Jul 17

What chips inside CECH-4000?

Our first post about PS3 UltraSlim, four months ago, dealt with the successive shrink steps and eventual integration of the Cell Broadband Engine (the CPU) and the RSX graphics processor (GPU), analoguous to the merging of the Emotion Engine and Graphics Synthesizer into EE+GS which paved the way for PS2 Slim.

In the meantime we’ve learnt about the complexities and diminishing returns of shrinking the two main PS3 chips and other practical obstacles making a single die Cell+RSX unlikely in the near future. One question we’ve asked no one seemed willing to answer is if Cell and RSX on one chip is too cumbersome, would combining them on a SoC (System-on-a-Chip) or in a SiP (System-in-a-Package) be a possible and attractive step forward for Sony Computer Entertainment?

Well, we’ve just got a glimpse of the lines SCE is apparently thinking along: Sony’s new PS Vita, according to Canadian semiconductor technology analysts and advisors Chipworks, contains a true SiP component, consisting of a processor (one with an embedded GPU in case you’re interested) with memory stacked on top, making very efficient use of scarce PCB real estate. Detailed technical analysis here.

Is the new PS3 going to use RAM on top of Cell, VRAM stacked onto the RSX, or a combination of both? Unless SCE is going to announce this at some semiconductor conference such as ISSCC (but the 2012 edition of that has already taken place) we’ll have to wait until CECH-4000 PlayStation 3 gets to market — some time shortly after GamesCom 2012? — and gets a teardown by iSuppli or iFixit, with the silicon X-rayed by researchers of Chipworks,


  1. jeff_rigby

    Ram on top of Cell, not going to happen as RAM is heat sensitive. Handheld is a different animal in many ways.

    The TSV technology that is making handheld SOCs cheaper to make can also be applied to Game consoles but stop thinking of 3D stacking until heat is not an issue. Think 2.5D side by side and/or outside the box. A large multi layer substrate between SOC chip (CPU, GPU and more on one SOI Silicon chip) and the motherboard. The substrate would have custom 256 bit wide memory attached to it.. Using TSVs in the substrate to support connections between SOC chip and memory. Short trace length and 256 bit wide memory would allow cheaper DDR3 memory to be used in a PS3 instead of XDR RamBuss memory.

    We already know a PS3 Cell refresh to 32nm is going to require a redesign as XDR and Flex I/O interfaces on each side of the chip can’t be downscaled. Multiple thru Cell connections using TSVs will be needed rather than current connections at each end of the chip. A new memory interface to L2 cache would be needed.

    This is only one of a possible number of changes that could come because of what TSVs make possible. With faster cheaper memory, how about adding a faster USB or SATA interface.

    IBM as of Dec 2011 is making a all on one SOI wafer @32nm CPU-GPU-eDRAM for WiiU and what else as both GF8 and IBM are both making these chips.

    1. Yuri

      Thanks — interesting! What do you foresee will be inside the new, slimmer PS3 to be produced this year?

  2. Jeff_Rigby

    What do I foresee? That’s too much of a stretch for me to guess. I’d rather look at what has been done with the Xbox360S SoC in 2010 (designed over 3 years according to the authors of the article to use technology that would be available to fab in 2010) http://www.hotchips.org/wp-content/uploads/hc_archives/archive22/HC22.23.215-1-Jensen-XBox360.pdf. IBM started work with AMD in 2008 incorporating AMD GPUs into/on SOI with their CPUs.

    IBM used a later model PPC CPU and a custom designed GPU in the 360S to mostly hardware emulate (with some software emulation) the Xbox 360 chipset. The next refresh if it’s to last and if it’s going to have HDMI pass-thru (in the leaked Xbox 720 powerpoint coming this season, PS3 4000 chassis is also scheduled for the same time) will need to support low power modes (California to have laws requiring low power modes within 18 months or so). This requires a major hardware redesign and the next refresh may use much later AMD & IBM hardware and software emulate the Xbox where possible. It may support a partial AMD HSA design. Add 4 SPUs to the Xbox design and it could mostly software emulate a PS3.

    Then to support this with rumors and leaks, the Microsoft Domain registration of microsoft-sony.com and sony-microsoft.com which occurred the same month as the Digitimes rumor:

    http://www.ps3news.com/playstation-3-psn-news/rumor-sony-to-start-ps4-production-at-end-of-2011/ which we know can’t be a PS4, so if accurate on the other points, it must be the PS3 4000 chassis and it’s stated to have motion control built in most likely using the Sony Depth camera patented Feb 2012.

    BUT and it’s a big one, the pictures supplied by the South American blog of the PS3 4000 chassis does not have any additional plugs like HDMI pass-thru or a separate Depth camera port. Fake pictures or everything I’m speculating is bogus. There is the possibility that a faster USB port will be used for the depth camera.

    The PS3 3000 chassis (2010) did have an unannounced refresh of the Cell processor but it wasn’t a die size reduction but a change in the manufacturing method that reduced cost 15% as well as reduced trace length inside the cell design which made it more efficient and reduced heat. Your cite links a very good article on the reasons for no Cell die size reduction to 32nm or below. A major redesign would be needed.

    On 9/2010 Sony filed a patent for designs to use PPU and SPUs in a Multi-processor architecture. This is the same month that a Linked in profile for a Sony employee has work on next generation started and the same month as the Xbox 720 powerpoint draft PDF (has X86 as the CPU) and another rumor disclosure has 2010 when Microsoft decided to go with X86 rather than PPC & AMD GPU. IF you haven’t heard there are multiple rumors that look very credible that both Xbox 720 and PS4 will be AMD Full HSA SoCs using 2 packages of 4 Jaguar CPUs each with 2 meg cache each package. This is a 2014 design to be released 1st quarter 2014 most likely because of DDR4 availability.

    A multi-processor architecture is used by the Xbox 360S and AMD HSA Fusion APUs/SoCs. So Sony has a patent for designs to incorporate PPC and SPUs into a more modern design. Cell can not be used with other processors using Multi-processor cache/processor CrossBar switches. http://www.freepatentsonline.com/20100312969.pdf

    So patents and rumors support this speculation but that’s all we have.

    1. Yuri

      Wow, thanks. I remember that news about the Cell process change for the CECH-3000B (it was described as coreless which is of course a funny choice of words — or is this not the redesign you meant?). I was also vaguely aware of the Xbox 360′s CPU & GPU shrink and integration developments but I wasn’t aware of all of these details about emulation and such. I’d also never seen this great document.

      Now about the next generation, I guess you’re not suggesting PS4 and Xbox720 are going to be the same thing (the stakes here are too high but it would be an outrageous trick to pull off) but they may very well have a high degree of commonality (with each other and with PCs), right?

      And for CECH-4000, it may very well use the same chip versions as the CECH-3000B, only with a yet more compact motherboard PCB redesign, correct?

  3. Jeff_Rigby

    Originally posted by Yuri: “Now about the next generation, I guess you’re not suggesting PS4 and Xbox720 are going to be the same thing (the stakes here are too high but it would be an outrageous trick to pull off) but they may very well have a high degree of commonality (with each other and with PCs), right?”

    These appear to be the most accurate rumors, all know the Sony SoC Thebe project name.

    Originally Posted by http://www.evga.com/forums/tm.aspx?high=&m=1659042&mpage=1#1660845
    “Durango -> “I don’t know the name”
    4 Power7 cores + 2 ARM cores(2010) -> 4 AMD Jaguar cores(2012) [4 packages of 4 = 16 CPUs
    48+16 VLIW4 SPs(2010) -> 192 GCN SPs(2012)

    Orbis -> Thebe-Jaguar
    4 AMD Steamroller cores(2011) -> 4 AMD Jaguar cores(2012) [4 packages of 4 = 16 cpus
    128 GCN SPs(2011) -> 192 GCN SPs(2012)
    (year) is the date it was changed to…

    Sony’s version will be optimized for SonyGL 3.0/OpenGL 4.2
    Microsoft’s version will be optimized for DirectX 11.1

    The Jaguar SoC for both consoles will not be the same as the Jaguar SoC for consumers.

    AMD Jaguar vs Sony/Microsoft Jaguar
    H1 2013 vs H1 2014
    LP-DDR3/DDR3 vs DDR4(estimated clock is 2133 MHz 10-10-10-20 1.2v)
    For x86 consumer markets vs specialized versions tweaked for each company
    HSA 3/4ths vs HSA Complete”

    Originally Posted by http://www.forbes.com/sites/briancaulfield/2012/07/24/key-console-player-leaves-amd-for-nvidia/
    “The work with Sony is known inside AMD as ‘Project Thebes.”

    Originally Posted by http://www.neogaf.com/forum/showpost.php?p=37364710&postcount=1265
    “The PS4 AMD project called as Thebe. Previously it used to be based on Themesto and Callisto based chips but now that has been revised. They moved on to a chip called Jaguar replacing the Streamroller. They moved on to TSMC 28nm solution from the 32, which the streamroller is.

    The whole thing basically is APU solutuion, they made the changes considering the 10 year product life cycle and to keep the initial product costs at minimum.

    As of now it is called as the Thebe Jaguar project or TH-J.”

    Originally Posted by http://www.neogaf.com/forum/showpost.php?p=37365190&postcount=1269
    “Also, I just saw some specification for the next Xbox thing, mostly codenames and all. Bobcat, Jaguar. Jaguar which is apparently advanced and it is also being used in the next Xbox project called as Kryptos.

    So looks like there are 2 Jaguar CPC’s (Core plus cache) with 4P/2MB. Clocked at 1.6Ghz(~1.25V).”

    Posted by Jeff_rigby: It appears the AMD APUs were named for moons of Jupiter and Thebe would be the proper spelling (Might explain Orbis Latin for orbit) but the Forbes article spells it Thebes which is a city in Egypt. Kyptros is hidden in Greek and was chosen as the name for the Kyrptos sculpture in front of the CIA building in Langley. Entering Kryptos + Thebes in Google and we find that hidden in the Kryptos sculpture Kryptos, the sculpture at CIA Langly is the Howard Carter account of the King Tut discovery in Thebes Egypt.

    There are now several news articles including the domain registration microsoft-sony.com by Microsoft that some partnership or appearance of a partnership between Microsoft and Sony started sometime before 6/2011. The domain name registration would be a protective measure taken before a potential leak could cause others to register those domain names.



    Also since the efficiencies of a AMD Fusion HSA design require using OpenCL as that is what is supported by the AMD libraries this creates issues with Microsoft’s Direct X (Direct3D & Direct Compute). This partnership might just be Microsoft “reluctantly” agreeing to support OpenCL and possibly OpenGL rather than Direct Compute/Direct3D on the Xbox720.

    Originally posted by Yuri: “And for CECH-4000, it may very well use the same chip versions as the CECH-3000B, only with a yet more compact motherboard PCB redesign, correct?”

    Posted by Jeff_rigby: That’s possible. At some point Sony is going to have to scrap the Cell and RSX and start over with more modern hardware like Microsoft did with the Xbox 360S. My opinion is that they delayed this till they could incorporate two major redesign changes into one. That makes sense now but does not make sense in two years when they will either have to stop selling the PS3 or redesign for the low power modes required in California.

    XTV and Augmented Reality are coming in a big way (to all Networked CE equipment, Smart TVs, Blu-ray players and more) after September of this year when Webkit2 APIs are finished for Linux and Unix. This is the reason the Xbox will have a browser for the first time later this year. The PS3 will have WebGAF HTML5 apps, full Browser with; SVG, HTML5 with Playready DRM and WebGL. Augmented Reality will leverage the browser native support libraries. PSN will expand and offer many new services including a IPTV streaming service like Netflix. h.265, the 4K codec, will be released Jan 2013 allowing 4K blu-ray and higher resolution IPTV streaming at lower bandwidth.

    The above was known in 2010 and was the reason the Xbox 720 powerpoint document had a Xbox361 XTV version with HDMI pass-thru scheduled for release this holiday season. This is also why I feel that the PS3 4000 refresh will be more than using the “same chip versions”.

  4. Jeff_Rigby


    High speed custom array designer with roughly ten years of design experience primarily on the IBM games processors for the PlayStation3 and Xbox360. Owned the delivery of arrays in the 90nm, 65nm, 45nm, 32nm, and 22nm technologies as well as led the design hardware characterization work for all arrays on a 32nm product.

    PlayStation 3
    Team Members: Elizabeth Gerhard
    - Owned the array design for a 1.6GHz TLB array in the 90nm node.
    - Co-owned the delivery of a 128K custom L2 Cache array in the 90nm node.
    - Led a team that delivered the L2cache array macro in the 65nm and 45nm nodes.
    - Owned the register file cell qualification analysis in the 22nm node
    - Owned the design of a 3.2GHz one read one write custom register file array in the 22nm node.

    Xbox 360
    Team Members: Elizabeth Gerhard
    - Co-owned the delivery of a 128K custom L2 Cache array in the 90nm node.
    - Led a team that delivered the L2cache array macro in the 65nm and 45nm nodes.
    - Owned the IO and Core generation for a compilable 6T SRAM system in the 45nm node that generated all the SRAM’s for the GPU.
    - Led a team that defined the simulation methodology used to verify a compilable 6T SRAM system.
    - Served as array lead for a 32nm test site; interfacing with project management to set expected delivery dates, resource needs, and risks associated with design and technology.
    - Owned the delivery of a 3.2GHz 128×64 six read two write custom register file array in the 32nm node.
    - Owned the delivery of a 3.2GHz 72×66 six read two write custom register file array in the 32nm node.
    - Worked on a team to define the array design methodology in the 32nm node.
    - Owned the register file cell qualification analysis in the 32nm node.

    Hardware Characterization
    Team Members: Elizabeth Gerhard
    - Owned the array hardware characterization work for several 32nm test sites.
    - Led an international team to complete the design hardware characterization on a 32nm project.
    - Interfaced between three different groups within IBM to ensure proper allocation of parts, test methodology, and product characterization results pertaining to yield.
    - Developed a tool to automate the data analysis portion of the characterization work and allow more flexibility within the team.
    - Defined the test plan used to characterize all of the arrays

    Posted by jeff_rigby : It confirms part of what I have been saying, a Xbox 360 refresh @32nm is coming and most likely the Oban chip produced by IBM (10,000 wafers Dec 2011) and GloFlo8 is for the XBOX 360 not the WiiU or Xbox 720.

    For the PS3, just testing for 22nm which doesn’t tell us much. In early 2010 news had Sony skipping the 32nm process node and a guess was they were waiting for 22nm. This testing of 22nm listed might be early 2010 and then nothing as there would be a completely different design/cache array and different project if my speculation is correct. Really it tells us nothing of PS3 plans.


    1) PS3 4000 Refresh is using the same chips in the PS3 3000 chassis
    2) Oban is a Japanese name for a large blank gold coin and the international team might be Sony-Microsoft-AMD

  5. Yuri

    Excellent. The stuff you can find on the web… I’m surprised so many companies working on confidential projects apparently don’t have rules about what you can write on your LinkedIn profile.

    As for myself, I only knew Oban as a scotch whiskey from a town by the same name on the west coast of Scotland, but I doubt that has anything to do with this.

    Please keep it coming. Thoroughly interesting stuff to read! The plot thickens…

  6. Jeff_Rigby

    Those who built on Charlie’s (SimiAccurate article) reporting on Oban, did they embellish with guesses and add that it has a 6000 series GPU or did they hear it had a 6000 series GPU. We now can be comfortable the Xbox 720 is going to have a 2014 8000 series Full HSA GPU design @28nm (H1 2014 in one “leak”). (AMD’s Chips in 2014 to Support CPU-GPU Context Switching.. Sony filed a patent in 2010 about CPU-GPU context switching.) Context switching absolutely needed to efficiently use parts of the GPU as a compute element (split the GPU up into multiple GPUs/CUs (supported by OpenCL 1.2)) in games and switch back to traditional GPU use when needed. http://www.xbitlabs.com/news/cpu/display/20120202102405_AMD_Promises_Full_Fusion_of_CPU_and_GPU_in_2014.html

    The point; if they heard a 6000 series GPU is in the Oban SoC, the Xbox 361 will be software emulating not hardware emulating the Xbox 360 chipset. This because low power modes require OS changes to control the CPU & GPU for the different modes and redesigning the PPC and GPU for this would be less cost effective than using more modern hardware. This could not be done for the Xbox 360S in 2010 with design started 2 years earlier so it had to wait till now (my guess). No way to know if this is true now because we don’t know what parts are guesses and embellishments and what was actually leaked.

    But what if accurate but it’s about the next Xbox (361) not 720: Next-gen Xbox will launch fall of 2013 with this GPU http://www.gamingexaminer.com/next-gen-xbox-will-launch-fall-of-2013-with-this-gpu/12815/#.UCiZ9aFA2FB

    “The new GPU for the next-gen Xbox (whatever it is finally named) will be from the 2011′s AMD 6000 series, much like the Radeon HD 6670, this GPU will support DirectX11, 1080p HD, multi-display output and 3D. This according to IGN, who is citing sources close to the project.”

    So not for Xbox 720 but might be for Xbox 360 2012 refresh? This assuming the “source” is accurate. This points to using a A6 AMD APU, remove the X86 processors and put in PPC.

    “The A6-Series A6-3400 accommodates a handful of low-level improvements, such as power gating, which purpose is to reduce processor’s power consumption. For instance, power gating is used to turn off power for idle CPU cores, graphics processing unit and/or Universal Video Decoder. The chip also supports a few low-power states. ”

    Oban is CPU-GPU-eDRAM (in news articles) on SOI process @32nm and the A6 is CPU-GPU+I/O on SOI process @32nm Oban is being made by both IBM and AMD (GloFlo) forges

    Any Xbox 360 Emulation developed for the A6 AMD SoC using PPC processors would work with minor changes to support Backward Compatibility for the Xbox 720 AMD SoC. And the reverse being true with the Xbox 720 Powerpoint having Microsoft supporting BC using 3 PPC processors. It’s sorta chicken and egg but the work would have been done for one and could be used on the other.

    Same applies to the PS3 – PS4 if they do the same using an A6 for PS3 with two 1PPU4SPU Multi-processor packages. Wouldn’t such a A6 with 2PPCs and 8 SPUs be able to emulate both Xbox 360 and PS3? Probably and was the reason I think for the microsoft-sony.com domain registration by Microsoft

    AMD (AMD & GloFlo) and IBM have been sharing technology since 2008. I’d guess Memory controllers and Multi-processor architecture a part of this as a move to HSA and Fabric computing (OpenCL & HSA Foundation) has been obvious to them. http://semiaccurate.com/forums/showpost.php?p=167246&postcount=128

    This is just speculation with supporting cites, it really depends on cost-benefit decisions by both Microsoft and Sony and they have more information than we have.

  7. Yuri

    Striving for synergies and compatibility (by emulation?) between the next Xbox360 design iteration and the next-generation Microsoft console makes sense to me, as it does for a (near?) future PS3 design iteration and PS4/Orbis. But while compatibility between the next Xbox and the next PlayStation would surely boost their market uptake, I’d think it could still be a poor business case because MS and Sony would have to split/share licensing income and together they can’t hope to earn as much as they would individually if each would go their own way. After all, console games are one of those territories where mutually incompatible, proprietary formats are accepted as a fact of life. In fact, the ‘format war’ that erupts every couple of years is business as usual and, I suspect, even needed to create the necessary hype for each platform.

    Years ago, after the 8-bit and 16-bit cycle had been run and 32-bit was coming up, when it was becoming apparent that there was space for only one profitable or ‘winning’ platform in the market at any time, I expected future platforms to run ‘out of phase’ so each could reach profitability in its own lifetime. This didn’t happen, not successfully at least. Market forces seemed to force the platform owners to minimize time difference between their respective launches, ensuring maximum competition and associated media attention.

    Then again, it seems PS4 and Xbox720 or whatever they’ll be called (I still like the idea of an Xbox365 or an Xbox1080) are not going to be the same or even fully compatible, merely not completely different, so developers will have a higher degree of commonality between the platforms they’re coding for, and that seems like a sound idea because that allows them to spend their money where it matters most.

    Perhaps the idea is to reduce the otherwise almost prohibitively high cost of developing a new platform for Sony and Microsoft?

    Do we know for sure those domains were actually claimed by Microsoft, not by some domain squatter?

  8. jeff_rigby

    Posted by Yuri: 1) Perhaps the idea is to reduce the otherwise almost prohibitively high cost of developing a new platform for Sony and Microsoft? 2) Do we know for sure those domains were actually claimed by Microsoft, not by some domain squatter?

    Posted by jeff_rigby:
    1) Yes and if you are going for the most cost efficient most powerful gaming SoC made with building blocks produced by AMD; the design would be the same for both PS4 and Xbox 720, at least for the SoC main silicon. Since it’s already essentially the same why not use the same silicon. You can put some FPGA on the silicon and “tweek” it for each even though it comes out of the forge exactly the same.

    2) yes but there are a number of areas where it might appear that Sony and Microsoft were in a partnership and might cause Microsoft to register the domain microsoft-sony.com.

    Power Modes are not supported by anything in the 360S SoC. This is the BIG BIG issue and could not be resolved until the A6 design which supports power modes!

    Power modes are the reason both Xbox 360 and PS3 need TOTAL from the ground up redesigns. Within about 18 months it will be required in California like the 2009 requirement for TVs. HDMI Pass-thru with overlay for full XTV support also implies Power modes; Standby and instant on when answering a skype call or for web lookup of Movie or TV information. The 2010 Xbox 720 powerpoint had a Xbox 361 supporting XTV (1080P and HDMI pass-thru) to be released this 2012 Season.

    So it’s not just a minor redesign of the 360S that is already nearly a A6 APU except for the partial HSA efficiencies (memory controller and cache), the A6 also has power mode support! Hibernate – Memory has to have power even if the console is powered down in a standby state with one CPU running at a reduced clock speed.

    So a redesign for power modes occurs at this refresh cycle (32nm) or at the next @28nm or 22nm in 2 years. 2012 and with the AMD A6 APU is the FIRST point at which AMD has support for power modes built into their APUs. My guess is that it is occuring this time mainly because both will want to support XTV with HDMI pass-thru and we already know Microsoft planned to do so for 2012. Sony and the PS3 is not a sure bet but the 1PPU4SPU 2010 patent seems to indicate plans for a total redesign using a Multi-processor design (not CELL) which could use the same A6 APU.

    Both the PS3 and Xbox 360 are getting a refresh this year for the first time ever at the same time. Both will have a browser for the first time ever this year. And we have microsoft-sony.com as well as the Digitimes rumor of a PS4 in 2012 which must be a PS3 4000 chassis that has a Kinect like interface.

    Pictures claimed to be the PS3 4000 chassis do not have HDMI pass-thru or a Kinect like port. 1) The Model info sticker does not match previous designs (grey plastic with black lettering) and 2) There is no Model sticker on any of the far pictures, only on one closeup. So it’s possible that the current pictures are fake. It’s also possible that my guess is off by 2 years for one or both.

  9. Yuri

    Either way, a game console switching to entirely new hardware emulating the old hardware within the same generation is going to be a first, isn’t it? I mean emulating previous generations for backward compatibility has been done a couple of times (with varying amounts of hardware acceleration) but emulating the current generation sounds like a new thing to me.

  10. Jeff_Rigby

    More fuel to fire speculation:


    “We had a few “operatives” that were working with HardOCP to mine information.at E3 2011.”

    “It is highly likely that Microsoft’s 3rd generation Xbox will be sporting a new IBM cell processor.”

    “Looking towards the Sony PS4, we hear it is still unsettled between some kind of Bulldozer (would most likely be an APU) variant and a newer updated 32nm IBM cell processor. This one appears to have flip flopped back and forth a few times.”

    Both Microsoft and Sony had decided on AMD GPU and X86 processors for PS4 and Xbox 720 by late 2010 (This only from one post which seems to be believable). So the 2011 flip flopping and Microsoft as well as Sony having a updated 32nm Cell processor (1PPU4SPU) had to be for BC and/or PS3-Xbox 360 refresh.

    June 2011 E3 with the above published July 7 2011 was about the same time the Digitimes PS4 rumor (July 4) and microsoft-sony.com domain registration (by July 15) occurred.

  11. jeff_rigby

    The Sony 2010 patent for PPU+SPU multi-processor architecture was published AFTER Sony had decided to go with AMD X86-GPU….WHY? No one but IBM, Toshiba and Sony can use SPUs. Was it to keep IBM from giving this technology to Microsoft like IBM did with the PPU they developed for Sony’s Cell? http://www.freepatentsonline.com/20100312969.pdf

    I suspect something took place around 7/2010 when Sony filed the patent. Remember 2010 was when Microsoft was deciding to go with X86 over PPU. In any case the Sony patent gives Sony control only if the PPU+SPU is used in a Multi-processor architecture like AMD’s SoCs and APUs. (They covered multiple combinations and Multi-processor architectures.)

    So Sony can deny Microsoft the use of SPUs but why would they do so if “Cell is dead” and SPUs are not going to be used because they can’t compete with GPGPUs? Why file the 2010 patent or publish the patent? My best guess is that 1PPU4SPU (or some # of SPUs) are going to be used in a PS3 refresh or PS4 or both.

    The microsoft-sony.com domain name suggests some partnership between Microsoft and Sony and using SPUs in both next generation Consoles would support this.

  12. Yuri

    Bear in mind that there is often considerable time lag between the moment of ‘invention’, the start of the patent submission, the actual filing and the eventual awarding/publishing — in companies I’ve worked for typically 18 months between the first draft and final awarding. That would put the inital writing of this paper somewhere in 2009.

  13. jeff_rigby


    Stakeholders (game console manufacturers who voluntarily become part of the Energy Star rating system) commented that earlier models may not meet performance requirements. This means there was a consensus arrived at for newer models and they all complied.

    EPA proposes that game consoles brought to market prior to January 1, 2011 are excluded from the scope of the performance requirements.

    Voluntary Requirements agreed to by Game console Manufacturers
    Auto power off
    Standby power .5W
    Active navigation menu 35W
    Active streaming Media 45W

    The PS3 3000 chassis consumes
    Standby .5W
    Active Navigation Menu 61 watts
    Playing game 72-79 watts

    Xbox 360S Valhalla needs 67 watts at the dasboard and 80 watts while gaming.

    So the PS3 4000 chassis has to cut XMB Menu power consumption from 61 watts to 35 watts or less I think that confirms a total redesign if Sony is going to be part of the Energy Star rating system for game consoles. Same for the Xbox 360.

  14. james braselton

    hi there i want too know is what flash drive or ssd is going in the 16 gb ps3 ultra slim

  15. Jeff_Rigby

    Most likely the same SSD Flash drive as going in the PS4 for the same reason. To comply with Energy Star third tier power ratings for an active XMB screen @35 watts the hard disk must be sleeping and the PS3 4000 chassis has to have a more modern CPU and GPU that can be turned off/partially turned off and/or clocked slower.


    It will be in the SoC or on the motherboard. ALL game consoles made from July 2012 on should voluntarily comply which will require SSD Flash for the OS and background task switching/cache. This now includes the PS3 4000 chassis, the Xbox 361 and the coming PS4 and Xbox 720. If they don’t then it puts them at a Sales disadvantage (casual users may choose the more efficient game console) in the short term and within 18 months they can’t be sold in California.

    Third tier EPA Gold Star voluntary compliance @ 35 watts for an active menu screen started July 1 2012 and the PS3 4000 chassis FCC filing occurred July 2012. Most likely Sony wanted to be first to tout their compliance.

    This is still speculation but from a business standpoint most likely needed.

  16. Yuri

    I saw your recent discussion (a week ago) at ‘Gaf. I think I’ll make an update of the shrink roadmap (see very first post) — skipping 32/28nm, going direct to 22nm, no integration envisaged anymore — and post it here for discussion, or perhaps share it with you first, though I know it’ll get a lot more complicated then : )

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>