Sony's Next Generation Portable unveiling - PSP2 in disguise

Foundry for sure.
I wouldn't know about design, but wouldn't Sony have the workforce (and competence) for that task?

I don't know, they've traditionally partnered with Toshiba on that.

But PS1, PS2, and PSP had way more custom logic than PSVita appears to have.
 
I assume it's for some level of contribution to the design of the SoC too if they went with Samsung.

Apparently, Sony still had their own customizations added on the processor side (software/hardware?) with their + on the 543MP4+.
 
So, you wouldn't consider the whole vita SoC to be custom then given that it isn't an off the shelf part?

Just because I said MORE custom logic doesn't mean I consider a custom SoC to have NO custom logic, much less do I not think a custom SoC to be a substantial workload. But look at what is being compared, PSVita has:

- Cortex-A9 quad-core with NEON that's probably fully generic ARM IP
- SGX543MP4 with what likely amounts to small modifications

We don't know yet what off the shelf IP is used auxiliary to the two, ie if they use an ARM L2 controller or not. But CPU and GPU are the two most critical blocks in the SoC and they were clearly not designed for Sony. I don't know about the peripheral IP that such an SoC would have but I'd be surprised if in light of this Sony rolls their own decode blocks. We'll see if it bothers with audio hardware at all beyond the basic codec functions, or if that's not better delegated to cores.

Compare with the custom IP on the systems I listed. Note when I say "custom" here I mean not used in any other device (derivatives found in arcade hardware notwithstanding)

PS1: Geometry vector coprocessor (probably a hardcoded DSP), GPU, SPU, media decoders
PS2: MIPS core extension (128-bit integer SIMD), vector coprocessors, GPU, SPU, media decoders
PSP: Vector coprocessor, GPU, media decoders and whatever the hell is in VME

Can you really not see the point I'm getting at? While they licensed standard ISAs for their CPUs they all had custom vector coprocessors (far from a meager effort) and custom GPUs and a variety of other blocks that you can't find anywhere else.

To date PS3 would probably be the least customized design, using an IBM PPU + SPU Cell core that's hardly PS3-centric and a GPU that's just about one-off from an off-the-shelf part. But given how little Cell has surfaced in other products you can surmise that Sony at least played a bigger role in its development than say, MS did in that of Xenon's.

Re: Samsung, I would expect they're supplying the OLED display too.
 
Just because I said MORE custom logic doesn't mean I consider a custom SoC to have NO custom logic, much less do I not think a custom SoC to be a substantial workload. But look at what is being compared, PSVita has:

- Cortex-A9 quad-core with NEON that's probably fully generic ARM IP
- SGX543MP4 with what likely amounts to small modifications

We don't know yet what off the shelf IP is used auxiliary to the two, ie if they use an ARM L2 controller or not. But CPU and GPU are the two most critical blocks in the SoC and they were clearly not designed for Sony. I don't know about the peripheral IP that such an SoC would have but I'd be surprised if in light of this Sony rolls their own decode blocks. We'll see if it bothers with audio hardware at all beyond the basic codec functions, or if that's not better delegated to cores.

Compare with the custom IP on the systems I listed. Note when I say "custom" here I mean not used in any other device (derivatives found in arcade hardware notwithstanding)

PS1: Geometry vector coprocessor (probably a hardcoded DSP), GPU, SPU, media decoders
PS2: MIPS core extension (128-bit integer SIMD), vector coprocessors, GPU, SPU, media decoders
PSP: Vector coprocessor, GPU, media decoders and whatever the hell is in VME

Can you really not see the point I'm getting at? While they licensed standard ISAs for their CPUs they all had custom vector coprocessors (far from a meager effort) and custom GPUs and a variety of other blocks that you can't find anywhere else.

To date PS3 would probably be the least customized design, using an IBM PPU + SPU Cell core that's hardly PS3-centric and a GPU that's just about one-off from an off-the-shelf part. But given how little Cell has surfaced in other products you can surmise that Sony at least played a bigger role in its development than say, MS did in that of Xenon's.

Re: Samsung, I would expect they're supplying the OLED display too.

Yes, I see what you're saying, but it doesn't change that the SoC itself is a custom design composed of (probably) standard(ish) blocks in a custom configuration. The degree of specific custom logic is neither here nor there, it's the whole picture that matters imo.
 
The reason why the degree matters, in this context, is because I'm reasoning how much Sony may have needed outside help this time. They could have possibly done the SoC work largely themselves, with some help from the IP holders themselves.
 
The reason why the degree matters, in this context, is because I'm reasoning how much Sony may have needed outside help this time. They could have possibly done the SoC work largely themselves, with some help from the IP holders themselves.

I'd suggest that it's less a case of needing help and more a case of doing what is most cost effective.

John.
 
http://www.xbitlabs.com/news/multim...an_Act_Like_Controller_for_PlayStation_3.html
Don t know if it s been posted but this could open the ps3 to interesting stuffs like mmo and cool new rpg ... that could be a seriously cool replacement to the mouse/keyboard
Except you'd be designing you game for a very niche audience, so it'll likely see very little adoption in software, meaning less reason to buy one - the usual Catch 22 that afflicts console peripherals. It'd be a great interface option in some cases, like Wuu. Drawing LBP content on a screen directly is about as intuitive as it could get, and easy to manage than Move as people's fine motor skills are more developed at the finger level. I can't see it gaining traction though, and I don't see it's a Sont strategy either that they'll actively pursue. It's more an observation, "devs can do this," but they probably won't. Maybe Wuu ports will make their way onto the PS3<>PSV combo?
 
If they don't release until well into 2012, how close is it going to be to the release of products with A15 and PowerVR Series 6?
 
In early 2012, it's launching on the wrong side of process transition. I am afraid this thing will be decimated within a year at best, low level access notwithstanding. This will be the last gen of portable gaming devices.

It's only hope is to make use of it's new control features and strong interaction with the rest of the ecosystem.
 
In early 2012, it's launching on the wrong side of process transition.
Well, if they decide a 3D Display is worth the trouble, they can easily release an upgraded variant with (at least?) twice the 3D performance on 32nm. For low-level compatibility reasons they're stuck with SGX and cannot move to Rogue but there's theoretically plenty of life left to SGX. We don't know how well it scales at the maximum of 16 cores but obviously performance won't suddenly stop increasing at 6 or 8 cores and you'll get higher clocks via 32/28nm and High-K as well. If they triple performance (e.g. twice the cores at 1.5x the clocks) with a move to 28nm then they should be able to come close to the ST-Ericsson A9600 and beat it with low-level access. Power consumption would still be lower than the original 45nm SoC if they don't increase CPU performance. We'll see if they're smart enough to do that.

This will be the last gen of portable gaming devices.
I agree completely that there won't be any further generations (or they will fail horribly) but I'm not yet ready to write off this generation. Also it's safe to say that with the possible exception of Apple (if they decide to scale up SGX in 2012 further before going to Rogue in presumably early 2013), the PSP2 will still have the graphics lead (before even considering low-level access) for one full year.
 
Well they could prepare for the process transition just like they did with the PSP. Limit the clocks then later boost them when the new model has acceptable power draw.
 
Well they could prepare for the process transition just like they did with the PSP. Limit the clocks then later boost them when the new model has acceptable power draw.

Huh? How would games programmed for the higher clocks work on the original models?:???:
 
Well, if they decide a 3D Display is worth the trouble, they can easily release an upgraded variant with (at least?) twice the 3D performance on 32nm. For low-level compatibility reasons they're stuck with SGX and cannot move to Rogue but there's theoretically plenty of life left to SGX. We don't know how well it scales at the maximum of 16 cores but obviously performance won't suddenly stop increasing at 6 or 8 cores and you'll get higher clocks via 32/28nm and High-K as well. If they triple performance (e.g. twice the cores at 1.5x the clocks) with a move to 28nm then they should be able to come close to the ST-Ericsson A9600 and beat it with low-level access. Power consumption would still be lower than the original 45nm SoC if they don't increase CPU performance. We'll see if they're smart enough to do that.
I am not sure that they will want to upset the console model just for the sake of 3D.
I agree completely that there won't be any further generations (or they will fail horribly) but I'm not yet ready to write off this generation. Also it's safe to say that with the possible exception of Apple (if they decide to scale up SGX in 2012 further before going to Rogue in presumably early 2013), the PSP2 will still have the graphics lead (before even considering low-level access) for one full year.
This gen won't fail horribly, but it would be lucky to break even if it hw is a loss leader. 3DS is already there. The real problem is utility. When you already have a smartphone, why get another device which doesn't do anything your phone doesn't do?
 
Back
Top