PlayStation 4 (codename Orbis) technical hardware investigation (news and rumours)

Status
Not open for further replies.
Hi patsu,

The kind of remote play that you suggest requires both parties to have to full game, correct? But that does make some sense - in this case the only data that needs to be transferred is controller data (in real-time) and game-state (which can be done in advance).
 
In my view, basic peer-to-peer RemotePlay arrangement should only be used for LAN gaming.
In this case, the Vita does not need a copy of the RemotePlayed PS4 game.

To switch the game host across WAN between 2 PS4s on-the-fly, they can either:

* Mandate both parties to own the game: only need to share action and state data like co-op.
or
* Both parties Gaikai from the servers so that the host doesn't have to stream 1080p game presentation upstream (to another PS4). In this case, both parties don't need a local copy of the game technically speaking. It just happens that the host already bought a copy but he got stuck. This approach is only applicable if Gaikai's instant demo platform works as advertised.



As you can see, I don't believe in peer-to-peer RemotePlay across WAN.
 
Anyone that is familiar with the ins and outs of manufacturing/lead times/distributions etc, do you think it is remotely possible that Sony could do a worldwide holiday release?

Was it purely stock levels that stopped the PS3 from having a simultaneous release? I saw something about difficulty with distribution in Europe.
 
Anyone that is familiar with the ins and outs of manufacturing/lead times/distributions etc, do you think it is remotely possible that Sony could do a worldwide holiday release?

Was it purely stock levels that stopped the PS3 from having a simultaneous release? I saw something about difficulty with distribution in Europe.

If I remember correctly, the PS3 had to deal with blue laser diode shortages, which obviously wouldn't exist today. But the question really is, what is a "worldwide release"? Same day? Same week? Same month? Expecting Apple-style releases is impractical because the hardest part of manufacturing any modern iPhone is the exterior, not the SoC which can usually have pretty good yields because the surface area is so small.

The other issue is getting software available for launch especially in Europe, where you have to deal with multiple languages.
 
In my view, basic peer-to-peer RemotePlay arrangement should only be used for LAN gaming.
In this case, the Vita does not need a copy of the RemotePlayed PS4 game.

To switch the game host across WAN between 2 PS4s on-the-fly, they can either:

* Mandate both parties to own the game: only need to share action and state data like co-op.
or
* Both parties Gaikai from the servers so that the host doesn't have to stream 1080p game presentation upstream (to another PS4). In this case, both parties don't need a local copy of the game technically speaking. It just happens that the host already bought a copy but he got stuck. This approach is only applicable if Gaikai's instant demo platform works as advertised.



As you can see, I don't believe in peer-to-peer RemotePlay across WAN.

Dave Perry specifically mentioned in the presentation that gaikai tech is built directly into PS4 allowing for it to act as a kind of gaikai server in its own right. And that it is how the wan based remote play is implemented. You may have missed it but i watched the presentation again last night so it is fresh in my mind, i didnt really pick up on thhis the first time. I dont think we really need to speculate with that in mind. It will be used for both Remote Play on Vita and for allowing for users to take control of your game remotely over the internet. There is no reason not to use the same method for both, you will just see reduced latency over wifi. It may not work as well for helping friends with fast reflex action sequences but for things like puzzles which was the example given in the presentation.
 
Last edited by a moderator:
Dave Perry specifically mentioned in the presentation that gaikai tech is built directly into PS4 allowing for it to act as a kind of gaikai server in its own right. And that it is how the wan based remote play is implemented. You may have missed it but i watched the presentation again last night so it is fresh in my mind, i didnt really pick up on thhis the first time. I dont think we really need to speculate with that in mind. It will be used for both Remote Play on Vita and for allowing for users to take control of your game remotely over the internet. There is no reason not to use the same method for both, you will just see reduced latency over wifi. It may not work as well for helping friends with fast reflex action sequences but for things like puzzles which was the example given in the presentation.

Using some of local Gaikai hosting for remote play is going to be a disaster for most people. 720p compressed requires something like a 5+ Mbit/s connection. There's a lot of people who have upload speeds capped in the 512 Kbit/s to 2 Mbit/s speeds. Most ADSL providers for example have limits of 896 Kbit/s for upload while having anywhere from 1.5 Mbit/s to 40 Mbit/s download speeds.

So, I think it's very doubtful that WAN based remote play will use that. It'd be much smarter to require both parties to have the game from both a practical and business POV.

From a practical POV it allows for the most responsive gameplay. As well, how is a friend going to help you if he doesn't even have the game to play?

From a business POV it would prevent one person with a high upload bandwidth from sharing his game with multiple people. Say one person in a group of friends buys the game. Then everyone plays his copy remotely instead of buying the game, assuming they don't mind any input lag that would exist and he has large chunks of time where he doesn't use his PS4 (at work or while sleeping for example).

Regards,
SB
 
Dave Perry specifically mentioned in the presentation that gaikai tech is built directly into PS4 allowing for it to act as a kind of gaikai server in its own right. And that it is how the wan based remote play is implemented. You may have missed it but i watched the presentation again last night so it is fresh in my mind, i didnt really pick up on thhis the first time. I dont think we really need to speculate with that in mind. It will be used for both Remote Play on Vita and for allowing for users to take control of your game remotely over the internet. There is no reason not to use the same method for both, you will just see reduced latency over wifi. It may not work as well for helping friends with fast reflex action sequences but for things like puzzles which was the example given in the presentation.

Yes, PS4 RemotePlay is Gaikai. So is the demo server platform. It's just how they choose to use it. The server version or the client version, depending on what issues they encounter.

It's subjected to business deals though.
 
If I remember correctly, the PS3 had to deal with blue laser diode shortages, which obviously wouldn't exist today. But the question really is, what is a "worldwide release"? Same day? Same week? Same month? Expecting Apple-style releases is impractical because the hardest part of manufacturing any modern iPhone is the exterior, not the SoC which can usually have pretty good yields because the surface area is so small.

The other issue is getting software available for launch especially in Europe, where you have to deal with multiple languages.

They could surely include most native English countries then for day 1 release (Only assuming that they could manufacture enough). So US,UK,Ireland, Australia, Canada.
 
They could surely include most native English countries then for day 1 release (Only assuming that they could manufacture enough). So US,UK,Ireland, Australia, Canada.

Maybe if we moved to a digital-only system that would be possible, but I don't think developers or publishers really want to deal with more that one SKU for Europe. Publishers would have to start dealing with having too many copies if the German version and not enough of the French or Spanish version. It's too much of a headache that waiting a few weeks for all localizations to be done is preferable to them.
 
Why would you keep PS1, PS2, PSP and Vita emulators in the cloud instead of on the PS4?

That may be possible but Sony's ambition is to enable access for catalog titles across PCs, tablets and phones as well using Gaikai.
 
Dave Perry specifically mentioned in the presentation that gaikai tech is built directly into PS4 allowing for it to act as a kind of gaikai server in its own right. And that it is how the wan based remote play is implemented. You may have missed it but i watched the presentation again last night so it is fresh in my mind, i didnt really pick up on thhis the first time. I dont think we really need to speculate with that in mind. It will be used for both Remote Play on Vita and for allowing for users to take control of your game remotely over the internet. There is no reason not to use the same method for both, you will just see reduced latency over wifi. It may not work as well for helping friends with fast reflex action sequences but for things like puzzles which was the example given in the presentation.

It's interesting because in a way it is the actualisation of the grid computing idea that Sony have been playing with since the PS2. Distribute the game server infrastructure amongst all the machines that run it so that they can be local servers on a global scale.

With clever load balancing you can ensure that no one machine is overloaded and exists within the capping infrastructure exposed by the ISP. If the load is overbalanced then swap it back to the main cloud servers operated by Sony.

It may well be that they can make it happen this time around, they just need to shift a lot of boxes, the more there are then the more balanced the structure is. Though I guess they might hit the travelling salesman problem before they reach the optimum number of localised servers...
 
It's interesting because in a way it is the actualisation of the grid computing idea that Sony have been playing with since the PS2. Distribute the game server infrastructure amongst all the machines that run it so that they can be local servers on a global scale.

With clever load balancing you can ensure that no one machine is overloaded and exists within the capping infrastructure exposed by the ISP. If the load is overbalanced then swap it back to the main cloud servers operated by Sony.

It may well be that they can make it happen this time around, they just need to shift a lot of boxes, the more there are then the more balanced the structure is. Though I guess they might hit the travelling salesman problem before they reach the optimum number of localised servers...

Erm, how would you even load balance game streaming across multiple machines. One machine has to do all the rendering and video stream encoding. I wouldn't even begin to be able to guess how you could distribute that across multiple machines.

Regards,
SB
 
He probably meant passing the game states/saves from player 1 to player 2 (or a Gaikai server), and reverse streaming. In this way, the controlling player or a server is always the host. For best effect, the game will probably need to allow "anywhere save".

I remember IBM funded an outfit called Butterfly networks (?) to support a unified, scalable MMO world. It's now defunct. I heard the games need to be written in certain way to take advantage of their infrastructure. It's a MMO though; not a streaming service like Gaikai.

For Playstation, what may be interesting is how they are going to offer PS3 b/c via Gaikai as some media outlet have reported. The PS3 run-time and the app server may be different entities.
 
No turning back now but I kinda wish Sony went with 6 GBs gddr5 and spent the extra ($30?) on the GPU? Say an additional 6 CUs for a total of 24. That would have been a more balanced design but I'm guessing that would require too much change to the motherboard.
 
No turning back now but I kinda wish Sony went with 6 GBs gddr5 and spent the extra ($30?) on the GPU? Say an additional 6 CUs for a total of 24. That would have been a more balanced design but I'm guessing that would require too much change to the motherboard.

What really? have a 192bit bus? GPU compute workloads are commonly bottlenecked by throughput. So you want way more CU's then a 7850 + 8 core CPU with way less bandwidth and an expectation of nextgen Compute shaders.........


it might just be me but that doesn't compute.
 
Which one is cheaper in a short and long run, producing a system with 384bit (6GB ) or 256bit (8GB)? How much the added complexity would cost them? Of course I know it's impossible for Sony to change the design from 256 to 384bit, but let just say that this is happening at the early stage of designing the PS4, is 384bit is even possible without breaking the bank?
 
Memory controller and bus takes a lot of die space. I think it would be more cost effective to pick 7gbps chip to get up to 224 GB/s on a 256-bit wide bus. In fact they will most probably choose 7gbps chip, but will underclock/volt them to get the most units and reliability. I still think they could settle at 6gbps for 192 GB/s but even 176 GB/s is really nice with that amount on a console.

1 Million PS4 = 16 Millions high end GDDR5 chips. Good luck with that.

It's probably a better case than blu-ray diode when PS3 launched, but still challenging in production.
 
Status
Not open for further replies.
Back
Top