be the true next-gen console core, mark my freaking words! I can post the tweets where I was 'wrong' before way before what the majority where saying.

Man your a flip flopper, First everything was going to be on a GPU and now its all about the CPU, So what GPU are you going to use with that CPU, im sure mobile soc developers will be lining up to give you mobile SOC lever GFX.

Also love how you compare tomorrows ARM CPU to yesterdays AMD CPU.

Next what memory are you going to Use, who's IP will that be? You can wave your hands all you like but what your sprouting is high risk and high cost (pro tip, integration is always costly). You want to see what high risk, high cost looks like, just look at the PS3. Now ask yourself do MS or Sony want another PS3?

 

[Globalisateur]

ARM Ares next-gen Cortex A75 *successor* will be unveiled in May. Clocked at 3.0 GHz it will be 2-to-3x of Jaguar while offering far superior perf/watt and perf/mm2 metric over Ryzen, amongst a host of other advantages including the ability to produce a *true* 8 core console, thanks to Dynamic IQ, and be coupled with low power OS cores.

Ryzen is a server/desktop 'big' core. It is never going to be used in a console which is a low power, consumer embedded device. *Never*.

Disagree with me, whoever, all you want (was also told I was wrong when I said day-one NX would be ARM and cartridges..), but you've read this and can keep it in mind when it inevitably becomes true.

Microsoft and Sony don't give a damn about AMD Ryzen premium IP and need to respond to ARM-based competitive challenges from Apple, Google and others in the future.

ARM not Ryzen will be the true next-gen console core, mark my freaking words! I can post the tweets where I was 'wrong' before way before what the majority where saying.

We need to think about the tools and software for developing games with any hardware. For PS4, Cerny began by studying X86 and the software (before convincing Sony to use it) and found it was a mature platform with plenty of tools meaning it's easier and quicker to make a game on it, and optimizing is also easier.

Also PS5 is rumored to have PS4 BC. Because of those reasons I don't think Cerny is going to pick ARM for his PS5 (as main CPU, the ARM secondary processor will still be invisible from developers POV) even if it's a slightly faster CPU (on paper).

 

[function]

ARM Ares next-gen Cortex A75 *successor* will be unveiled in May. Clocked at 3.0 GHz it will be 2-to-3x of Jaguar while offering far superior perf/watt and perf/mm2 metric over Ryzen, amongst a host of other advantages including the ability to produce a *true* 8 core console, thanks to Dynamic IQ, and be coupled with low power OS cores.

Better performance than Zen (2) on what process, and at what frequency, and with what level of core count scaling?

You just say things, but never cite anything beyond vague PR releases and useless benchmarks from which you extrapolate to a ludicrous degree.

Ryzen is a server/desktop 'big' core. It is never going to be used in a console which is a low power, consumer embedded device. *Never*.

Well it's made it into 15W laptops already, even with 10 active CU's on SoC for graphics. After a 15W laptop, a 150W console on a more power efficient process isn't a huge stretch.

Disagree with me, whoever, all you want (was also told I was wrong when I said day-one NX would be ARM and cartridges..), but you've read this and can keep it in mind when it inevitably becomes true.

Who the hell ever told you that the next Nintendo portable wouldn't support cartridges and absolutely wouldn't use ARM?

Nintendo used ARM and carts in their next portable. Fuuuuuuuk who could have seen that as a possibility.

Microsoft and Sony don't give a damn about AMD Ryzen premium IP and need to respond to ARM-based competitive challenges from Apple, Google and others in the future.

"Premium IP".

Zen is in 15W laptops that undercut Intel competitors, and it has completely replaced Jaguar in AMDs product stack.

ARM not Ryzen will be the true next-gen console core, mark my freaking words! I can post the tweets where I was 'wrong' before way before what the majority where saying.

The self promoted "myth of you" is not a reasonable or interesting point of consideration for the rest of the forum.

Stick to copy and pasting groundless claims from the comments section of tech websites and presenting them as positions of your own.

 

[Pixel]

Disagree with me, whoever, all you want (was also told I was wrong when I said day-one NX would be ARM and cartridges..), but you've read this and can keep it in mind when it inevitably becomes true.

Who told you that you were wrong? I certainly not the majority of members on this forum. 3DS, DS and Gameboy Advanced all used ARM based cpus.

 

[Syferz]

As soon as Nintendo merged their hardware divisions, I knew it would be ARM. A15 matched jaguar cores clock for clock iirc, A57 is superior clock for clock (switch is clocked too low to take advantage of it), A75 is a very cool core that would be good for a gaming console, and if it's successor is more efficient and higher performance, I can absolutely see Microsoft adopting it. Sony will keep with AMD from everything I've been hearing and that's fine, CPUs just don't matter in game consoles.

 

[Shifty Geezer]

It makes even less sense for MS to adopt ARM given BC with XB1 and PC cross compatibility. Not saying it won't happen, but it's the least likely one to change.

 

[msia2k75]

One thing i wonder if will they (by they i mean Sony and MS) keep a monolithic die or going with a MCM?