Is PS3's scaler working now?

[Hezz]

After reviewing the presentation it looks like the SCC is a high end programmable analog and digital I/O interface. It seems logical that scaling can be programmed here but it would be programmed with software on the SCC making the SCC flexible enough to use in many situations.

Ideas?

 

[DeanA]

Can someone explain to me why you guys seem to be so desperate for this 'unidentified chip' to be Toshiba's SCC?

I see posts saying 'well, it's got the same die size as SCC' and I can barely believe what I'm reading. That kind of assumption is pure comedy gold.

Dean

 

[Crossbar]

I see posts saying 'well, it's got the same die size as SCC' and I can barely believe what I'm reading. That kind of assumption is pure comedy gold.

Well the SCC is one of few I/O chips specifically designed to work in cooperation with the Cell so if there is no other information around I don´t see it as such a far fetched assumption, though all previous speculations have always concluded that the SCC contains a lot of redundant functionality from the PS3 point of view.

But if you can confirm that the SCC has never been part of the PS3 we can establish a brand new starting point of our speculations.

 

[Shifty Geezer]

Can someone explain to me why you guys seem to be so desperate for this 'unidentified chip' to be Toshiba's SCC?

I'd have to dig up the old discussions, but there was more reason to think that big hunk of silicon was SCC than any other chip. It's certainly big for an IO controller, no? I don't recall anyone offering convincing alternatives. What would you suggest it might be, even without insider insight? What other options are there? A Sony custom IC? In which case, why so big? What does it do? Inquiring minds want to know!

 

[expletive]

So is this a game by game implementation or can i now safely hook a pS3 up to my old-as-dirt CRT set via component and play resistance at 1080i (instead of 480p)?

 

[Gradthrawn]

So is this a game by game implementation or can i now safely hook a pS3 up to my old-as-dirt CRT set via component and play resistance at 1080i (instead of 480p)?

game-by-game still. But, Resistance at 480p is still a damn good game, IMO. That's coming from someone who's played Resistance extensively at 480p (along with MotorStorm). But the detail does pop-out out a lot more, IMO, at 720p.

 

[Truespeed]

Can someone explain to me why you guys seem to be so desperate for this 'unidentified chip' to be Toshiba's SCC?

I see posts saying 'well, it's got the same die size as SCC' and I can barely believe what I'm reading. That kind of assumption is pure comedy gold.

Dean

It's not speculation. The chip is indeed the SCC. The following text is from the Linux PS3 install documentation:

PS3 storage devices includes HDD, Blu-ray Disc drive (BD drive) and flash ROM. These devices are connected via CBE’s companion chip. Accessing these devices is performed by dedicated hypervisor calls. These hypervisor calls consist of simple straightforward methods: open, close, read, write, ioctl.Most of all methods are asynchronous, that is, methods will return immediately after call, and then the caller must wait for its completion via other method. These completions are notified by virtualized interrupts. The status of results of these asynchronous methods can be retrieved via the specific hypervisor call. Since the BD drive is basically ATAPI device, Linux can issue ATAPI commands by ioctl. Some of ATAPI commands have been rejected by the hypervisor call because of security issues.

 

[DeanA]

It's not speculation. The chip is indeed the SCC. The following text is from the Linux PS3 install documentation:

So you're taking a pretty generic term from the PS3 Linux installation guide and interpreting it as confirmation of a specific machine part? Hmmm.. a unique approach to investigation, that's for sure. Although the 'companion chip' term was used in news articles dating back to 2004.. like these:

http://www.newsfactor.com/story.xht...ut-Cell-Chip&story_id=28682&category=hardware
http://www.hpcwire.com/hpcwire/hpcwireWWW/04/1203/108902.html
http://www.extremetech.com/article2/0,1697,1732975,00.asp

OMG! I seem to have confirmed the existence of SCC before Toshiba even announced it!!! What a scoop!

Oh, and don't forget this link...!

http://www-03.ibm.com/technology/ets/portfolio/offering_cell_micro.html

It would appear from this that IBM offer custom design of... wait for it... 'companion chips' for CELL. Wow! IBM must have designed SCC after all, not Toshiba!

Dean

 

[War]

So you're taking a pretty generic term from the PS3 Linux installation guide and interpreting it as confirmation of a specific machine part? Hmmm.. a unique approach to investigation, that's for sure. Although the 'companion chip' term was used in news articles dating back to 2004.. like these:

http://www.newsfactor.com/story.xht...ut-Cell-Chip&story_id=28682&category=hardware
http://www.hpcwire.com/hpcwire/hpcwireWWW/04/1203/108902.html
http://www.extremetech.com/article2/0,1697,1732975,00.asp

OMG! I seem to have confirmed the existence of SCC before Toshiba even announced it!!! What a scoop!

Oh, and don't forget this link...!

http://www-03.ibm.com/technology/ets/portfolio/offering_cell_micro.html

It would appear from this that IBM offer custom design of... wait for it... 'companion chips' for CELL. Wow! IBM must have designed SCC after all, not Toshiba!

Dean

Best post ever

 

[patsu]

A quick question for DeanA, if you can answer it

I frequent the cbe-oss-dev mailing list (just trying to learn).
Does the spidernet module = gigabit ethernet card or the large companion chip ?

I suspect it's the former, but I can't find a definitive blurb about it.

 

[DeanA]

Does the spidernet module = gigabit ethernet card or the large companion chip?

From what I've read on that list, spidernet is a module that's normally integrated into the southbridge.. but it's described at http://readlist.com/lists/vger.kernel.org/linux-kernel/65/327431.html as not being on PS3. From the descriptions of it, I can't see why it would be needed on PS3 anyway.

I'm afraid I don't know more than that... sorry!

Dean

 

[patsu]

Argh ! Another dead end... thanks to the hypervisor. Did the leaked SDK turn up anything informative for a normal programmer ? Anyone ?

 

[Truespeed]

So you're taking a pretty generic term from the PS3 Linux installation guide and interpreting it as confirmation of a specific machine part? Hmmm.. a unique approach to investigation, that's for sure. Although the 'companion chip' term was used in news articles dating back to 2004.. like these:

http://www.newsfactor.com/story.xht...ut-Cell-Chip&story_id=28682&category=hardware
http://www.hpcwire.com/hpcwire/hpcwireWWW/04/1203/108902.html
http://www.extremetech.com/article2/0,1697,1732975,00.asp

OMG! I seem to have confirmed the existence of SCC before Toshiba even announced it!!! What a scoop!

Oh, and don't forget this link...!

http://www-03.ibm.com/technology/ets/portfolio/offering_cell_micro.html

It would appear from this that IBM offer custom design of... wait for it... 'companion chips' for CELL. Wow! IBM must have designed SCC after all, not Toshiba!

Dean

Excited aren't we? But you forgot one thing. You neglected to post a picture of your hypothetical companion chips. IBM does supply services that allow for custom Cell companion chip design and fabrication, but I've yet to hear or see any of these companion chips. So, to summarize, all you've provided are links to text that make mention of possible Cell companion chips that may or may not exist. Is that the best you can do? Should I use the obligatory OMG to emphasize my point?

Now, the chip in question is labeled CXD2973GB. According to iSuppli (I hear they're pretty credible and accurate) this chip is manufactured by Toshiba and referred to as the I/O Bridge controller. The cost of this chip is $59 USD, only $30 less than the powerful and expensive Cell BE. It would seem to me that $59 USD is quite expensive for a simple I/O bridge controller.

Now, I know this is not definitive proof that this chip is indeed the Toshiba Super Companion chip, but that fact that it's a $59 USD I/O bridge controller chip manufactured by Toshiba does point to this chip being the Toshiba SCC.

 

[Truespeed]

Argh ! Another dead end... thanks to the hypervisor. Did the leaked SDK turn up anything informative for a normal programmer ? Anyone ?

This is from the SDK file list

C:\cell\samples\graphics\SpecialFeatures\1080pUpscaling
1080p_upscaling.vcproj
main.cpp
Makefile
readme_e.txt
readme_j.txt

Make of it what you will.

 

[DeanA]

Excited aren't we?

Why yes! Yes, I am!!

Now, the chip in question is labeled CXD2973GB. According to iSuppli (I hear they're pretty credible and accurate) this chip is manufactured by Toshiba and referred to as the I/O Bridge controller. The cost of this chip is $59 USD

You hear 'credible and accurate', while I look at the figures and think 'way off the mark'. Sure, they can pull figures out of any available orifice when trying to price standard parts like Wi-Fi/Bluetooth modules, but with a closed IC that's not (to my knowledge) been stripped down and analysed, they're just making prices up off the top of their heads.

Now, I know this is not definitive proof that this chip is indeed the Toshiba Super Companion chip, but that fact that it's a $59 USD I/O bridge controller chip manufactured by Toshiba does point to this chip being the Toshiba SCC.

You're right.. it's not definitive proof at all. There's no proof it's a $59 IC, and - even with your leaps of faith - there's still no proof that it's the Toshiba SCC you so desperately want it to be.

My personal view is that it's more likely to be a custom part for PS3 (developed by Toshiba) which draws some elements from SCC design work, but which omits a large amount of stuff which simply isn't needed in PS3.

But hey, if you want to believe it's the SCC from the information you have, I get the feeling there's nothing I can say that would change your mind.

Cheers,
Dean

 

[Truespeed]

You hear 'credible and accurate', while I look at the figures and think 'way off the mark'. Sure, they can pull figures out of any available orifice when trying to price standard parts like Wi-Fi/Bluetooth modules, but with a closed IC that's not (to my knowledge) been stripped down and analysed, they're just making prices up off the top of their heads.

Well, considering iSuppli has already identified Toshiba as the supplier of the chip I think they may know what they're talking about. The process of part identification and part cost is sort of their business. As for pulling numbers out of their orifice, well, if their previous BOM reports are any indication of their accuracy then their orifices are very well lubricated.

You're right.. it's not definitive proof at all. There's no proof it's a $59 IC, and - even with your leaps of faith - there's still no proof that it's the Toshiba SCC you so desperately want it to be.

My personal view is that it's more likely to be a custom part for PS3 (developed by Toshiba) which draws some elements from SCC design work, but which omits a large amount of stuff which simply isn't needed in PS3.

But hey, if you want to believe it's the SCC from the information you have, I get the feeling there's nothing I can say that would change your mind.

Well, I think we've made some progress. It went from a Cell BE companion chip with no provenance to:

- An IC developed by Toshiba
- An IC that may draw some elements from the SCC

I do believe the iSuppli cost of $59 USD for this IC is within reason and one has to wonder why Sony would pay this much. With Sony scrambling to reduce the component costs of the PS3 as quickly as possible, why would they continue to spend $59 for this chip? Unless, of course, it was something more than a simple I/O bridge controller. If the current evolution of the PS3 firmware is any indication of the feature set catching up to the hardware capabilities then I think we'll see more and more of the functionality of this secret chip uncloaked in future updates.

As IP TV becomes more prevalent I think we'll start to see the shroud slowly pulled off this chip.

 

[Shifty Geezer]

It's worth arguing this point from the flip-side. What is the minimum functional the IO chip would need to do, and what's the minimal size and price of such a chip? If the thing on PS3's mobo is larger (and pricier) than that, what is it doing extra? I don't know zip about IO controller chips and maybe they're all this big, in which case it's not a big deal. But I'm not the only one looking at that big package and thinking there's substantial silicon in there, which leads me to ponder what all that silicon is responsible for.

 

[DeanA]

As IP TV becomes more prevalent I think we'll start to see the shroud slowly pulled off this chip.

My own thoughts are that there's going to be no additional functionality specific to this chip being exposed, simply because such functionality doesn't exist. The multitude of AV inputs to SCC shown on page 10 of the SCC overview PDF from HotChips simply have no equivalent on PS3, while the audio output is routed directly through RSX on PS3. If a companion chip was going to have any dealing with that, then audio would not be passed from XDR through RSX.. it'd be sent straight out to the companion chip via the 5GB FlexIO connection. I won't even go into the scalers that are mentioned on page 13 - I think everyone's more than familiar with PS3's scaling capabilities already, thanks to Farid's article from back in January.

Anyway, I'm sure time will show one of us to be right...

All the best,
Dean

 

[Crossbar]

- I think everyone's more than familiar with PS3's scaling capabilities already, thanks to Farid's article from back in January.

I thought that excellent article only concerned developer controlled scaling of PS3 games. Not that I find the scaling likely to be done by the northbridge, which I already said.

I agree with Shifty, what magic does the northbridge chip with a size just slightly smaller than Cell contain then? If it´s not the SCC it must be full of Sony custom specific logic. Some USB, ethernet and similar logic does not need a 200 m2 chip.

What does it mean that the chip was shrunk in the PAL version, was this the first chip to be moved to 65 nm?

 

[Shifty Geezer]

What does it mean that the chip was shrunk in the PAL version, was this the first chip to be moved to 65 nm?

One other theory is that the chip that features in the original PS3 was a stop-gap, thrown in oversized processor with most of it's functions ignored, and the new chip isn't a 65nm varient but a cut-down processor that's lost all the unused transistors from the original chip. In such a case the ratio of shrinkage would just be coincidental. That begs the question why such a stop-gap chip was used though.