PS3 internals

[Rolf N]

Definately not. It only works in the 60Hz range.Did you look at the image in the post you were quoting?
Did you notice some forum members here use Japanese PS3s in European households without step-down converters or anything similar?

It is a universal PSU, and it definitely works just fine with 50Hz AC.

[Arwin]

*See this (http://www-03.ibm.com/industries/media/doc/content/bin/RapidMind_Raytracer_Demo.pdf). There's also a video floating around somewhere.

Neat!

And yes, the PSU is definitely universal.

[jayco]

I know this is stupid, but in the "Higher Definition Video" on playb3yond, now says that RSX can do 2 trillion operations per second, the thing is, that they had changed it, because in the same video, it said that RSX could do 100 billion operations per second.

Before (http://www.youtube.com/watch?v=sknTNbM8vTo)

After (http://www.playb3yond.com)

I mean, the diference between 100 billion and 2 trillion (200 Billion, am i right?) is HUGE.

[london-boy]

I know this is stupid, but in the "Higher Definition Video" on playb3yond, now says that RSX can do 2 trillion operations per second, the thing is, that they had changed it, because in the same video, it said that RSX could do 100 billion operations per second.

Before (http://www.youtube.com/watch?v=sknTNbM8vTo)

After (http://www.playb3yond.com)

I mean, the diference between 100 billion and 2 trillion (200 Billion, am i right?) is HUGE.

Uhm as figures, they're both as useless as each other... So i wouldn't worry too much about it...

[Shifty Geezer]

I mean, the diference between 100 billion and 2 trillion (200 Billion, am i right?) is HUGE.2 trillion is 20x as large as 100 billion, or 1900 billion more.

[mrboo]

I know this is stupid, but in the "Higher Definition Video" on playb3yond, now says that RSX can do 2 trillion operations per second, the thing is, that they had changed it, because in the same video, it said that RSX could do 100 billion operations per second.

Before (http://www.youtube.com/watch?v=sknTNbM8vTo)

After (http://www.playb3yond.com)

I mean, the diference between 100 billion and 2 trillion (200 Billion, am i right?) is HUGE.

In one video there refering to only RSX ( The 100 Billion comment, maybe talking shader ops? ), while in the other video there talking about PS3( The 2 trillion comment ) as a whole system.

[jayco]

In one video there refering to only RSX ( The 100 Billion comment, maybe talking shader ops? ), while in the other video there talking about PS3( The 2 trillion comment ) as a whole system.


Oh, i missed that.

[Rolf N]

2 Trillion what?
You'd have to count a lot of things you really shouldn't be counting to reach that kind of number. Not relevant, just PR (see sig). It's simply a more memorable form of "it's really fast!".

[bobthebub]

Sorry for digging up an old thread but reading some comments in the HS thread about PS3 scaling prompted me to google to see if any more info had been discovered about that big IO chip (CXD2973GB) in the PS3 that I have always reckoned to be a Toshiba Super Companion Chip (http://www.hotchips.org/archives/hc17/2_Mon/HC17.S1/HC17.S1T3.pdf) (based on size, isuppli cost estimates and wishful thinking).

A quick dig came up with the following email from a Linux mailing list which I reckon to be fairly close to a confirmation that there is indeed an SCC in there from some guy called Geoff Levand that appears to work for Sony.

usb: fix ohci-hcd quirk in cell south bridge (http://lists-archives.org/linux-usb-devel/14872-usb-fix-ohci-hcd-quirk-in-cell-south-bridge.html)

There are currently three platforms that use this chip, the IBM
Cell Blade, the Toshiba Cell Reference Set (Celleb), and the
Sony PS3. The chip itself is on a system bus (IOIF), but each
platform represents it somewhat differently.


From reading some of the responses to that email it looks like the Hypervisor adds a degree of complication to programming for the SCC so I wondered if this could explain the delay in getting the functionality out the door (if it's there that is of course)

[one]

A quick dig came up with the following email from a Linux mailing list which I reckon to be fairly close to a confirmation that there is indeed an SCC in there from some guy called Geoff Levand that appears to work for Sony.

usb: fix ohci-hcd quirk in cell south bridge (http://lists-archives.org/linux-usb-devel/14872-usb-fix-ohci-hcd-quirk-in-cell-south-bridge.html)Great find! I'd though it's a bit overkill for the PS3, but if PS3 has it it may be a subset of it.

[bobthebub]

Great find! I'd though it's a bit overkill for the PS3, but if PS3 has it it may be a subset of it.


It would seem a bit of overkill but it may have been designed in before nVidia came on board and they would have had enough problems with bolting that on without getting Toshiba to design a new southbridge at the same time, perhaps Sony agreed to buy this (large) chip off Tosh as part of the deal when the cell visualiser thingy fell through.

If it is indeed an SCC under that heatspreader it does make the PS3 seem a bit of a FrankenConsole, three great components that were never really designed from the ground up to work as a system cobbled together with a mix of memory and buses then a complete PS2 tacked on the side!

[Nesh]

Whats an SCC?

[jonabbey]

Whats an SCC?

It's the Toshiba SuperCompanionChip, a hot southbridge/IO chip for use with Cell.

See 'Cell companion chip gets hot demo' (http://articles.techrepublic.com.com/2100-10877_11-5833453.html) for some nice words about it.

[temesgen]

Reading thru the Toshiba white paper it appears that among other things the SCC can do image processing... could this chip be intended to do the scaling for PS3? This chip has to be very expensive I cant understand for the life of me why Sony would include a chip like this and not something as simple a as a 5 dollare scaler... Unless they plan to use it to do i/p processing prior to sending info to the AV or HDMI port

[nonamer]

http://www.hotchips.org/archives/hc17/2_Mon/HC17.S1/HC17.S1T3.pdf

At 12.71mm by 12.71mm, that's not a cheap chip.

[Crossbar]

http://www.hotchips.org/archives/hc17/2_Mon/HC17.S1/HC17.S1T3.pdf

At 12.71mm by 12.71mm, that's not a cheap chip.

I really cannot see the logic behind including an SCC chip as is. There is so much redundant functionality provided with regard to the needs of the PS3, and besides it only provides an ATA HDD interface, the PS3 uses a SATA HDD interface.
I do not rule out a heavily modified SCC though.

[-tkf-]

I really cannot see the logic behind including an SCC chip as is. There is so much redundant functionality provided with regard to the needs of the PS3, and besides it only provides an ATA HDD interface, the PS3 uses a SATA HDD interface.
I do not rule out a heavily modified SCC though.

Modifying something like this chip takes time?

I wonder if the reason why Sony isn´t using any of the exotic features in this chip is simply because they play to exchange it for something else? If not i hope they get it "online" and working...

[Kryton]

I really cannot see the logic behind including an SCC chip as is. There is so much redundant functionality provided with regard to the needs of the PS3, and besides it only provides an ATA HDD interface, the PS3 uses a SATA HDD interface.
I do not rule out a heavily modified SCC though.

And the USB, and the Gb Ethernet, ... also has a scaler

[bobthebub]

Even if there is an SCC in there (anyone fancy peeling the heatspreader off thiers?) would the scaler be usable, the SCC could do HDMI out but is it actually connected to the HDMI port and could it read the framebuffer from RSX fast enough (it would have to go through cell and RSX to get to the GDDR).

Is there any way the video out from RSX could be directly connected to this mystery southbridge because it doesn't look like it from the M/B pics above.

And the USB, and the Gb Ethernet, ... also has a scaler

The Gb ethernet on PS3 also seems to be at least based on the SCC port as some googling also found evidence the PS3 linux ethernet driver "gelic" is a copy of the one for the SCC (spider_net).


I really cannot see the logic behind including an SCC chip as is. There is so much redundant functionality provided with regard to the needs of the PS3, and besides it only provides an ATA HDD interface, the PS3 uses a SATA HDD interface.
I do not rule out a heavily modified SCC though.

The fact that there is this big chip there 'and' an outboard SATA controller suggests to me that Sony did not get a redesigned southbridge for PS3 as surely they would have integrated SATA during the redesign.

[Shifty Geezer]

Even if there is an SCC in there (anyone fancy peeling the heatspreader off thiers?) would the scaler be usable, the SCC could do HDMI out HDMI is by a SiI chip.

[bobthebub]

HDMI is by a SiI chip.

I thought that was an HDCP transmitter but I am not sure what difference it makes, the SCC can output a DVI style signal, if that needs to go through an HDCP/HDMI chip it doesn't really alter whether the SCC can ouput video without going back through RSX.

However If we know the Sil chip is not connected at all to the Cell South bridge then that would be a clue.

http://www.hotchips.org/archives/hc17/2_Mon/HC17.S1/HC17.S1T3.pdf

At 12.71mm by 12.71mm, that's not a cheap chip.

Found the following - looks like someone has measured this chip:-

PS3 VS Wii, Comparisons of Core LSI Chip Areas (http://techon.nikkeibp.co.jp/english/NEWS_EN/20061127/124495/)

Nikkei Electronics has measured chip areas of microprocessors and graphics LSIs mounted on the main boards of the "PlayStation 3" (PS3) and "Wii." As we removed heat spreaders stuck on the upper surface of respective product packages

The chip, which works as a south bridge in the PS3, measured 13 x 13 mm

[nonamer]

So, how close are we to confirming the SCC is in the PS3 now?

[Carl B]

Just two days ago I was wondering about this same thing. Wow, SCC inclusion would certainly be something. The first question of course is "why?"

RSX was put into motion in 2003, so I certainly don't think there's accidental feature overlap where it could otherwise have been hedged. It could just be that Toshiba didn't design anything further custom for the PS3 beyond what they are doing for Cell reference designs in general, and there's business/politics going on behind-the-scenes to help Toshiba out. Not to mention, perhaps as with Cell in general, part of the ecosystem drive as a whole involves a mass fabbing effort to reduce the cost of the SCC, and part of that involves putting it in the PS3.

Anyway this is definitely an interesting topic to watch develop.

[Shifty Geezer]

Which chip are people talking about? The one labelled as a Sony controller chip? If it's the Toshiba SCC chip, why isn't it labelled as such? Customized varient? The design and size of this package seems to match the Toshiba SCC, and would be an additional substantial cost for PS3, explained in PS3's high price.

As for inclusion, the mind boggles! It's an IO feed. It'd be ideal to supply a dozen TV streams to PS3 for digital recording...but there's no input. Are we looking at intended PVR functionality via a USB addon? What exactly does the SCC do?! It seems a massive* bit of silicon for just redirecting signals, yet surely it doesn't handle scaling or decoding as that's what Cell's supposed to do.

* Edit : Actually for massive, I can't find any transistor counts for the SCC. Could be a big lot o' pins for what'sa pretty tiddly piece of silicon inside.

[Shifty Geezer]

...and there's business/politics going on behind-the-scenes to help Toshiba out. Not to mention, perhaps as with Cell in general, part of the ecosystem drive as a whole involves a mass fabbing effort to reduce the cost of the SCC, and part of that involves putting it in the PS3.I think that's kinda reaching.

Toshiba : "What about us?"
Sony : "Well, what do you want in PS3?"
Toshiba : "We've got a great SCC."
Sony : "How will that help PS3?"
Toshiba : "It won't. But we want these things being massed produced so we can get costs down to sell it to other devices."
Sony : "Ah, okay then. In it goes. Has any other company got some new chips to add to our console that won't do anything useful, but will help the chip to get established...? [/open invite]"

[bobthebub]

What exactly does the SCC do?! .

The Toshiba hotchips white paper that has been linked a few times here gives a really good overview of the capabilities of the SCC http://www.hotchips.org/archives/hc17/2_Mon/HC17.S1/HC17.S1T3.pdf

edit: You also asked about transistors shifty, it's a 90nm part around 160mm2, almost the size of Xenon which is 165Million transistors, I guess the transistor density could be quite different but would it not cost a vaguely similar amount to churn out if it's a similar size.

[nonamer]

HDMI is by a SiI chip.

Are we certain of this? The SCC supports encryption internally, thus removing the need for any HDMI chip.

EDIT: It seems to support HDMI directly. From here on page 13: http://www.hotchips.org/archives/hc17/2_Mon/HC17.S1/HC17.S1T3.pdf

[Carl B]

I think that's kinda reaching.

Well, reaching or not, the chip seems to be in there. And even though there's nothing rock-solid indicating it's the SCC shown at hotchips, at 13mmx13mm, it has to be something at least very similar. The Gigabit Ethernet driver supports this theory also. And when you look at what the SCC *is*, there are indeed - in any circumstance - some unused functions. Prominent in this list would be the memory controller intermediary for the use of DDR2 with Cell. That just simply has nothing to do with anything as far as PS3 is concerned. Although I never expect that mentioned feature to come into play, I do expect either some wacky future unlocking of potential on PS3 via the SCC, or a much smaller revision of this SB chip to come in on future PS3s. The chip presently in there, for whatever reasons it is in there, seems less than the ideal from a cost/functionality perspective. For now I have to stick with the 'Cell ecosystem' theory, although I'm sure a more focused revision of the chip is in the works.

[bobthebub]

Prominent in this list would be the memory controller intermediary for the use of DDR2 with Cell. That just simply has nothing to do with anything as far as PS3 is concerned.

My reading of the hotchips paper indicates that the DDR2 is required as some kind of local buffer for the scaling type functionality, if there isn't *any* on the PS3 motherboard it may not be able to scale anyway.

[Carl B]

Yeah, reviewing the doc I do see that also.

Well, I guess a good question would be if on the existing pics of the PS3 motherboard layout, DDR memory modules are visible in close proximity to the SB chip.

EDIT: Luckily, we're in the right thread to get those questions answered. Here's the link to the motherboard layout:

http://pc.watch.impress.co.jp/docs/2006/1111/ps3_32.jpg

Indeed, it seems as if the SB does have an associated outboard DDR2 memory module associated with it; and then next to that, a Samsung module that if I didn't know better I would say looks like traditional DDR.

EDIT 2: The chip in between the Samsung module and the SCC must just be something else, because the Samsung chip definitely seems to be RAM from the text indicators on the chip itself.

[nonamer]

Could be flash RAM, not DDR2 RAM.

And it would be strange to need memory to do scaling. Existing scaling chips don't use that at all.

[Carl B]

And it would be strange to need memory to do scaling. Existing scaling chips don't use that at all.

Well, those were my original thoughts as well - that scaling (if done on this chip) would/could be handled as a straight real-time dynamic stream - but nonetheless a memory buffer does seem to be an integral aspect of the SCC design for some reason. Maybe there's a video 'input' aspect to this console yet... use of the SCC would make more sense in that instance.

[Npl]

Am I right that the chip can decode Digital TV? What would be actually needed to enable this functionality except a physical connector?

I mean being able to record TV with the PS3 would be a big plus.. from a "Media Center" POV atleast.

also I find 4 USB Ports a bit lacking. doesnt the PS3 have 4x USB front and 2 back ?

[Mmmkay]

EDIT: Luckily, we're in the right thread to get those questions answered. Here's the link to the motherboard layout:

http://pc.watch.impress.co.jp/docs/2006/1111/ps3_32.jpg

Indeed, it seems as if the SB does have an associated outboard DDR2 memory module associated with it; and then next to that, a Samsung module that if I didn't know better I would say looks like traditional DDR.

EDIT 2: The chip in between the Samsung module and the SCC must just be something else, because the Samsung chip definitely seems to be RAM from the text indicators on the chip itself.

That memory is 128MB of NAND Flash RAM.

http://images.dailytech.com/nimage/2999_large_marvell.jpg
K9F1G08U0A

[Carl B]

Great cross-referencing Mmmkay. I think it's interesting to note that the Japanese build has a Samsung 625 and the American a Samsung 622. And I guess the adjacent module is still on the table then for being DDR2, although I wish we could read some of the text on it.

[bobthebub]

. And I guess the adjacent module is still on the table then for being DDR2, although I wish we could read some of the text on it.

It's an SCEI cxd4302gb (http://images.dailytech.com/nimage/2998_large_io_bridge.jpg), Chipworks couldn't work out what it was.

Check this out though :- Chipworks PS3 basic product teardown (http://www.chipworks.com/uploadedFiles/Technical_Competitive_Analysis/Capabilities/Sony_Playstation_report%20(2).pdf)

Fairly comprehensive list of the main components

[Mmmkay]

Great cross-referencing Mmmkay. I think it's interesting to note that the Japanese build has a Samsung 625 and the American a Samsung 622. And I guess the adjacent module is still on the table then for being DDR2, although I wish we could read some of the text on it.
No probs.

Unfortunately the other chip is one of those mystery components.

Chipworks identified the component as CXD4302GB
http://www.chipworks.com/uploadedFiles/Technical_Competitive_Analysis/Capabilities/Sony_Playstation_report%20(2).pdf

[Carl B]

Thank you for that link, it's awesome!

Yeah RAM wouldn't bare the Sony label though, so indeed another mystery chip/controller for the pile.

[Crossbar]

Well, those were my original thoughts as well - that scaling (if done on this chip) would/could be handled as a straight real-time dynamic stream - but nonetheless a memory buffer does seem to be an integral aspect of the SCC design for some reason. Maybe there's a video 'input' aspect to this console yet... use of the SCC would make more sense in that instance.

It would seem a little odd if this mysterious SCC chip or whatever, would handle the scaling of output from the RSX as it is apparent from the PCB that there is a direct connection between the RSX and the Silicon Image chip. Maybe there is more to this but it seems a little bit odd.

[nonamer]

It would seem a little odd if this mysterious SCC chip or whatever, would handle the scaling of output from the RSX as it is apparent from the PCB that there is a direct connection between the RSX and the Silicon Image chip. Maybe there is more to this but it seems a little bit odd.

http://img.hexus.net/v2/gaming/screenshots/blockdiagram.jpg

If this is accurate, then video output is directly connected to the RSX, with the "SCC" being merely a southbridge. That's a rather wasteful use though for such a large chip, unless there is someway way to send video data back to the Cell. I can't help but think that they were forced to slap this thing together without fully integrating the parts.

[bobthebub]

I can't but think that were forced to slap this thing together without fully integrating the parts.

I think you are probably right (Hence my frankenconsole comment yesterday), if this chip is an SCC it could completely sum up the whole history of the PS3.

They start out with an elegant all integrated design , XDR, flexio, lots of AV capability, scaling functions etc, then they have to go to Nvidia for RSX, suddenly the scaler can't see the framebuffer, they are left with a big semi-useless southbridge and no time/money/co-operation from Toshiba to replace the southbrige or create a dedicated scaler.

PS3 ends up late, expensive and lacking in some key functionality.

I am genuinely wondering whether it might be wise to wait for a 2nd gen machine to see if some of this gets sorted.

Hope that's all wrong though and they are ironing out the kinks on the scaler as we speak, ready for the euro launch.

[Carl B]

I think the problem is in viewing the SCC as soley a scaler solution. If its functions include video input, then it makes more sense. If they don't, well we'll just see what the deal is as time goes on... either way the full story eludes us as yet, and there are obviously a lot of plans Sony has for the PS3 beyond what's been revealed so far.

[bobthebub]

I don't think we will see video input into the PS3 as anything other than digital data which can come via Gig ethernet or USB, once it's onboard even if SCC could do the decoding cell can as well so it's just duplication of functionality.

Sony may have had (or may still have) intentions of creating a PSX style PS3 media box where the SCC could come into it's own and there would be obvious benefits in having a common platfom for the two but at the moment it looks like it could be a waste of silicon as the output will still likely be via RSX so a good chunk of the SCC would never be utilised in a PS3 based box.

... either way the full story eludes us as yet, and there are obviously a lot of plans Sony has for the PS3 beyond what's been revealed so far.

Agreed.

[tuna]

If it is for video input, maybe it could be for Eyetoy?

[one]

also I find 4 USB Ports a bit lacking. doesnt the PS3 have 4x USB front and 2 back ?That's the E3 2005 spec. The current SKU has 4.

[Rolf N]

Wow, that's an ugly hack.If it is for video input, maybe it could be for Eyetoy?It's not necessary for Eyetoy software. The PS2 can do Eyetoy games without special hardware in the system, so PS3 Eyetoy software should be perfectly fine with using an SPE or two.

For purposes of the PS3 this chip is a kludgy interim solution -- it's too big and too expensive to stay in. The PS3 needs a single SATA port, video outs, Networking, USB, a scaler and some LPC interface to read firmware and stuff. That SCC plus extra NIC plus extra TMDS plus extra whatever is expensive overkill and needs to be simplified asap.

[one]

Prominent in this list would be the memory controller intermediary for the use of DDR2 with Cell. That just simply has nothing to do with anything as far as PS3 is concerned.I think that RAM interface is for workstations, such as the Cell blade. IIRC the Mercury Cell hardware for military use had a big RAM attached to it and if it's based on the IBM board I think it's likely that it has SCC in it.
I don't think we will see video input into the PS3 as anything other than digital data which can come via Gig ethernet or USB, once it's onboard even if SCC could do the decoding cell can as well so it's just duplication of functionality.

Sony may have had (or may still have) intentions of creating a PSX style PS3 media box where the SCC could come into it's own and there would be obvious benefits in having a common platfom for the two but at the moment it looks like it could be a waste of silicon as the output will still likely be via RSX so a good chunk of the SCC would never be utilised in a PS3 based box.Interesting, but as far as I read the Hot Chips paper for SCC, I don't think it has a scaler function. Probably you saw the page 13 with "Scaling" in the processing steps, but in the page 14 and 15, the actual video processing seems to occur in Cell and SCC only does its job just like a normal southbridge but with higher bandwidth. In the Toshiba 48 MPEG2 streams decoding demo it was Cell that did scaling. Can anybody point me to where it explicitly says SCC has a scaling function?

[Nesh]

I think that's kinda reaching.

Toshiba : "What about us?"
Sony : "Well, what do you want in PS3?"
Toshiba : "We've got a great SCC."
Sony : "How will that help PS3?"
Toshiba : "It won't. But we want these things being massed produced so we can get costs down to sell it to other devices."
Sony : "Ah, okay then. In it goes. Has any other company got some new chips to add to our console that won't do anything useful, but will help the chip to get established...? [/open invite]"

Thats not very logical though. Someone has to compensate the extra cost of the chip existing in the machine unused even if the marginal cost is lower. Sony is either increasing the cost of the PS3 further increasing their losses with each machine sold, or Toshiba is losing money with each cheapper to produce chip being thrown unused in the PS3.

Toshiba could have just as well produced the chip in more quantities than needed just to reduce the marginal cost and throw the extra chips in the dump. Thats definetely still a cost without a revenue unless Sony pays them on top for something that is useless. Its no different than implementing in a PS3 and compensate the cost with zero benefit.

The only way I see this actually being beneficial for them is a very weird behavior of production cost, with total cost being reduced more than the cost of the extra useless/unused chip produced. But that sounds impossible to me.

There should definetely be some use in it, atleast some inefficient use due to some bad planning and designing the architecture. Otherwise all manufacturers would have increased supply more than demanded to reduce costs

[Fox5]

Thats not very logical though. Someone has to compensate the extra cost of the chip existing in the machine unused even if the marginal cost is lower. Sony is either increasing the cost of the PS3 further increasing their losses with each machine sold, or Toshiba is losing money with each cheapper to produce chip being thrown unused in the PS3.

Toshiba could have just as well produced the chip in more quantities than needed just to reduce the marginal cost and throw the extra chips in the dump. Thats definetely still a cost without a revenue unless Sony pays them on top for something that is useless. Its no different than implementing in a PS3 and compensate the cost with zero benefit.

The only way I see this actually being beneficial for them is a very weird behavior of production cost, with total cost being reduced more than the cost of the extra useless/unused chip produced. But that sounds impossible to me.

There should definetely be some use in it, atleast some inefficient use due to some bad planning and designing the architecture. Otherwise all manufacturers would have increased supply more than demanded to reduce costs

Sony could pay them at or near cost for the chip, thus Toshiba gets to benefit from increased production without suffering losses.
Or Sony could just have acted like the government and given Toshiba a lucrative contract as a type of pay off.

[Nesh]

Sony could pay them at or near cost for the chip, thus Toshiba gets to benefit from increased production without suffering losses.
Or Sony could just have acted like the government and given Toshiba a lucrative contract as a type of pay off.

But isnt that stupid? Paying Toshiba for something useless just to put it in their product?

[Fox5]

But isnt that stupid? Paying Toshiba for something useless just to put it in their product?

If Toshiba is giving them a sweet deal on it (at or near cost) it may be cheaper or the same price as an equivalent off the shelf solution.

And if Sony is losing out on the deal, yes it would be stupid, but that's why I said like a government. Businesses can do stupid things too just to honor contracts.

[inefficient]

If Toshiba is giving them a sweet deal on it (at or near cost) it may be cheaper or the same price as an equivalent off the shelf solution.

And if Sony is losing out on the deal, yes it would be stupid, but that's why I said like a government. Businesses can do stupid things too just to honor contracts.


If it was an off the shelf solution from Toshiba, why is the chip in question labled as a Sony chip (SCEI cxd4302gb) and not Toshiba?

There is no chip labed "Toshiba" in the PS3 itself. The only Toshiba chip is the sixaxis controler.

If the chip labled SCEI cxd4302gb really is based on the SCC then in all likely hood Sony bought the designs and is fabbing it.

The question is, why would Sony want to fab a chip that big that has features they would never use in PS3? Two answers I can think of is: They didn't have enough time to complete their stripped down version before launch but will have a smaller version in the future. Or they want to use the component in their other systems as well so they want to drive costs down by mass producing it.

[Carl B]

The chip could be fabbed at Sony/Toshiba's joint fab and simply be packaged differently, while still being a Toshiba design. For example, none of the Cell chips bare IBM anywhere on them, yet we know that IBM is the only source of these at the moment.

I think knowing that the SouthBridge likely is an SCC (or variant) - simply due to the die size similarities - the question then goes to: "why?" I think there may be a couple of different angles in that regard.

[Shifty Geezer]

From that HotChips, presumably this SCC is running all the I/O - USB, Gigabit, etc. That makes it a necessary component, although a big one. It's bound to scale in price much better than lots of smaller chips, and is maybe more reliable? How else would you get a dozen different IO devices attached to the FlexIO?

[Jim_Fear]

The chip could be fabbed at Sony/Toshiba's joint fab and simply be packaged differently, while still being a Toshiba design. For example, none of the Cell chips bare IBM anywhere on them, yet we know that IBM is the only source of these at the moment.

I think knowing that the SouthBridge likely is an SCC (or variant) - simply due to the die size similarities - the question then goes to: "why?" I think there may be a couple of different angles in that regard.

I thought Cell was also being fabbed in Sony's Nagasaki plant?

[Crossbar]

From that HotChips, presumably this SCC is running all the I/O - USB, Gigabit, etc. That makes it a necessary component, although a big one. It's bound to scale in price much better than lots of smaller chips, and is maybe more reliable? How else would you get a dozen different IO devices attached to the FlexIO?

While we are talking about attaching devices to the FlexIO, how is the EE+GS chip attached?
Could this chip(SCC?) be emulating the interface of the southbridge of the PSTwo as well or have anyone seen any description of how it is done?

If you look at this semi-professional document (http://cell-industries.com/SCC_Details1.pdf), there is an overview of the internal bus structure of the SCC, it seems pretty flexible.
That document claims that the SCC is used in the PS3, I remember when the document was released some year ago that it was dismissed as nonsense, but maybe there is some truth in it after all?

Edit: I found this nice picture over here (http://www.semicon.toshiba.co.jp/product/micro/cell/pdf/toshiba_cell_reference_set_200604.pdf):
http://img250.imageshack.us/img250/3423/scc1xm.jpg

[inefficient]

Unused:
3x TSIF Digital Video In
Analog Video In
Analog Audio In
3x Video Out (HDMI,D4,A-DAC)
IEEE1394 Firewire
1x P-ATA - Can't use, thats why there is a separate S-ATA controler

Used:
USB2.0
Analog Audio Out
SPDIF Audio Out
1x P-ATA - For the Bluray drive
G-Ethernet

It just seems wrong...

[Arun]

NVIDIA => NRE2 (http://www.beyond3d.com/forum/showthread.php?t=29406) => Southbridge.
Hmm! That'd make more sense than the PSP2 for that specific block of NRE, given the timeframes.


Uttar

[Carl B]

I thought Cell was also being fabbed in Sony's Nagasaki plant?

Well, but not just yet. Cell will start being fabbed there on the 65nm node; the line should honestly be just about ready now though.

[Crossbar]

NVIDIA => NRE2 (http://www.beyond3d.com/forum/showthread.php?t=29406) => Southbridge.
Hmm! That'd make more sense than the PSP2 for that specific block of NRE, given the timeframes.
Good catch! Could be.

[bobthebub]

While we are talking about attaching devices to the FlexIO, how is the EE+GS chip attached?
Could this chip(SCC?) be emulating the interface of the southbridge of the PSTwo as well or have anyone seen any description of how it is done?


The PS3 VS Wii, Comparisons of Core LSI Chip Areas (http://techon.nikkeibp.co.jp/english/NEWS_EN/20061127/124495/)article that measured the PS3 southbridge covered that topic a little as well.

The "EE + GS" chip loaded to boost the PS3's compatibility with the PS2 is connected with this south bridge via another bridge LSI. In short, the "IOP" input/output processor LSI used in the PS2 has been replaced with these bridge LSI, south bridge and Cell chips.

[Nesh]

If Toshiba is giving them a sweet deal on it (at or near cost) it may be cheaper or the same price as an equivalent off the shelf solution.

And if Sony is losing out on the deal, yes it would be stupid, but that's why I said like a government. Businesses can do stupid things too just to honor contracts.

Government is a business operating at a loss without risks

[-tkf-]

Unused:
3x TSIF Digital Video In
Analog Video In
Analog Audio In
3x Video Out (HDMI,D4,A-DAC)
IEEE1394 Firewire
1x P-ATA - Can't use, thats why there is a separate S-ATA controler

Used:
USB2.0
Analog Audio Out
SPDIF Audio Out
1x P-ATA - For the Bluray drive
G-Ethernet

It just seems wrong...

It makes sense in a Cell way. SCC is the perfect Cell partner, yes it can do more than it does in the PS3, but i wouldn´t be surprised if the costs for making a smaller chip with less functions AND the Cell connectetivity wouldn´t be all that different considering that the SCC is already developed and ready to roll. It´s gonna take some time for the manufactoring price to catch up on the price to develop such a chip. Just a though of course...

[Crossbar]

It makes sense in a Cell way. SCC is the perfect Cell partner, yes it can do more than it does in the PS3, but i wouldn´t be surprised if the costs for making a smaller chip with less functions AND the Cell connectetivity wouldn´t be all that different considering that the SCC is already developed and ready to roll. It´s gonna take some time for the manufactoring price to catch up on the price to develop such a chip. Just a though of course...

Very possible, from the die size, the external SATA-controller and all other evidence, it seems very likely it is indeed the SCC chip in there. Another price related possibility is that since the PS3 is hardly using about half of the functions in the SCC, mind you I don´t know the die area of the corresponding logic, Sony may be taking care of malfunctioning chips from Toshiba, where some function not used by the PS3 is broken. That would help yield and price as well.

Something that strikes me when looking at the part list of the board is that there are so many major chips that have not been tailored/optimised for the PS3, such as the

The SCC chip
The Silicon Image HDMI Transmitter
The Marvell SATA Controller
The Complete PSTwo chipset (+ RDRAM memory)


The benefit of using all these more or less standard components is that you get some reliable and proven circuitry in there, and you may get them faster than if you were to design them ground up, the drawback is that they are pretty expensive, which result in a pretty expensive console (another popular topic:smile: ).

One conclusion from this could be that Sony do not plan to manufacture any high volumes of this console according to this spec. (not a surprise I know). Because when you go over a certain number it would have been worth the effort to design some more stream-lined custom circuits. Then you could argue that this is a hack job because Sony was under stress and they needed to get something out of the door, so they took a short track and used some standard circuitry to reduce the risk.

Maybe there is some truth in that considering the risk part, but this console has been under design for about 6 year so I think that if they wanted to have a custom made southbridge in the PS3 right from start they would have done that. I think Sony decided to concentrate some major engineering job into the first cost-reduced version instead of putting a large effort into the first PS3 version which has a limited life cycle anyhow. Mind you, the effort(cost) may have been large enough already through the design of Cell and RSX.

I am predicting some major redesign coming this fall together with the die shrinked Cell and RSX.
The PSTwo chip will diappear possibly some parts of the GS will be integrated together with a SATA controller in a new custom designed 65 nm South-Bridge. Possibly the Silicon Image chip as well, but I don´t know if there are some particular requirement of that I/O circuitry that would make it harder.
Anyway, once they´ve done that, I think Sony will ramp up the PS3 pretty aggressively and be more price competetive.

So I think Sony has the current price point just because they can, and at the same time still sell millions of units (how many remains to be seen), the current design is only intended to establish a foothold on the market and to be on that market a relatively short time and in relatively low volumes. Sony is prepared to take the piss for having an expensive console at start and save some major design efforts for the cost-reduced and high volume version coming later on, together with some attractive game titles.

Time will tell. :)

[Capeta]

Something that strikes me when looking at the part list of the board is that there are so many major chips that have not been tailored/optimised for the PS3, such as the

The SCC chip
The Silicon Image HDMI Transmitter
The Marvell SATA Controller
The Complete PSTwo chipset (+ RDRAM memory)


The benefit of using all these more or less standard components is that you get some reliable and proven circuitry in there, and you may get them faster than if you were to design them ground up, the drawback is that they are pretty expensive, which result in a pretty expensive console (another popular topic:smile: ).

One conclusion from this could be that Sony do not plan to manufacture any high volumes of this console according to this spec. (not a surprise I know). Because when you go over a certain number it would have been worth the effort to design some more stream-lined custom circuits. Then you could argue that this is a hack job because Sony was under stress and they needed to get something out of the door, so they took a short track and used some standard circuitry to reduce the risk.

Maybe there is some truth in that considering the risk part, but this console has been under design for about 6 year so I think that if they wanted to have a custom made southbridge in the PS3 right from start they would have done that. I think Sony decided to concentrate some major engineering job into the first cost-reduced version instead of putting a large effort into the first PS3 version which has a limited life cycle anyhow. Mind you, the effort(cost) may have been large enough already through the design of Cell and RSX.

I am predicting some major redesign coming this fall together with the die shrinked Cell and RSX.
The PSTwo chip will diappear possibly some parts of the GS will be integrated together with a SATA controller in a new custom designed 65 nm South-Bridge. Possibly the Silicon Image chip as well, but I don´t know if there are some particular requirement of that I/O circuitry that would make it harder.
Anyway, once they´ve done that, I think Sony will ramp up the PS3 pretty aggressively and be more price competetive.

So I think Sony has the current price point just because they can, and at the same time still sell millions of units (how many remains to be seen), the current design is only intended to establish a foothold on the market and to be on that market a relatively short time and in relatively low volumes. Sony is prepared to take the piss for having an expensive console at start and save some major design efforts for the cost-reduced and high volume version coming later on, together with some attractive game titles.

Time will tell. :)

I don't see the Silicon Image chip disappearing. HDMI has been out for a long time already yet almost EVERYONE buys these chips. I don't see SONY designing their own HDMI chip. Also does SONY even have a SATA chip for sale? I doubt that they do unless they've been selling them to SATA board makers. Silicon Image also makes SATA chips.

[Kryton]

Unused:
3x TSIF Digital Video In
Analog Video In
Analog Audio In
3x Video Out (HDMI,D4,A-DAC)
IEEE1394 Firewire
1x P-ATA - Can't use, thats why there is a separate S-ATA controler

Used:
USB2.0
Analog Audio Out
SPDIF Audio Out
1x P-ATA - For the Bluray drive
G-Ethernet

It just seems wrong...

Why do I get the feeling the PS2 chipset is wired to the analog connections? Could possibly explain 2 scaler fixes at once :S

Edit: Also, haven't Sony hinted at PVR functionality with many people questioning how they could provide it?

Edit2: What is the card reader connected to in the premium SKU, that could use PATA (especially for CF)?

Edit3: Does the SCC support HDMI 1.3 and HDCP? The Silicon Image chip maybe there purely for this.

[Capeta]

PVR functionality will need an external adapter, likely a USB device similar to those for PCs.

[inefficient]

Edit2: What is the card reader connected to in the premium SKU, that could use PATA (especially for CF)?

USB would be the most obvious solution.

[Crossbar]

I don't see the Silicon Image chip disappearing. HDMI has been out for a long time already yet almost EVERYONE buys these chips. I don't see SONY designing their own HDMI chip. Also does SONY even have a SATA chip for sale? I doubt that they do unless they've been selling them to SATA board makers. Silicon Image also makes SATA chips.I made a reservatÃ*on about the Silicon Image chip, but I would be very surprised if you coudnt find some SATA VHDL-ip for sale, I am sure Nvidia could help out in that matter if that was the case.

[Rainbow Man]

I don't understand why sony put in a separate SATA controller when they could just have used one of those parallel to dserial converter things that wrere so common when SATA was brand new. Those were by all reports cheap effective and had no discernible performance impact.

Heck the first raptor harddrive was by far the fastest unit at the tiem and it used one integrated right on the PCB as I recall..

Peace.

[mtkc]

I don't understand why sony put in a separate SATA controller when they could just have used one of those parallel to dserial converter things that wrere so common when SATA was brand new. Those were by all reports cheap effective and had no discernible performance impact.

Heck the first raptor harddrive was by far the fastest unit at the tiem and it used one integrated right on the PCB as I recall..

Peace.

The Marvell SATA chip you see on the PS3 board is PATA to SATA bridge, not a dedicated SATA controller. Code is 88SA8040. http://www.marvell.com/products/storage/sata/SerialATAII_88SM4140.pdf

This lends even more credibility to the I/O chip being the SCC.

[bobthebub]

Just found another teardown (sorry if old, couldn't find it on here) that may be of some interest, a lot of pics and some more chip close-ups, seems to suggest the PS1 chip is still on board which I didn't think would have been necessary.

Techrepublic - Cracking open the PS3 (http://content.techrepublic.com.com/2346-10877_11-41891-1.html)

[Shifty Geezer]

a lot of pics and some more chip close-ups, seems to suggest the PS1 chip is still on boardWhat makes you think that? As the PS1 chip has been replaced in the PS2, I find it incredibly improbably they'd have it in a PS3!

[bobthebub]

One of the pics is labelled as follows:-

System battery connector (white) and PlayStation 1 chip (http://content.techrepublic.com.com/2346-10877_11-41891-113.html)

No idea if they are correct or not though.

[INKster]

One of the pics is labelled as follows:-

System battery connector (white) and PlayStation 1 chip (http://content.techrepublic.com.com/2346-10877_11-41891-113.html)

No idea if they are correct or not though.

"CRX" is usually part of a Sony optical disc drive name, so that chip may have something to do with it, since it starts with "CRX" too.

[Rangers]

There's a pic in there of "1GB of flash memory". Does the PS3 have that much flash? The Xbox360?

Never mind I guess it is 128 Mbyte. Still, didn't know about that.

[Fox5]

There's a pic in there of "1GB of flash memory". Does the PS3 have that much flash? The Xbox360?

It's probably 1Gigabit and contains the OS and stuff.

[Crossbar]

There's a pic in there of "1GB of flash memory". Does the PS3 have that much flash? The Xbox360?

Never mind I guess it is 128 Mbyte. Still, didn't know about that.
The PS3 mainboard contains 256 MByte flash memory in total, here (http://www.chipworks.com/uploadedFiles/Technical_Competitive_Analysis/Capabilities/Sony_Playstation_report%20(2).pdf)is a nice component breakdown.

[Arwin]

The PS3 mainboard contains 256 MByte flash memory in total, here (http://www.chipworks.com/uploadedFiles/Technical_Competitive_Analysis/Capabilities/Sony_Playstation_report%20(2).pdf)is a nice component breakdown.

I'm wondering if it conains a built in backup (it can format an unformatted disc and such) or if that's all just for straightforward OS features. The PSP also has 2 flash banks, if I remember correctly.

[Hezz]

Now that we have heard about "Home". Is it possible that the DDR that seems connected with the SCC is anouther memory space for Home to run without using XDR. Or is it more likely that this extra DDR is used for helping the SCC to emulate PS2 I/O hardware compatibility.

Also, is this DDR only on the developer SDK consoles or is it on all PS3 models.

[Sousuke]

The PS3 mainboard contains 256 MByte flash memory in total, here (http://www.chipworks.com/uploadedFiles/Technical_Competitive_Analysis/Capabilities/Sony_Playstation_report%20(2).pdf)is a nice component breakdown.
from that pdf, there seems to be a 128Gb SRAM on the WIFI board - but I guess thats a typo ;)

Hezz: I think you're a little confused, what are you talking about? :shock:

[Panajev2001a]

from that pdf, there seems to be a 128Gb SRAM on the WIFI board - but I guess thats a typo ;)

Hezz: I think you're a little confused, what are you talking about? :shock:

It's a typo in the Chipworks's PDF, if you look that part up (IS42S32400B) you'll see it is a 128 Mbits SDRAM chip.

[Hezz]

Yeah, reviewing the doc I do see that also.

Well, I guess a good question would be if on the existing pics of the PS3 motherboard layout, DDR memory modules are visible in close proximity to the SB chip.

EDIT: Luckily, we're in the right thread to get those questions answered. Here's the link to the motherboard layout:

http://pc.watch.impress.co.jp/docs/2006/1111/ps3_32.jpg

Indeed, it seems as if the SB does have an associated outboard DDR2 memory module associated with it; and then next to that, a Samsung module that if I didn't know better I would say looks like traditional DDR.

EDIT 2: The chip in between the Samsung module and the SCC must just be something else, because the Samsung chip definitely seems to be RAM from the text indicators on the chip itself.


Sorry guys, this is an idea got from some of the discussion a page or two back in the thread. There seemed to be a DDR chip in close proximity to the SB (southbridge ?). Or the SCC as Sony has been calling it.

I guess since Sony has stated that they want to built a media PC based on the PS3, the overkill of the SCC seems in order as it would allow more DDR2 to be installed which I would assume would be used by Linux or other software than games. Since game hardware specs would need to remain standardized.

This lead me to speculate that perhaps Sony had intended for a small DDR chip to function as a RAM space for Home. But most likely it is just used for some kind of PS2 hardware I/O emulation.