Does Cell Have Any Other Advantages Over XCPU Other Than FLOPS?

[Bobbler]

Only it is. Unless you can point me to another single-chip processor that can build a 3D scan of a human torso in 3 seconds . Obviously for existing apps there's other solutions, but there's some stuff Cell is just better at. Whether these will actually be of use in CE space is another matter.

I wouldn't call Cell cheap. Certainly not as cheap as existing CE solutions. Long term the price will drop but that won't be for a long time.

Quite the opposite to you, I think Cell offers a costly premium solution for high-end products. Sony and Toshiba will be able to use Cell as a single-chip solution at first, offering excessive numbers of features to make a point of having that versatility to hand. I can't see them using Cell to do the tasks of existing cheaper alternatives.

I think you missed the part where I mentioned for the CE realm (maybe I worded it poorly, as it was meant as solely in the CE realm and not talking about elsewhere). And yes, you're right, there are a few cases where Cell does seem to be rather magical (), but the CE realm is not one of them -- CE devices are yet to be limited by processing capabilities in the features they offer; CE devices don't offer features because they just aren't necessary or companies don't think the consumers want them at this point. They might start offering more features in the future, but it won't be because Cell opened up a lot of new opportunities that weren't really present before hand. CE devices really don't require all that much power -- there are benefits to using a Cell (portability of code, general freedom a more general processor allows, the SoC nature of Cell, and of course the ever powerful marketing), but none of them are that the Cell will allow things that were previously not possible with chips out now.

Also you missed the part where I said Cell will be cheap for Sony and probably Toshiba (not cheap, at least not for a while, for anyone else -- I wasn't trying to imply that, but maybe I did) -- Sony especially, because it's going to be fabbing millions of them and the many of the toss-out ones can be used for CE devices; the only true garbage chips for Sony are the ones with actual damaged PPEs/ringbus/etc and/or all of the SPEs busted. Any chip that has 1, 2, 3, 4, 5, 6, or 8 SPEs functional (along with PPE and the rest) can be used -- that makes their overall usuable yields pretty high, making it cheap in a sense that they'd have to make those Cells regardless of yields for the PS3 anyways (of course they probably planned it this way from the start -- Sony seems to be all over subsidizing different parts of its self lately).

I'd love to be wrong and have Cell usher in a new era of CE devices, but the logical part of me says that because there are chips out that are more powerful than what CE devices currently use (and still cheap, and low power) that could be used in CE devices, but aren't means that the power of Cell would likely go to waste in CE devices, instead of offering new possibilities. I'd believe Cell would offer a lot more if CE devices were struggling to find chips of suitable power, but there are tons out there that are cheap and can do a lot of things (more things than current CE devices are doing). Don't get me wrong, there are benefits to using Cell, it's just the benefit of the additional power of the Cell is not really a useful one to a manufacturer or consumer.

Sony is in a good position for the Cell, which is what's important. I hope they do end up using the extra Cells in their TVs and such, and I hope they don't squander the marketing potential that it could offer. The multiple indentical cores on the Cell make it fantastic for what Sony seems to have planned for it, and in that way I think it would be foolhearty to say that Cell is anything but a very well designed chip.

 

[expletive]

Now try finding an ARM which even gets close to a single SPUs performance.

I think argument has been all along that at this point, do we need that level of programmable performance when theres dedicated ASICs that are more than capable of the CE tasks we've been able ot dream up at this point.

Embedded systems are not like desktop systems, they obviously want to reuse code but there is no "legacy" to worry about.

Cells hurdle for mass CE adoption will not be compatibility, it will be cost, power consumption, software, and LOTS of competition.

ARM has most of the mobile phone market wrapped up but certainly not the entire CE sector. They don't seem to have had many design wins in the console sector...

Take a SetTopBox or cheapo DVD apart and you can find all sorts of different devices, all ASICs or off the shelf parts, you'll not find any FPGAs. There's all sorts of different processors in use, PowerPC, MIPs, Transputers and more.

IBM do a range of STB chips based on a low end PowerPC which includes all manner of I/O ports, gfx, MPEG2 audio / video decoding, decryption, Dolby hardware etc. They've very cheap in big volumes and you'll probably find some in those cheapo DVD players. They use around 2W, that was a few years ago...

Then we wanted HD and funky 3D GUIs, so ATI developed a whole new chip.

Then they wanted H.264 so a new batch of chips will be developed.

Now they're talking about VOIP, maybe even with video, the existing chips can decode this but can't encode so that's another new chip...

Each revision is costing 10's of millions to develop. With Cell you could do those changes in software.

I think the point to take from this though is that, as a CE manufacturer, i didnt develop anything. I just bought off the shelf parts that someone else developed. NO risk, low cost, perfect match for my product requirements.

My STB of 2 years ago would have required lots more than just a software upgrade to get it where it is today. It would have needed HDMI, a HArd drive, DVR, different chips on the video output side, firewire, etc. Having a scalable Cell inside would not have changed the fact that i needed a new box to get what i wanted (DVR, HDMI, etc).

I dont think its reliable to assume that to upgrade a CE device to all new functionality that all you need is more processing power, theres usually a lot more to it than that.

STBs also require other chips e.g. demodulators, usually ASICs because no CPU or DSPs are powerful enough. If you can do that on the processor you can save money. CE is all about saving money.

Cell is too big, too expensive and too hot for that end of the market at the moment. Once they have done a die shrink and made the "mini Cell" or "micro Cell" they've talked about it'll be in the right range.

Doesnt this require the assumption that the rest of the chip market stands still while Cell is being shrunk? Who's to say that when that time comes there still wont be more attractive dedicated ASIC designs available?

It'll turn up in high end stuff first and it'll probably be used in place of DSPs, FPGAs and ASICs in industrial stuff. Think mobile phone base stations or big printers. Cells have a big advantage over ASICs and FPGAs in that they are much easier to program and they'll be a lot cheaper.
You can bet STI will be pushing Cell into CE, after all it was in part specifically designed for it.

Where are you getting the information that Cells are easier to program for and will be cheaper than FPGAs and ASICs? How is an FPGA harder to devleop for than a cell? Youre starting from zero with both arent you?

ASICs are cheap if you are producing millions of them, if not you have the problem of paying $15 million + in development costs. FPGAs cost less but they're harder to develop and you'll need a circuit board.

Problem here is that for panasonic, for example, to put a dedicated ASIC in their TV, they are not paying 15 million to develop it or worry about manufacuring. They are getting a price per chip in 10k quantities, and being done with it.

 

[aaronspink]

Here's the Sony roadmap for Cell. The PSX DVD recorder, which is a CE product, used 90nm EE+GS in 2003. That Mini-Cell will compete with ARM and likes, and normal Cell will compete with x86, both in the CE space. But who knows, ARM-based Cell is not impossible.

The PSX DVD recorder was a market failure...

Mini-Cell will have a larger die size and higher processing cost for minimal to no gain in usable features and performance. The likelyhood of cell competing for any CE devices with ARM both within and without sony are slim to none. OTOH, if sony really wants to do it, I'm sure their competitors will be happy and drive Sony's margins to zero for the vast majority of the CE devices sony sells.

Aaron Spink
speaking for myself inc.

 

[aaronspink]

Wow! Never thought I'd see such a big fight about if CELL can actually improve CE products. I mean aaronspink I respect that you think a mini-Cell is overkill, but do you honestly think that there will be NO improvements on using a one SPE CELL chip?

Yes. The issue is that the feature set for the CE devices will not be improved by using a cell instead of many of either the current generation asic or their proliferations. I've spent time thinking about it and cannot come up with 1 viable feature advantage and there don't appear to be any cost advantages in the CE space (likely only cost disadvantages).

Aaron Spink
speaking for myself.

 

[scooby_dooby]

So to break it down there is no lack of processing power in CE devices right now, just lack of good ideas?

That's basically what you're saying right?

 

[expletive]

The PSX DVD recorder was a market failure...

Mini-Cell will have a larger die size and higher processing cost for minimal to no gain in usable features and performance. The likelyhood of cell competing for any CE devices with ARM both within and without sony are slim to none. OTOH, if sony really wants to do it, I'm sure their competitors will be happy and drive Sony's margins to zero for the vast majority of the CE devices sony sells.

Aaron Spink
speaking for myself inc.

Also, the argument has been made that CE devices that use cell will be able to be upgraded down the line due to the extra processor power. This would require Sony to use a version of the Cell that has more processing power than is required for the device at its launch. This would put Sony at an additional cost disadvantage for that device since they are using a higher cost Cell in that device for the possibility it may be upgraded in the future.

So for example, Sony makes a TV that, with all its features, would need a 1:2 Cell. So do they put in a 1:4 cell so they can possibly upgrade software down the line(not that i believe a TV would ever be upgraded) or do they stay cost-effective and cost-competitive by putting in the Cell that gives them just the power they need(1:2 in this example)?

I dont think you can make the argument both ways, either they are able to drive their costs down by making smaller Cells or make the 'upgradeability' argument.

I don't believe the Cell has an advantage in either regard anyway but i thought this was worth throwing out there.

 

[aaronspink]

So to break it down there is no lack of processing power in CE devices right now, just lack of good ideas?

That's basically what you're saying right?

Pretty much. All the enhancements people have come up with fit comfortably within the existing processing envelope. The primary driver in the CE space always has been cost, with the secondary driver being cost, with the tertiary driver being cost, eventually features come into it, but features are at best a temporary advantage often lasting less than 6 months.

Remember that we are talking about devices that in mass production have a BOM of under $50.

 

[expletive]

So to break it down there is no lack of processing power in CE devices right now, just lack of good ideas?

That's basically what you're saying right?

Personally, I think its a few things. A lack of demand for esoteric features that extends beyond what we have now, a cost driven market, abundant competition in the dedicated ASIC space, AND a lack of great ideas.

 

[Vince]

The primary driver in the CE space always has been cost, with the secondary driver being cost, with the tertiary driver being cost, eventually features come into it, but features are at best a temporary advantage often lasting less than 6 months.

Right, like Trinitron. Easily fit within the cheaper/6 month advantage catagory....

Remember that we are talking about devices that in mass production have a BOM of under $50.

Wow, I had no idea that >30" Plasma/LCD/Microdisplay HDTVs, HD-DVD players, and next gen DVR's had such cheap component costs. Amazing how a DVR can have a BOM under $50, but XBox1 with a HDD couldn't in large part due to the one major common component.

Amazing how your argument is entirely based on the segment of the maket that Samsung and Sony don't cater to.

 

[mckmas8808]

Remember that we are talking about devices that in mass production have a BOM of under $50.

Aaron everything CE does not revolve around $50 products. If that was the case why are we getting HD movie players?

 

[AlphaWolf]

Right, like Trinitron. Easily fit within the cheaper/6 month advantage catagory....



Wow, I had no idea that >30" Plasma/LCD/Microdisplay HDTVs, HD-DVD players, and next gen DVR's had such cheap component costs. Amazing how a DVR can have a BOM under $50, but XBox1 with a HDD couldn't in large part due to the one major common component.

Amazing how your argument is entirely based on the segment of the maket that Samsung and Sony don't cater to.

So you're suggesting that somehow cell is going to replace CRTs and LCDs?

 

[mckmas8808]

So you're suggesting that somehow cell is going to replace CRTs and LCDs?

Could CELL not help the visuals of LCD and/or Plasma screens?

 

[AlphaWolf]

Could CELL not help the visuals of LCD and/or Plasma screens?

Nope.

 

[Vince]

So you're suggesting that somehow cell is going to replace CRTs and LCDs?

How hard is it to think before posting? I mean, I'm convinced you're posting just be an ass, since this comment is beyond ridiculous.

Perhaps I'm suggesting, inline with he previous comment which I quoted (hint), that in the CE arena there have been technological improvements which cost more than the competition, but still basically stole the marketplace for 20-odd years.

 

[Vince]

Nope.

Of course it could. For example, Sony's HDTV's already contain discrete chips sets under their WEGA Engine HD, the DRC-MFv2 LSI, New Panel Driver LSI, IPF, and they contain the same S-Master digital amplifier LSI that's in my ES series reciever.

So, you're telling me Sony couldn't replace all these ASICs with, say, a 4 or 6-SPE Cell that can't be used in the PlayStation3? Are you saying that it can't do the above tasks? Can it not apply even more advanced algorithms and additional features that an ASIC, intrinsically, can't support? Why did you post this again?

 

[aaronspink]

Aaron everything CE does not revolve around $50 products. If that was the case why are we getting HD movie players?

Because the movie studios want to get everyone to replace their video libraries. In order for this to happen in a timely manner, the manufacturing cost of the players are going to have to drop enough for $200 players quickly, then then to $100 players, and then to $50 black friday deals.

Right now the players have a significant cost because they are not in mass production. From a hardware perspective, the difficulty of playing a HD movie and a non HD movie is fairly minimal, and as such, I expect the hardware cost to come down FAST. The primary cost factor for at least the first 1-2 years will be the cost of the actual drive mechanism itself.

Content is king, and you don't sell a whole lot of content when it costs $1000 to play it.

Aaron Spink
speaking for myself inc.

 

[aaronspink]

Right, like Trinitron. Easily fit within the cheaper/6 month advantage catagory....

Trinitron was a step function in physical display technologies and didn't deliver a significant portion of the market to sony. But that is orthogonal to the discussion at present, or do
I need to remind you of Betamax?

Wow, I had no idea that >30" Plasma/LCD/Microdisplay HDTVs, HD-DVD players, and next gen DVR's had such cheap component costs.

90+% of the BOM cost in flatpanel display is actually in the flat panel which pretty much everyone pays the same cost for.

Amazing how a DVR can have a BOM under $50, but XBox1 with a HDD couldn't in large part due to the one major common component.

DVR BOM costs are currently in the $75-85 dollar range with the single biggest cost currently being the HD. There is incredible pressure to push the DVR cost downwards by companies like Comcast, DirecTV, etc since they can charge the same price to the consumer but make for money.

Amazing how your argument is entirely based on the segment of the maket that Samsung and Sony don't cater to.

Samsung targets a significant portion of the market that is very price sensetive. They have a very effective strategy of pricing just a marginal amount more than the generic market and relying on they brands built in equity to close the gap. Plus they are the primary manufacture for many of the ingredients both them and their competitors use.

Sony also caters to the mainstream significantly, as that is where the majority of the sales are made.

Aaron Spink
speaking for myself inc.

 

[seismologist]

Someone want to wrap up this thread with the consclusions we've made. So far I've come up with.

Cell is good for physics. Good for high end CE devices. Bad for $50 DVD players...anything else?

 

[scooby_dooby]

Bad for ANY dvd player, or CE device not made by SOny or Toshiba as the costs of the CELL processor would not be justified when cheaper custom solutions are readily available.

 

[Vince]

Trinitron was a step function in physical display technologies and didn't deliver a significant portion of the market to sony. But that is orthogonal to the discussion at present, or do I need to remind you of Betamax?

It didn't? I could have sworn Sony basically dominated the '80s and early '90s with Trinitron.

"Sony dominated the market for traditional, cathode ray tube TVs, with its Trinitron line commanding a premium." - Tom Edwards, Pacific Media Associates.​

And how is it orthogonal to our discussion? It was a logical extention of your comment; perhaps you just didn't think it through?

90+% of the BOM cost in flatpanel display is actually in the flat panel which pretty much everyone pays the same cost for.

A term: "Product differentiation" ... lost on Aaron. In a time of increasing commoditization, one needs to find a way to differentiate himself; technology is ultimately the way it's going to be done. You just helped make the case for my position, thank you.

DVR BOM costs are currently in the $75-85 dollar range with the single biggest cost currently being the HD.

Can I get a link or quote? Thank you.