Support NeoGAF

[KeplerL2]

You're stuck in this mindset like it's a hard fast rule.

Every generation has different parameters and conditions that have to be examined individually.

• Diminishing returns on silicon
• Increased costs for advanced chipsets
• Pandemic recovery
• Global economy
• The seeming discontinuation of Xbox as a platform in Year 5
• GTA6
• EA FC, Madden, and CoD all eventually going current-gen only

If Sony was sticking to the roadmap you think they are, the PS5 Pro would have come out in 2023 instead of 2024.

If you actually looked at the trends, each PS generation has lasted longer than the one before it, with PS3 to PS4, and PS4 to PS5 both being 7.

The PS6 won't come out until 2028 at the earliest. The PS Handheld will either come out next year or in 2027.

The handheld is coming at the same time or slightly later than the PS6.

 

[Three]

You motherfuckers were crying like little bitches about how the Series S was "holding back gaming" and now you masturbate over a low-spec Playstation SKU, which developers will have to target.

The fundamental differences being

1) this is for handheld games
2) has the same amount of memory
3) most importantly, it isn't mandatory so doesn't hold back anything

This is a new flag like 'base' mode or 'trinity' mode." He also pointed out this isn't required yet

 

[nial]

You're stuck in this mindset like it's a hard fast rule.

Every generation has different parameters and conditions that have to be examined individually.

• Diminishing returns on silicon
• Increased costs for advanced chipsets
• Pandemic recovery
• Global economy
• The seeming discontinuation of Xbox as a platform in Year 5
• GTA6
• EA FC, Madden, and CoD all eventually going current-gen only

The need for continued momentum in unit sales is exactly what overcomes all of this. Going from 2028 instead of 2027 changes absolutely nothing for the pandemic recovery or diminishing returns. Nothing.

If Sony was sticking to the roadmap you think they are, the PS5 Pro would have come out in 2023 instead of 2024.

PS5 Pro is purely aimed at the hardcore base. It's rarely a factor when we're clearly talking about devices with wide market appeal, PS5 Pro is not one of them.

If you actually looked at the trends, each PS generation has lasted longer than the one before it, with PS3 to PS4, and PS4 to PS5 both being 7.

If we actually look at trends, no, PS2 lasted 6 years and 8 months, PSP lated 7 years, PS3 lasted 7 years in the West and 7 years and 3 months in Japan (longest PlayStation generation yet), PS Vita lasted 7 years and 2 months until discontinuation, and PS4 lasted 7 years in the West and 6 years and 8 months in Japan (exactly like PS2). It's been a very consistent trend since 2000.

The PS6 won't come out until 2028 at the earliest. The PS Handheld will either come out next year or in 2027.

I don't see any benefit for Sony on launching two devices with unified libraries 2 years from each other.

 

[Darsxx82]

The fundamental differences being

1) this is for handheld games
2) has the same amount of memory
3) most importantly, it isn't mandatory so doesn't hold back anything

The handheld is coming at the same time or slightly later than the PS6.

 

[HeisenbergFX4]

The handheld is coming at the same time or slightly later than the PS6.

You likely know better than me but sounds like they want to launch them together if possible

 

[Zathalus]

You'd hope the PS6 will launch in 2027. A SoC design that is finalised this year would look pretty dated in 2028 or later.

 

[saintjules]

The handheld is coming at the same time or slightly later than the PS6.

If it's at the same time, that would be interesting. Would wonder what the strategy of that will be.

Maybe it's a gimmick where you can use both the Console and Handheld simultaneously for certain games.

 

[Three]

Ok? What does this change if it's not mandatory. It's like saying the switch is holding back PS5.

 

[pacman4000]

nah, even for RDNA3 and 4 you can do *64. the dual issue FP32 of RDNA3/4 has shown basically zero performance benefits.
the PS5 Pro shows the exact raster performance you'd expect from a 16TF RDNA2 gpu, while technically being a 32TF RDNA3+ gpu.

Technically the Pro has, IIR Cerny correctly, a 16TF "RDNA2+" GPU with no dual issue implemented.

 

[nial]

How long until we get another NeoGAF > Reddit > NeoGAF

 

[Darsxx82]

Ok? What does this change if it's not mandatory. It's like saying the switch is holding back PS5.

If it's not a mandate, it completely changes the concept and perspective of the product...

Good luck then with native support for third-party games and even Sony's first-party games developing PS6 games.
This point would leave it in the position of a "complementary device" or a PS Portal 2 .

 

[Ev1L AuRoN]

A PSP binary compatibility with PS5 looks far-fetched to me. PS5 hardware is not meant to be mobile, and reducing it will crawl its performance to values much lower than more modern mobile chips. Meaning, games will need to be ported or, at very minimum, recompiled to this new SKU. At this point, it makes sense to go the ARM route.

 

[nial]

A PSP binary compatibility with PS5 looks far-fetched to me. PS5 hardware is not meant to be mobile, and reducing it will crawl its performance to values much lower than more modern mobile chips. Meaning, games will need to be ported or, at very minimum, recompiled to this new SKU. At this point, it makes sense to go the ARM route.

I think that could be why Sony is doing this 2.5 years from the actual launch. It will require a ton of testing, QA and actual developer patching if needed.

 

[Three]

A PSP binary compatibility with PS5 looks far-fetched to me. PS5 hardware is not meant to be mobile, and reducing it will crawl its performance to values much lower than more modern mobile chips. Meaning, games will need to be ported or, at very minimum, recompiled to this new SKU. At this point, it makes sense to go the ARM route.

Why if It's an AMD SoC? The steamdeck is binary compatible (for the most part) with much more powerful PCs.

 

[Ev1L AuRoN]

Why if It's an AMD SoC? The steamdeck is binary compatible (for the most part) with much more powerful PCs.

Sony goes to extreme lengths to ensure binary compatibility in their products, just read about the Pro consoles, PS4 e PS5 pro.
On PC, you have the shader compilation pass to deal with, that is what made them binary compatible. On consoles in the other hand, the shaders come compiled, which save a lot of CPU time and help to deal with stutters.

TBH I don't fully understand either, but from what I understand. Being binary compatible mean an easier time for the developers, and it keeps costs to a minimum to support the new SKU, specially being a more niche product that won't move the same numbers as their main system.

When the architecture changes. They will need to resource to emulation or a wrapper to translate the code to the new SKU. Historically, Sony always shies away from emulation, preferring hardware compatibility. To be precise, for this to work like a PC handheld, Sony must port over their API to be hardware-agnostic or the very least compatible with the new architecture.

 

[UltimaKilo]

The handheld is coming at the same time or slightly later than the PS6.

I hope you are right on that timeline.

 

[Karim]

You'd hope the PS6 will launch in 2027. A SoC design that is finalised this year would look pretty dated in 2028 or later.

I don't follow you. PS6 is a totally different hardware and we already know the rumoured target specs.

 

[Zathalus]

I don't follow you. PS6 is a totally different hardware and we already know the rumoured target specs.

Some want the PS6 to launch later, but the SoC design is getting finalised this year so releasing it later then 2027 won't get you a more powerful console, it would just look dated.

 

[nial]

I don't follow you. PS6 is a totally different hardware and we already know the rumoured target specs.

He's talking about this:

PS6 is design complete and in pre-si validation already, with A0 tapeout scheduled for late this year.

If it follows PS5 schedule: Q4 2025 A0 tapeout, H1 2026 1st party dev kits, H2 2026 B0 tapeout, H1 2027 3rd party dev kits, H2 2027 launch.

 

[DevonExplorer]

Handheld only has 1/3 the memory bandwidth of the PS5 (although more cache and improved memory compression), it's pretty obvious that a mode that cuts bandwidth in half is meant to simulate the handheld.

What is odd is that PSSR is disabled in this low power mode though. Think you said in another thread that the PS handheld has AI upscaling, so why not use it in this mode as well then.

 

[nial]

What is odd is that PSSR is disabled in this low power mode though. Think you said in another thread that the PS handheld has AI upscaling, so why not use it in this mode as well then.

Same reason there's no DLSS on BC Switch 1 games on Switch 2. There's no PSSR on base PS5, either, this would only apply to PS6 games on the PS handheld.

 

[kevboard]

Technically the Pro has, IIR Cerny correctly, a 16TF "RDNA2+" GPU with no dual issue implemented.

it's RDNA3+ (some RDNA4 stuff like RT)

so it should have dual issue, but maybe they even turned it off, as it's as I said completely useless for gaming

 

[Loxus]

it's RDNA3+ (some RDNA4 stuff like RT)

so it should have dual issue, but maybe they even turned it off, as it's as I said completely useless for gaming

It's RDNA2.x

3:14 mark.

 

[kevboard]

It's RDNA2.x

3:14 mark.

ok, but like I said, it's not like it makes a difference. you basically have to half RDNA3 tflops values to have a decent comparison to RDNA2, that's all I'm saying. AMD's dual issue fp32 can not be used by games in any meaningful way... or by most programs in fact.

 

[sachos]

You motherfuckers were crying like little bitches about how the Series S was "holding back gaming" and now you masturbate over a low-spec Playstation SKU, which developers will have to target.

Came here to say this, im not a fan of the handheld trend because it has the potential to hold back the biggest games, specially if it has to run PS6 games too. But hey, it is what it is.

 

[rickybooby87]

Riky vindicated?

 

[Shane89]

this is stupid

 

[poppabk]

Why would this be on the PS5 pro though if it was handheld related?

 

[FingerBang]

So in recognizing that, maybe you should think about what Sony should be doing with the PS6 as it relates to tech.

I don't think Sony, nor MS or Nintendo, can really do much at this moment in time. They're limited by the slowdown of Technology after several decades of exponential growth/progress.

Nintendo had to drop their gimmicks and made the most iterative product ever (still great, don't get me wrong). They can improve software, reduce friction, but I can't see how new hardware in 3-4 years will be able to offer generational improvements.

Some kind of technological breakthrough is needed and I'm confident it will happen. Just not by the next generation. Unless we're willing to pay 1000+ for a more substantial jump in performance.

Wouldn't be shocked to see the PS5 gen last even longer than 8 years tbh

 

[FireFly]

I don't think Sony, nor MS or Nintendo, can really do much at this moment in time. They're limited by the slowdown of Technology after several decades of exponential growth/progress.

Nintendo had to drop their gimmicks and made the most iterative product ever (still great, don't get me wrong). They can improve software, reduce friction, but I can't see how new hardware in 3-4 years will be able to offer generational improvements.

Some kind of technological breakthrough is needed and I'm confident it will happen. Just not by the next generation. Unless we're willing to pay 1000+ for a more substantial jump in performance.

Wouldn't be shocked to see the PS5 gen last even longer than 8 years tbh

They could get to maybe 2.5X faster in rasterisation and 3.75X faster in RT, and then use that to enable Path Tracing at 1080p, while upscaling to 4K via FSR 4/5. The same game would run on the PS5 with limited RT effects.

 

[Zathalus]

ok, but like I said, it's not like it makes a difference. you basically have to half RDNA3 tflops values to have a decent comparison to RDNA2, that's all I'm saying. AMD's dual issue fp32 can not be used by games in any meaningful way... or by most programs in fact.

Seems like RDNA4 fixed that issue.

9060 XT - ~25 TF
5060 Ti - ~ 25 TF
7700 XT - ~ 35 TF

All three of the above GPUs perform very close to each other in raster.

 

[Physiognomonics]

Came here to say this, im not a fan of the handheld trend because it has the potential to hold back the biggest games, specially if it has to run PS6 games too. But hey, it is what it is.

The bigger games will be cross gen with PS5 for many years and this handheld will have the same amount of memory as PS5. it'll just run the PS5 games at lower resolution / framerate. In the case of XSS we can see now, the main problem is memory.

 

[dgrdsv]

9060 XT - ~25 TF
5060 Ti - ~ 25 TF

9060 XT @ 3150MHz is 25.8 TFs
5060 Ti @ 2700 MHz is 24.9 TFs
The latter is some 5-10% faster so there's still some deficit per TF in comparison.
And yes RDNA's VOPD is still not very usable. The advancements RDNA4 made come from perf/clock improvements which are mostly down to memory subsystem more than from increasing the math efficiency.

 

[KeplerL2]

9060 XT @ 3150MHz is 25.8 TFs
5060 Ti @ 2700 MHz is 24.9 TFs
The latter is some 5-10% faster so there's still some deficit per TF in comparison.
And yes RDNA's VOPD is still not very usable. The advancements RDNA4 made come from perf/clock improvements which are mostly down to memory subsystem more than from increasing the math efficiency.

Yep, RDNA4 perf per clock/CU is mostly due to improved memory compression, memory prefetching, larger L2 cache, more powerful Scalar ALU and Dynamic VGPR support.

 

[Zathalus]

9060 XT @ 3150MHz is 25.8 TFs
5060 Ti @ 2700 MHz is 24.9 TFs
The latter is some 5-10% faster so there's still some deficit per TF in comparison.
And yes RDNA's VOPD is still not very usable. The advancements RDNA4 made come from perf/clock improvements which are mostly down to memory subsystem more than from increasing the math efficiency.

Yeah, but the TF comparison between RDNA4 and Blackwell is far closer than the previous generation, as can be seen from the 7700 XT.

 

[Lagspike_exe]

Yep, RDNA4 perf per clock/CU is mostly due to improved memory compression, memory prefetching, larger L2 cache, more powerful Scalar ALU and Dynamic VGPR support.

Shame it's 3nm. I thought 2nm would be possible in 2027/2028 in a package of that size.

 

[Loxus]

I don't think Sony, nor MS or Nintendo, can really do much at this moment in time. They're limited by the slowdown of Technology after several decades of exponential growth/progress.

Nintendo had to drop their gimmicks and made the most iterative product ever (still great, don't get me wrong). They can improve software, reduce friction, but I can't see how new hardware in 3-4 years will be able to offer generational improvements.

Some kind of technological breakthrough is needed and I'm confident it will happen. Just not by the next generation. Unless we're willing to pay 1000+ for a more substantial jump in performance.

Wouldn't be shocked to see the PS5 gen last even longer than 8 years tbh

"I'd like to take a moment to do something we very rarely do, which is talk about the future. Specifically, I'd like to talk about the future potential in each of these three key areas.

First there's rasterized rendering by which I mean the conventional rendering strategies that were all we had up through PS4 Pro or so. There's not a whole lot of growth left here, it mostly has to come from making the GPU bigger or memory faster.

Ray tracing is different, it's still early days for the technology and I suspect we're in for several quantum leaps in performance over the next decade.

Machine learning though has the greatest potential for growth and that's an area we're beginning to focus on. Some of that growth in machine learning will come from more performant and more efficient hardware architectures."


If we use RDNA4 as an example and what Mark Cerny said on PS5 Pro Technical Seminar above, i wouldn't worry about the PS6 performance to be honest and doesn't include all the other hardware improvements.


"Here we did not bring in the doubled floating point math from rdna3 because achieving that bonus in performance would require a recompile for PS5 Pro and as I said, having two versions of each compiled piece of code would create more work than we're comfortable asking the developers to do."


Again from the PS5 Pro Technical Seminar above, it's very certain the PS6 would utilize an optimized dual-issue for ps6 version of games, which can give more performance while keeping the relatively GPU small.

We all here worried about the PS6 performance and it may end up having the most improvements compared to the previous generations.

 

[Euler007]

I like the idea of PS6 being a portable that plays PS5 games and letting the PS5 carry-on for 3-5 more years.

 

[mayham2199]

I like the idea of PS6 being a portable that plays PS5 games and letting the PS5 carry-on for 3-5 more years.

lol the PS5 is going to carry-on for much longer than 3-5 more years.

 

[6502]

Is this handheld link all speculation? I have no insight but my guess would be this is a low power mode for less demanding games / apps rather than a handheld. It doesn't make sense for a ps5 slim / ps5s unless it is an all digital console with a reduced catalogue.

 

[Ebrietas]

lol the PS5 is going to carry-on for much longer than 3-5 more years.

It amazes me that people still don't get we are past "generations". The hardware advancements are no longer paradigm shifting for game development like they once were. This is normal as technology matures.

Devs can barely ship a game that pushes the ps5 to its limit (that also doesn't take 5+ years and 200 million dollars to make), but somehow in a few years time they will be ready to make games so advanced that ps5, or a ps6 portable, will be holding them back? The ps5 will absolutely be relevant for many more years to come. I don't even see ps4 completely disappearing for a long time.

 

[asdasdasdbb]

It amazes me that people still don't get we are past "generations". The hardware advancements are no longer paradigm shifting for game development like they once were. This is normal as technology matures.

The hardware is definitely holding things back. But yes, it's questionable how much devs care.

 

[Imtjnotu]

Sony should just call it the PSP R . r for rebirth.

none of this vita shit...and if it gives us a portable way to play GTA6, its a wrap.

id pay $500 for it seeing as how i already did it for the switch 2

 

[Topher]

Sony should just call it the PSP R . r for rebirth.

none of this vita shit...and if it gives us a portable way to play GTA6, its a wrap.

id pay $500 for it seeing as how i already did it for the switch 2

I'm hoping they call it PSP 2 and make it BC with PSP games.

 

[DeepEnigma]

I'm hoping they call it PSP 2 and make it BC with PSP games.

I'm hoping they call it PSPSPSPSPSP with cute little cat ears on the design. Preloaded with Stray 2.

 

[Imtjnotu]

I'm hoping they call it PSPSPSPSPSP with cute little cat ears on the design. Preloaded with Stray 2.

 

[nial]

I'm hoping they call it PSP 2 and make it BC with PSP games.

It will play the PSP classics available on PS Store, so basically that.

Ape Academy 2

The apes have graduated Mix it up with 100 fiendishly difficult mini-games.

store.playstation.com

 

[DevonExplorer]

The first plausible Sony handheld specs leaks emerge - but how capable can it be?

Quietly in the background, plausible technical specifications for Sony's next generation handheld have come to light. B…

www.eurogamer.net

Based on Kepler's information though, the in-development APU has 16 UDNA compute units and 32 ROPs - similar in configuration terms to the Strix Point processor we'll see this year in the ROG Xbox Ally X and the plethora of Chinese handhelds built on existing versions of the same core processor. There are key differences though - and these could prove crucial.

First of all, as mentioned, the Sony handheld's use of the UDNA architecture gives it a generational leap or two over Strix Point, which is using RDNA 3.5. Secondly, memory bandwidth has historically been a defining limiting factor for AMD handhelds - it's one of the key reasons why Steam Deck continues to measure up fairly well against much more modern AMD-based handhelds. According to Kepler's information, Sony attempts to address this with two improvements: faster LPDDR5X memory (9600MT/s vs 8000MT/s) along with an additional memory cache on the processor itself: 16MB of MALL (Memory Access at Last Level) cache. This will deliver one third of existing PS5 bandwidth, but the MALL plus architectural improvements should make a difference.

If this sounds like an amped version of the Z2 Extreme found within the ROG Xbox Ally X, Kepler begs to differ, suggesting that the new graphics architecture within the handheld has "way way way higher perf/CU". The 3nm process should also offer density and efficiency advantages over Z2 Extreme. Combined with further leaks that development PlayStation 5 hardware is receiving a mode with reduced bandwidth, the implication is that game makers will be able to unify their PS5 game development to support the new handheld and to start work on this sooner rather than later, using existing console dev kits.

There is another benefit for this handheld over all others - the UDNA architecture, said to a be a unification of AMD CDNA and RDNA graphics (though Kepler refers to it as just CDNA 5) should, in theory, be the only handheld using Radeon graphics to support AI upscaling, such as FSR4 or Sony's homegrown alternative, PSSR. Assuming a 1080p screen, this could be a highly useful feature to have, especially bearing in mind some of the results we've seen from DLSS on Nintendo Switch 2.

Right now, it's extremely difficult to come up with any kind of projected performance level for the handheld, but the mooted "reduced bandwidth" mode for PS5 dev kits would seem to suggest that Sony intends for the handheld to run current generation software, almost certainly at lower resolutions and/or lower frame-rates. This sounds somewhat like Sony's own take on the Series S, the difference being that we'd expect to see the handheld arrive in close proximity to Sony's actual next generation console - the PlayStation 6. You'd imagine that the platform holder would also be looking at the handheld to run variants of PS6 software too. This may sound optimistic but PS6 will be using the same architecture, leaning heavily into machine learning features, which the handheld should support.

Where this leaves the competitive landscape remains unclear. Kepler believes that the upcoming Z2 Extreme found within the ROG Xbox Ally X will be AMD's last major handheld processor for some time, with nothing using UDNA on the current roadmap. Meanwhile, reports continue to suggest that Microsoft has left the conversation - it will rely on third party companies to create devices instead, like the Ally X. However, Valve continues to bide its time until an actual generational leap in mobile hardware is available for a potential Steam Deck 2 - and there's no reason it couldn't tap into the same AMD technologies as Sony. Perhaps the UDNA architecture is the way forward?

 

[Lagspike_exe]

I still think they should have waited for 2nm. Oh well, at least they can release a model with extended battery life down the line.

 

[Scrawnton]

To me it depends on form factor, size, thermals, and price. I think Steam Deck is an ugly machine that reminds me of too much why I hate laptop gaming (heat, fan noise, pc jank).

I hope they go for efficiency and a sleek design. Give me that and I'm there.