(Grain of Salt) Moore's Law is Dead: PS6 SoC codenamed 'Orion', PS6 Portable SoC codenamed 'Canis'

...if the PS5 is below 1080p it's not the CPU that's the issue.
Alot of the time the ps5 falls below 1080p is due to it struggling with rt, even lumen which is generally pretty forgiving compared to other rt solutions is a heavy performance cost on the ps5 due to rdna2 being terrible at it and rt has a cpu cost aswell which can be shores up by zen6. Bring in an architecture designed around rt and micropolygon tech and maybe even pathtracing and it will run rt workloads much faster.
 
It's weird to think PS5 might be the one PlayStation generation I skip entirely, I never would have imagined as a kid skipping one entirely, but the only reason to get one is to play GTA6 day 1, it sounds like PS6 is not far off.
 
Alot of the time the ps5 falls below 1080p is due to it struggling with rt, even lumen which is generally pretty forgiving compared to other rt solutions is a heavy performance cost on the ps5 due to rdna2 being terrible at it and rt has a cpu cost aswell which can be shores up by zen6. Bring in an architecture designed around rt and micropolygon tech and maybe even pathtracing and it will run rt workloads much faster.

Alan Wake 2, no RT, sub 900p, uneven 60fps

FF7, 1080p 60fps, no RT

Assassin's Creed Shadows, dynamic res that can drop below 1080p 60fps, no RT
 
Alan Wake 2, no RT, sub 900p, uneven 60fps

FF7, 1080p 60fps, no RT

Assassin's Creed Shadows, dynamic res that can drop below 1080p 60fps, no RT
Alan wake 2 isn't well optimized on consoles even on the pro it struggles.

Ff7 is based on ue4 which wasn't great for openworld and is pretty outdated fully raster based honestly not exactly a showcase of a technically competent game.

Assassin's Creed meanwhile in rt mode runs at a solid 30fps add a much better cpu and much more performant architecture especially in rt I think you'll be surprised.

The majority of the games that actually drop below 1080p are on ue5 or use heavy rt which throttles rdna2.

Most importantly with an updated transformer model which the handheld will be designed around the game won't need to run at 1080p. It could run at 720p and still end up having better image quality than 1080p on th3 ps5 with taa.
 
Last edited:
It will probably not be 1080p native, but we can assume that the handheld will have something like an improved PSSR/FSR4, so upscaling from much lower resolution will still look pretty decent on the local small screen.
Everyone can't botch at n videos for fake frames then be ok with fake resolutions. That's literally the same argument.
 
The thing is where do they go? Most casual players (90% of their customers) would find it hard to tell Spider-Man on PS4 apart from Spider-Man 2 on PS5 outside of the jump to 60fps and that is now the base standard for PS5 games so there won't be that leap again on PS6 so you're relying on visual fidelity alone to sell your system. In leaked documents Insomniac doubled the budget along with hardware 6x the power of PS4 for Spider-Man 2 and they were contemplating "do we see this increase on screen?"

Pretty much this, the people that think this next box is going to have some magic power to drive visuals forward that would somehow kneecap the game from being ported down to a handheld running newer hardware bewilder me.

Console users are regularly choosing 60fps performance modes, taking the downgrade in visuals because they don't matter that much anymore. I think they'll take a downgrade to have their games with them, just like we've seen the Portal (an even more specific device) do well. Third-party devs are still going to target the Switch 2 as well when they make games.
 
Last edited:
Doesn't seem like the PS5 is going to have that big, beautiful tail; they've got to try to push out hardware that makes the Switch 2 look like a Texas Instruments calculator. For the record, PS4 got it's tail snipped by the Switch, so it didn't have a tail, either.
PS4 tail was killed by Covid. They almost stopped producing PS4s to focus on PS5 production. They could have sold many million more PS4, particularly at this low price.
 
Alan wake 2 isn't well optimized on consoles even on the pro it struggles.

Ff7 is based on ue4 which wasn't great for openworld and is pretty outdated fully raster based honestly not exactly a technically competent game.

Assassin's Creed meanwhile in rt mode runs at a solid 30fps add a much better cpu and much more performant architecture especially in rt I think you'll be surprised.

The majority of the games that actually drop below 1080p are on ue5 or use heavy rt which throttles rdna2.

Most importantly with an updated transformer model which the handheld will be designed around the game won't need to run at 1080p. It could run at 720p and still end up having better image quality than 1080p on th3 ps5 with taa.

RT in most console UE5 games (like 98%) is software lumen, it doesn't have big CPU requirements (compared to hardware RT). All games you mentioned are GPU limited to be at 1080p or below.

Now imagine hardware with LESS THAN HALF GPU power of PS5, this is what "PS5PS6 Portable" will represent.
 
I really don't understand the Series S doom mongering about PS6 portable. Sure, it's some extra effort for devs, especially in. Split screen modes. But it's totally viable. Easier than making cross gen versions for sure - which are still being made for many games.

The more acute problem of a PS6 Portable scenario is making it good value as a standalone console, since few people will double dip with two $600 consoles. So it needs a dock included for TV play. And a method to play when docked, so another controller or breakaway joycons.

And for the few people double dipping with a PS6 and PS6 Portable, there needs to be really seamless sync.

Basically, PS5 and PS6 Portable (with a dock) will be the base console, and PS6 will be the Pro console.
 
With the PS5 Pro available, there is no reason to release a XBSS variant.

When you think about it, it's doesn't really make sense making a handheld chip, when you can just put a cut down version of the PS6 chip within the handheld. This saves cost and makes the handheld cheaper as well.



Imo, Orion could just be the CPU chiplet and Canis the GPU chiplet. But that's just my opinion and speculation.
 
My problem with the PS6 handheld is that it needs to have

at least 50% more CPU than the PS5
and the same amount of RAM as the PS6 (2x or more 16gb). It needs to have these two components above the PS5, otherwise the PS5 itself will be capable of playing its games. This means the PS6 handheld can't be released before 2028.

Unless the PS6 is a Sony Nintendo Switch that's weaker than the PS5 Pro but has more RAM and CPU. This last option is the only technologically viable, but it means a downgrade compared to the PS5 Pro.
 
With the PS5 Pro available, there is no reason to release a XBSS variant.

When you think about it, it's doesn't really make sense making a handheld chip, when you can just put a cut down version of the PS6 chip within the handheld. This saves cost and makes the handheld cheaper as well.



Imo, Orion could just be the CPU chiplet and Canis the GPU chiplet. But that's just my opinion and speculation.

Not sure about Orion but Canis is monolithic.
 
I still think Sony's handheld is a PS5.
A handheld on par with the PS5 is more than enough to compete even with the Switch 3...
It would serve to extend the PS5 generation, even on par with the Kinect on the Xbox 360.

Mainly, it would serve to compete with the Switch 2 in the Japanese market.
 
I still think Sony's handheld is a PS5.
A handheld on par with the PS5 is more than enough to compete even with the Switch 3...
It would serve to extend the PS5 generation, even on par with the Kinect on the Xbox 360.

Mainly, it would serve to compete with the Switch 2 in the Japanese market.
I'm starting to think it doesn't exist. Or the PS6 project as a whole is a Switch-like.
There's no point in a PS6-S, and I highly doubt the performance x watts technology will support this, even the Xbox Series S handheld seems impossible to 2026.
 
I really don't understand the Series S doom mongering about PS6 portable. Sure, it's some extra effort for devs, especially in. Split screen modes. But it's totally viable. Easier than making cross gen versions for sure - which are still being made for many games.

The more acute problem of a PS6 Portable scenario is making it good value as a standalone console, since few people will double dip with two $600 consoles. So it needs a dock included for TV play. And a method to play when docked, so another controller or breakaway joycons.

And for the few people double dipping with a PS6 and PS6 Portable, there needs to be really seamless sync.

Basically, PS5 and PS6 Portable (with a dock) will be the base console, and PS6 will be the Pro console.

You don't have to be a Series S doommonger to question this because for starters the Series S is ~75W device and this would be like ~20W....which makes it objectively different in a less viable way.

The only way you make this work is if literally isn't the same platform and there is no mandatory requirement for games to be developed for the handheld. Or the main console itself is parred way down....~100W....which I doubt Sony would do.
 
Idgaf about portable if it's separate. If it's like switch then neat, but otherwise give me the premium couch experience.
 
If they don't make games mandatory for the portable version, this would be fine. However, the Series S being mandatory handicapped Microsoft's entire generation.
 
Last edited:
nexbox only having a 192bit bus really bothers me. 6 channels meaning if they use 4 gb clusters of gddr7 we get 24gb of ram but 4gb clusters are really expensive, so more like 2, meaning 12gb of ggdr7 if it even has gddr7. it clearly seems ms is building a cheap box here
 
nexbox only having a 192bit bus really bothers me. 6 channels meaning if they use 4 gb clusters of gddr7 we get 24gb of ram but 4gb clusters are really expensive, so more like 2, meaning 12gb of ggdr7 if it even has gddr7. it clearly seems ms is building a cheap box here
There's 3GB modules as well, which gives 18GB total. It's also speculated to have DDR as well.

I do believe it wouldn't be as expensive as many make it out to be. Probably $699-$799.
 
nexbox only having a 192bit bus really bothers me. 6 channels meaning if they use 4 gb clusters of gddr7 we get 24gb of ram but 4gb clusters are really expensive, so more like 2, meaning 12gb of ggdr7 if it even has gddr7. it clearly seems ms is building a cheap box here
Magnus has 12 Memory controllers though. It's likely to be 12 X 2 gb or 12 X 3 gb
 
There's 3GB modules as well, which gives 18GB total. It's also speculated to have DDR as well.

I do believe it wouldn't be as expensive as many make it out to be. Probably $699-$799.
3gb modules? doable?
Magnus has 12 Memory controllers though. It's likely to be 12 X 2 gb or 12 X 3 gb
12x2 most likely.


still 192 feels a step back and this is going to be a budget nexbox, i just feel like MS will fuck up their memory setup like they have every gen except for xbox360
 
Last edited:
Any hint/suggestion of MALL Cache possibly being on the Magnus chip (or PS6/Orion)? 192-Bit + 32-36Gbps G7 may seem slow but 768-864GB/s paired with ~64-80mb cache at ~1.5TB/s and ~1/8th GDDR latency would be very capable, especially with good cache hit rates that'll pretty much be a given in general gaming workloads. Not to mention less contention with the CPU or OS if the CPU/SoC die has its own DDR.

I'd be blown away if both the "nextbox" and the PS6 lack MALL (infinity l3) Cache on-die. I appreciate it can eat into die space but it's a nice efficiency win, I'd even think it'd be worth sacrificing a few CUs over if die space is in short supply. If we're seeing an aggressive push in both ML & RT, I wonder if a simple ~2x in bandwidth is enough to really feed these machines throughout the gen.

Frankly, we haven't seen really good bandwidth on a console relative to overall power since the base PS4. I'd be happy to sacrifice a few flops to keep a slightly lesser GPU very well fed. Really don't want another gen of bad anisotropic filtering and ugly DoF bloom and alpha fx. Getting ample data from A to Z should be a major priority in my opinion, especially with the direction we're heading. Plus with MALL Cache in the path you get the latency win and may even see a little power usage trimmed of the top.

I also hope the PS5 Pro in terms of the overall chip/memory balance isn't a sign of things to come on PS6. Despite the impressiveness of getting PSSR running with new instructions utilising the registers, the hardware balance as a whole just seemed odd. Can't help but think it would've been better to trim it down to 54CU out of 60 rather than 60 out of 64 and get some MALL Cache on there so that that 65-70%% GPU leap could be better realised. The whole thing just seems off kilter.

I'd have loved to see them try to balance it more and even mix it up a bit to further trim the fat and fit in a tight die space budget; something like this:

PS5 Pro ($599)
=========
54 CU (60 w/ 6 disabled)
2333MHz
224-Bit Bus
16.1TF Single-Issue FP32
N4P
----
32MB On-Die MALL Cache @ 1-1.5TB/s, ~1/8th Latency & Low Power
----
14GB GDDR6X (7x16Gb @ 21Gbps) @ 588GB/s
4GB DDR5 (1x32Gb)
----
1TB SSD
Wi-Fi 6e
----
HDMI 2.1b 48Gbps


PS5 Slim ($399)
==========
36 CU (40 w/ 4 disabled)
2233MHz
224-Bit Bus
10.3TF Single-Issue FP32a
N6
----
14GB GDDR6 (7x16Gb @ 16Gbps) @ 448GB/s
2GB DDR5 (1x16Gb)
----
1TB SSD
Wi-Fi 6e
----
HDMI 2.1b 32Gbps


Universal Disc Drive ($69)
================

I just hope PS6 doesn't half-arse the bandwidth (or effective bandwidth/throughput). I suspect those GDDR chips are the easiest thing to switch out 'last minute' and I can just see the moneymen wanting to save a bit by going to 28Gbps leaving us with 896GB/s on a 256-Bit bus. With which I'd like to see one or more of [MALL Cache], [Neural Texture Decompression] or [much more efficient memory usage from UDNA/RDNA5] to close the gap.
 
Last edited:
Any hint/suggestion of MALL Cache possibly being on the Magnus chip (or PS6/Orion)? 192-Bit + 32-36Gbps G7 may seem slow but 768-864GB/s paired with ~64-80mb cache at ~1.5TB/s and ~1/8th GDDR latency would be very capable, especially with good cache hit rates that'll pretty much be a given in general gaming workloads. Not to mention less contention with the CPU or OS if the CPU/SoC die has its own DDR.

I'd be blown away if both the "nextbox" and the PS6 lack MALL (infinity l3) Cache on-die. I appreciate it can eat into die space but it's a nice efficiency win, I'd even think it'd be worth sacrificing a few CUs over if die space is in short supply. If we're seeing an aggressive push in both ML & RT, I wonder if a simple ~2x in bandwidth is enough to really feed these machines throughout the gen.

Frankly, we haven't seen really good bandwidth on a console relative to overall power since the base PS4. I'd be happy to sacrifice a few flops to keep a slightly lesser GPU very well fed. Really don't want another gen of bad anisotropic filtering and ugly DoF bloom and alpha fx. Getting ample data from A to Z should be a major priority in my opinion, especially with the direction we're heading. Plus with MALL Cache in the path you get the latency win and may even see a little power usage trimmed of the top.

I also hope the PS5 Pro in terms of the overall chip/memory balance isn't a sign of things to come on PS6. Despite the impressiveness of getting PSSR running with new instructions utilising the registers, the hardware balance as a whole just seemed odd. Can't help but think it would've been better to trim it down to 54CU out of 60 rather than 60 out of 64 and get some MALL Cache on there so that that 65-70%% GPU leap could be better realised. The whole thing just seems off kilter.

I'd have loved to see them try to balance it more and even mix it up a bit to further trim the fat and fit in a tight die space budget; something like this:

PS5 Pro ($599)
=========
54 CU (60 w/ 6 disabled)
2333MHz
224-Bit Bus
16.1TF Single-Issue FP32
N4P
----
32MB On-Die MALL Cache @ 1-1.5TB/s, ~1/8th Latency & Low Power
----
14GB GDDR6X (7x16Gb @ 21Gbps) @ 588GB/s
4GB DDR5 (1x32Gb)
----
1TB SSD
Wi-Fi 6e
----
HDMI 2.1b 48Gbps


PS5 Slim ($399)
==========
36 CU (40 w/ 4 disabled)
2233MHz
224-Bit Bus
10.3TF Single-Issue FP32a
N6
----
14GB GDDR6 (7x16Gb @ 16Gbps) @ 448GB/s
2GB DDR5 (1x16Gb)
----
1TB SSD
Wi-Fi 6e
----
HDMI 2.1b 32Gbps


Universal Disc Drive ($69)
================

I just hope PS6 doesn't half-arse the bandwidth (or effective bandwidth/throughput). I suspect those GDDR chips are the easiest thing to switch out 'last minute' and I can just see the moneymen wanting to save a bit by going to 28Gbps leaving us with 896GB/s on a 256-Bit bus. With which I'd like to see one or more of [MALL Cache], [Neural Texture Decompression] or [much more efficient memory usage from UDNA/RDNA5] to close the gap.
There is no mention of mall cache, but RDNA5 GPUs has larger L2 cache similarly the Nvidia GPUs, ranging from 16MB to 64MB L2 cache.

If anything, I can see it having 24MB L2 cache at most.
 
Last edited:
I don't believe that it's a PS5 Portable, that would actually require every single game developed for the handheld to also have a native version for the main PS5.
I mean yeah, no issue, right? Isn't the main PS5 more powerful, anyway? I'm talking about the economics of developing for a platform that will have an almost (if not completely) reduced userbase by 2034, IF you want to release a game on a handheld that is just barely being replaced and will keep having a steady userbase up to, what? 2041? That's just not feasible.
The same reason a ton of cross-gen games in 2014-2015 skipped the Wii U despite being perfectly able to run them.
 
Last edited:
If PS6 portable won't have its own exclusive like PSP and PSVita did and just getting port of console version then the system just gonna be useless to me.

Thats the whole point. So they dont need separate studios to make enough exclusives for each console.

And the exact reason why Id buy a portable, to share my games across. Have then while i travel.

No1 expects exclusives for a portable system now that Nintendo and Steam have the library shared in hybrids
 
K KeplerL2 Do you currently know if target specs for the handheld device are necessarily as insanely expensive (say, $600 to $800) for 2027 as people like to theorize? I think people read PS5 and PS6 and immediately assume similar high level home hardware powering the thing, while that's not what sounds like to me.
 
Thats the whole point. So they dont need separate studios to make enough exclusives for each console.

And the exact reason why Id buy a portable, to share my games across. Have then while i travel.

No1 expects exclusives for a portable system now that Nintendo and Steam have the library shared in hybrids
But the point of hybrid is both TV mode and portable mode in one system, if I have to buy two separate systems for it there is no point and not having its own exclusive makes me less willing spend money on it.

When I bought Switch 1&2 I bought it for its exclusive games and portable mode was nice bonus.

It was nice option of two different way of playing games without needing to spend money for extra system.

If I'm spending money for separate system then it fucking better have exclusive games that makes it worth spending money on instead of being just port machine for console version.
 
Last edited:
K KeplerL2 Do you currently know if target specs for the handheld device are necessarily as insanely expensive (say, $600 to $800) for 2027 as people like to theorize? I think people read PS5 and PS6 and immediately assume similar high level home hardware powering the thing, while that's not what sounds like to me.
It's on N3P which is more expensive than N4P of Strix Point/Z2E, but the die should be much smaller and it's presumably going be subsidized unlike the Rog Xbox Ally so I don't think it will be very expensive.
 
It's on N3P which is more expensive than N4P of Strix Point/Z2E, but the die should be much smaller and it's presumably going be subsidized unlike the Rog Xbox Ally so I don't think it will be very expensive.
Then I wonder if they'll go $499.99 for the handheld and $699.99 for the home console (effectively taking PS5 Pro's spot on the market), with the PS5 Slim continuing to be manufactured as the most accessible entry point into the perpetual cross-gen PlayStation ecosystem that might await us until, at least, 2030.
 
If PS6 portable won't have its own exclusive like PSP and PSVita did and just getting port of console version then the system just gonna be useless to me.
Better tell Sony before they make this horrible mistake. FFS what a useless post.

I have been hoping I could move away from Sony. A portable system would be good for the kids though their library is not as kid friendly as Nintendo's. But then again my oldest will be a teen. The kicker if they do 2x skus is that you can do coop in the same room easily while not buying 2 of the same product. That may pull me in.
 
Thats the whole point. So they dont need separate studios to make enough exclusives for each console.

And the exact reason why Id buy a portable, to share my games across. Have then while i travel.

No1 expects exclusives for a portable system now that Nintendo and Steam have the library shared in hybrids
The issue many see is the resources needed to make these games run on a 15W TDP.

We barely got any PS5 exclusives because of the focus of GAAS without having dedicated Studios for those titles. Now your telling the current studios to spend extra time optimizing for the handheld?

And it's not like Sony couldn't make a PS4 handheld either. The hardware was available and nearly all of the Playstation exclusives came to the PS4 as well.

We got the PS Portal to avoid resources going into optimizing for a dedicated handheld. If Sony is going to be using resources to optimize for this new handheld at the expense of PS6 exclusives, I don't want it.

I for one don't think that PS5 Low Power mode is for a handheld either, it's just a PS5 eco mode for those with a high light bill.
 
If PS6 portable won't have its own exclusive like PSP and PSVita did and just getting port of console version then the system just gonna be useless to me.
Both playstation and nintendo have learnt having that type of seperation is a bad idea. You really need a unfified ecosystem for your devices sharing the same games.
 
Last edited:
But a portable ps6 is gonna be at best as powerful as a ps5 if even that. Won't that handicap the real ps6 if it needs to scale down to the portable version.
 
The issue many see is the resources needed to make these games run on a 15W TDP.

We barely got any PS5 exclusives because of the focus of GAAS without having dedicated Studios for those titles. Now your telling the current studios to spend extra time optimizing for the handheld?

And it's not like Sony couldn't make a PS4 handheld either. The hardware was available and nearly all of the Playstation exclusives came to the PS4 as well.

We got the PS Portal to avoid resources going into optimizing for a dedicated handheld. If Sony is going to be using resources to optimize for this new handheld at the expense of PS6 exclusives, I don't want it.

I for one don't think that PS5 Low Power mode is for a handheld either, it's just a PS5 eco mode for those with a high light bill.
I guess it depends how they approach the optimisation.

Maybe they could let the first part studios do there thing and not worry about the handheld and use the ps6 to it's fullest, then let studios like nixxes do the port down.

I'm still thinking supporting this thing will be optional, as the performance gulf between demanding ps6 games and this device will probably be huge so not all games will be ported down.
 
Last edited:
No, a 192-bit bus won't bottleneck a next-gen console if:


  • It's paired with GDDR7 @ 32 Gbps
  • You have good architecture (cache, compression, unified memory)
  • AI-driven rendering helps optimize workloads
  • With GDDR7 @ 32 Gbps:

    192-bit×32 Gbps=768 GB/s


So in a modern console context, this is a smart and forward-looking memory setup.

hmmm
 
Last edited:
the only reason to get one is to play GTA6 day 1, it sounds like PS6 is not far off.

I hope people getting a PS5 purely for one game will get the best out of it.

I don't have 100% faith in the optimization of the game, unless Sony is directly working with Rockstar to make this game the second coming of Christ.
 
Both playstation and nintendo have learnt having that type of seperation is a bad idea. You really need a unfified ecosystem for your devices sharing the same games.
But unlike Sony, Nintendo is not gonna make you buy separate system for portable mode…one system does both.
 
But unlike Sony, Nintendo is not gonna make you buy separate system for portable mode…one system does both.
Sony will make you choose, that's different. You will have to buy a separate system for either portable mode OR the powerful TV mode, not just if you want portable mode.
 
Doesn't seem like the PS5 is going to have that big, beautiful tail; they've got to try to push out hardware that makes the Switch 2 look like a Texas Instruments calculator. For the record, PS4 got it's tail snipped by the Switch, so it didn't have a tail, either.
PS4 got its tail snipped by Sony.

They discontinued it early to devote more production to PS5 during COVID restrictions.
 
Sony will make you choose, that's different. You will have to buy a separate system for either portable mode OR the powerful TV mode, not just if you want portable mode.
Nintendo give me system that has the option if I want play the game on my TV or play it on portable without paying extra for separate system.

With Sony if you don't have enough money to buy both then your only option is to buy one of them and that's gonna be your only way play your game.
 
Nintendo give me system that has the option if I want play the game on my TV or play it on portable without paying extra for separate system.

With Sony if you don't have enough money to buy both then your only option is to buy one of them and that's gonna be your only way play your game.
Yes, but it's largely impossible to ask SIE to make a hybrid device like the Switch when that would require their main PS6 to be less powerful than even the base PS5.
If SIE offers the option, maybe you will be able to buy a dock a well.
 
Yes, but it's largely impossible to ask SIE to make a hybrid device like the Switch when that would require their main PS6 to be less powerful than even the base PS5.
If SIE offers the option, maybe you will be able to buy a dock a well.
To me that just so damn pointless, make two separate system to share same library….it just waste of money.

And in the end devs still have to compromise in order for a game able to run decently on portable version.

Either make proper hybrid like Nintendo or don't fucking bother….this is just half-ass.
 
To me that just so damn pointless, make two separate system to share same library….it just waste of money.

And in the end devs still have to compromise in order for a game able to run decently on portable version.

Either make proper hybrid like Nintendo or don't fucking bother….this is just half-ass.
It's literally the only choice they have when 1. a full-on hybrid is not feasible and 2. neither are dedicated handheld games, something Nintendo agrees with.
It's an actual smart solution and it wouldn't surprise me to see consumers preferring the PS6P over the main PS6.
 
Top Bottom