CELL CHIP?? Big news for apple?

[blueflame]

Higher end powermac

I read someting recently that apple could be working on an even higher end workstation, for far more intensive use than the powermac, mabye this is its roots?
andreas

 

[SnakeDeath]

some dude told me the cell processor was going t
o be absurdly radical and bodaciously awesome.

I think That same dude told me it was FrEEKin SweeT!!! and totally Core! Can anyone set it straight whether it is radical or freekin sweet.

 

[GregA]

Personally I doubt that the Cell will ever be found in a Mac, but IBM could surely use what they learn from the Cell and put that knowledge back into the G5 and any future processors they make for the Mac.

If the cell processor is as good as rumours and marketting/advertising say... then Apple will be looking at their Power options and deciding if it's worthwhile...

IF the chip is amazing, I'm sure it'll find it's way into Apple products.

 

[wizard]

If what informaiton I've seen is true I don't think that a major rewrite is needed at all. Apple has been structuring its software for a long time to leverage multiprocessing. Some things would fit naturally onto Cells structure.

Yeah some apps would take a bit longer to leverage the capability, but some never will. Just as some applicaitons today never leverage SMP.

Dave

Yup

But i think that it would take a very long to re-write the OS and apps so maby they come up with something else better than the cell cpu (IBM has been working on the next gen PowerPC cpu's for a long time, started even before the G5 was out, so let's just sit and wait)

 

[GFLPraxis]

Don't think that it is going to happen
becasue: They have already "made" OS X Tiger and to port it to the cell it need re-writing.
But if it does happen then a lot of people will at least have a closer look at apple before buying their next PC (if not hope for the wintel people that longhorn will be a whole lot better than XP but not likely )

So? They made a special G5 edition of Jaguar. What prevents them from making a special Cell edition of Tiger?

 

[GFLPraxis]

Cell... Power Architecture
970... Power Architecture

What is this porting of OSX that everyone keeps talking about? The base instruction sets for both chips are the same. If anything, the process scheduler in OSX might need tweaked, but I doubt that this is a huge hurdle for Apple, Especially since OSX is based on Next, which in itself had proven to be incredibly portable...

Am I missing something?

Max.

Scroll back, I said the same thing and it was explained.

 

[GFLPraxis]

If what informaiton I've seen is true I don't think that a major rewrite is needed at all. Apple has been structuring its software for a long time to leverage multiprocessing. Some things would fit naturally onto Cells structure.

Yeah some apps would take a bit longer to leverage the capability, but some never will. Just as some applicaitons today never leverage SMP.

Dave

I disagree, the performance increase would be worth it.

After all, people aren't still writing programs for 68k Macs and using emulation. When Apple has an all-cell lineup, people will start writing for Cell procs.

 

[dguisinger]

I haven't seen any mention of cache. On chips like the P4 and I beleive the 970 as well, 1/3 to 1/2 of the die is cache, easily. Therefore external bandwidth is lower.

What appears to happen here is Sony said: Hey, we want to fit more processing power in, ditch the cache, lets add a massively unnessecary amount of bandwidth for ram. Problem is, once they lost the cache, they are dependent on that bandwidth actually being there. If anyone in electronics is here, let us know what the proper signal length is for the speeds they are talking. I beleive it was 6Gbytes per second to ram....that is insane, the memory chips will have to be within centimeters of the Cell chip..... the question becomes can Sony make it work at the speeds they are needing.

Also, the ATI and NVidia chips are massively parrallel and have a ton of floating point power (and are programmable using shaders!). Remember that if Sony doesn't use a dedicated graphics chip, they are pushing all of that onto this chip. The chip isn't dedicated to 3D, and an existing PowerMac G5 cannot do 3D as well as current generation GPUs. So the 10x the power of a 970 isn't very accurate in terms of graphics performance, we don't yet know what this means for real-time graphics. For the people claiming this will seal the coffin on the XBox, don't count on it. We are atleast 1 generation away on ATI and Nvidia gaphics processors.... you have no idea what Microsoft will be using. As far as graphics power goes, they could have nearly identical power when all is said and done. While Sony may have done this because they can reprogram the chip to do things other than GPU, on a graphics intensive game you will rarely do that. On the other hand, the XBox2 will have a next-generation GPU, with probably 6 or 8 pipelines. And at the same time they will have a dual-core PPC-based core for all other functions such as AI, physics, sound, etc. I wouldn't count them out. I would bet Sony went with the Cell to reduce costs, assuming they could eliminate the GPU. However the grave mistake is they are now VERY dependent on Rambus for XDR memory and the two busses they licensed. I bet in the end, it will cost them more to produce even though they saved themselves from the GPU, and the power will be about equal between the two products.

 

[nagromme]

If a $300 Playstation 3 can have 2 Cell Chips in it, then the Cell Chip is a hell of a lot cheaper than a G5 CPU.

Except that consoles are often sold at a loss, to make money on games.

I still theink the Cell could be economical due to economies of scale, though.

 

[GFLPraxis]

I haven't seen any mention of cache. On chips like the P4 and I beleive the 970 as well, 1/3 to 1/2 of the die is cache, easily. Therefore external bandwidth is lower.

What appears to happen here is Sony said: Hey, we want to fit more processing power in, ditch the cache, lets add a massively unnessecary amount of bandwidth for ram. Problem is, once they lost the cache, they are dependent on that bandwidth actually being there. If anyone in electronics is here, let us know what the proper signal length is for the speeds they are talking. I beleive it was 6Gbytes per second to ram....that is insane, the memory chips will have to be within centimeters of the Cell chip..... the question becomes can Sony make it work at the speeds they are needing.

Also, the ATI and NVidia chips are massively parrallel and have a ton of floating point power (and are programmable using shaders!). Remember that if Sony doesn't use a dedicated graphics chip, they are pushing all of that onto this chip. The chip isn't dedicated to 3D, and an existing PowerMac G5 cannot do 3D as well as current generation GPUs. So the 10x the power of a 970 isn't very accurate in terms of graphics performance, we don't yet know what this means for real-time graphics. For the people claiming this will seal the coffin on the XBox, don't count on it. We are atleast 1 generation away on ATI and Nvidia gaphics processors.... you have no idea what Microsoft will be using. As far as graphics power goes, they could have nearly identical power when all is said and done. While Sony may have done this because they can reprogram the chip to do things other than GPU, on a graphics intensive game you will rarely do that. On the other hand, the XBox2 will have a next-generation GPU, with probably 6 or 8 pipelines. And at the same time they will have a dual-core PPC-based core for all other functions such as AI, physics, sound, etc. I wouldn't count them out. I would bet Sony went with the Cell to reduce costs, assuming they could eliminate the GPU. However the grave mistake is they are now VERY dependent on Rambus for XDR memory and the two busses they licensed. I bet in the end, it will cost them more to produce even though they saved themselves from the GPU, and the power will be about equal between the two products.

IIRC, the PS3 will have an NVidia graphics card...
If the PS3 can beat out the XBox 2 in the graphics department, it WILL nail the coffin, because the Nintendo Revolution will have the most unique gameplay and the PS3 would have the best graphics.

So the death of the XBox 2 hinges on the PS3.

HOWEVER, I seem to remember that the XBox 2 was going to be released before the PS3, and if that is true, that too would make it more likely the PS3 would have better graphics.

 

[dguisinger]

IIRC, the PS3 will have an NVidia graphics card...
If the PS3 can beat out the XBox 2 in the graphics department, it WILL nail the coffin, because the Nintendo Revolution will have the most unique gameplay and the PS3 would have the best graphics.

So the death of the XBox 2 hinges on the PS3.

HOWEVER, I seem to remember that the XBox 2 was going to be released before the PS3, and if that is true, that too would make it more likely the PS3 would have better graphics.

Lets not forget the level of problems IBM had with 90nm. They just finally got that all going well. If they have to make the Cell in 65nm just to get the cost down low enough, Sony is in trouble. There are no guarentees that the launch of a process that it will go smoothly, and Sony's timeline is probably VERY dependent on that supply being steady. One small mistake and it could destroy their launch, we've seen what its done to Apple's availability. There is one thing thats nice about what MS is doing: Tried and true technologies. Infact in the past, thats all Console vendors have ever done.... until you got up around the dreamcast....fail. PS2 only succeeded because Nintendo wasn't doing so hot and Dreamcast missed big time. PS1 and PS2 (PS2 specifically) were hell to program for, and full potential wasn't realized for years. Sony is really betting on an experimental platform.... for their sake I hope it works as well as they claim at launch. If the development tools aren't there, or if IBM can't get the chip in quantities....goodbye market leader.

As far as your statement of Sony using a NVidia chip. I haven't seen it mentioned; and I would find it doubtful. There is nothing else beyond graphics that needs 256GFlops of floating point. Sound comes no where close, physics doesn't need more than 1-2GFlops max... there is really no point to making a chip so powerful unless you want to remove the graphics processor.

Remember, these machines won't be used for compressing live HD video and compressing with the latest lossless codecs for broadcast studios, or doing the latest Final Cut Pro editing.... gaming consoles are 3D graphics and decoding only (primarly, i know they are starting to add PVR, but thats low-usage compared to the power they have here).

 

[remingtonhill]

the death of the XBox 2 hinges on the PS3.

HOWEVER, I seem to remember that the XBox 2 was going to be released before the PS3, and if that is true, that too would make it more likely the PS3 would have better graphics.[/QUOTE]
<<>>

Microsoft will never let this happen.

 

[GFLPraxis]

the death of the XBox 2 hinges on the PS3.

HOWEVER, I seem to remember that the XBox 2 was going to be released before the PS3, and if that is true, that too would make it more likely the PS3 would have better graphics.

<<>>

Microsoft will never let this happen.

Why not? Unless they delay the XBox to let the PS3 come out first (yeah, right. They got killed on the first one, being last to the market), the PS3 will probably have better graphics. Cell + a bit more time for new things to come on the market = probably a more powerful system.

If the PS3 ends up with better graphics, and the Nintendo with better gameplay, the XBox 2 will go down the gutter.

 

[GFLPraxis]

Lets not forget the level of problems IBM had with 90nm. They just finally got that all going well. If they have to make the Cell in 65nm just to get the cost down low enough, Sony is in trouble. There are no guarentees that the launch of a process that it will go smoothly, and Sony's timeline is probably VERY dependent on that supply being steady. One small mistake and it could destroy their launch, we've seen what its done to Apple's availability. There is one thing thats nice about what MS is doing: Tried and true technologies. Infact in the past, thats all Console vendors have ever done.... until you got up around the dreamcast....fail. PS2 only succeeded because Nintendo wasn't doing so hot and Dreamcast missed big time. PS1 and PS2 (PS2 specifically) were hell to program for, and full potential wasn't realized for years. Sony is really betting on an experimental platform.... for their sake I hope it works as well as they claim at launch. If the development tools aren't there, or if IBM can't get the chip in quantities....goodbye market leader.

As I said, the XBox 2 hinges on the PS3.

If the PS3 doesn't come through as promised, the XBox 2 may very well displace it, and it will be a war with the NR vs XBox 2. But if the PS3 does come through, XBox will have weaker graphics and not have the extra capabilities of the NR, and will die.

As far as your statement of Sony using a NVidia chip. I haven't seen it mentioned; and I would find it doubtful. There is nothing else beyond graphics that needs 256GFlops of floating point. Sound comes no where close, physics doesn't need more than 1-2GFlops max... there is really no point to making a chip so powerful unless you want to remove the graphics processor.

http://gear.ign.com/articles/571/571598p1.html
To quote:
"The press release basically states that NVIDIA is and has been working on a GPU for Sony's system for some time now, and unfortunately not a whole hell of a lot more than that. This sort of came from left field as it sounded like Sony's Cell processor was going to be doing all of the work in its upcoming console, including the graphical duties."

And, http://www.techimo.com/newsapp/i12702.html

And, http://www.ferrago.com/story/4910

 

[anjaki]

Maybe I'm missing something, but I've been reading up about this chip for the past few weeks and a couple of things have struck me:

1: IBM has bet the house on this thing, which means that either it's going to be huge or it will spell the end of IBM.

2: It will need an OS, there are 4 main options:

• A version of Windows - not likely as no one wants to be held in a Microsoft half-nelson anymore.
• Linux - not likely as it not very consumer friendly, and lacks overall control, think of the government.
• A new OS - has anyone heard any rumours, I certainly haven't.
• OS X - Already prized as the best OS out there, PowerPC experience, links to IBM, Sony etc.

 

[Evangelion]

I was surprised at how much information the guy had crammed into that report when I found it, I'm glad it's now getting some attention around here... skip ahead to the Cell vx. The PC chapter for some Apple related
stuff.

I would take his ramblings with a huge grain of salt: Link. I think Ars Technica is propably the most respectable website when it comes to technical stuff. Hell, the guy who wrote the Cell-article talks about time-travel and anti-gravity on his website!

My thoughts and comment about the Cell: it looks very, very interesting. Will Apple use it? I don't know. But they would be stupid not to even consider it! While they would have to do some major re-designing of the hardware, it shouldn't be THAT hard in the end. As to software.... there shouldn't be that much problems. Hell, Cell even has Altivec!

 

[DavidCar]

There is a detailed explanation of Cell at ArsTechnica here, which is different from the link in the previous post. Also, a part II article is coming tomorrow with more details:

http://arstechnica.com/news.ars/post/20050207-4594.html

 

[dguisinger]

I would take his ramblings with a huge grain of salt: Link. I think Ars Technica is propably the most respectable website when it comes to technical stuff.

My thoughts and comment about the Cell: it looks very, very interesting. Will Apple use it? I don't know. But they would be stupid not to even consider it! While they would have to do some major re-designing of the hardware, it shouldn't be THAT hard in the end. As to software.... there shouldn't be that much problems. Hell, Cell even has Altivec!

Hmm, well I would bet that if Apple were to use it, it would be a variation. Fewer ALUs, more cache, allowing for slower bus speeds and cheaper ram. Further more it would probably have dual primary cores. IBM has a certain agreement that they usually require on their design projects....that they can license the technology they develop. Hence once upon a time when IBM officially made intel chips (Intel used to use several companies such as IBM, AMD, and NEC, all with similar arrangements), after the arrangement was over IBM made the BlueLightning, a 486 class processor that really kicked ass.

I would it put it past IBM to redesign a desktop class processor with similar design elements but in a more-consumer grade design for lower cost parts and better mass-production capabilities. In that case, Apple would have more say.

Lets also remember that these are just MACs, Multiply-Add-Accumilator, from what we gather. While used quite often, they aren't the only functions used in floating point. We'll see if most applications really need 8 parallel MACs on a chip. If 95% don't, then Apple won't spend the money for a premium waste of CPU....

 

[AlanAudio]

A 5x boost or bigger would virtually guarantee that anyone doing heavy work would use a Mac, and would also see a large surge in Mac-only games. That would utterly CEMENT the scientific users (UNIX compatabilitiy + far better performance) and video editors (who are mostly Mac already, this would just pull the few still on Windows) in the Mac platform, and begin to attract the gamers (the only market Apple has failed to touch).

It'd be a huge deal for Apple.

Yep … that certainly makes perfect sense. I also think that those users would find many other benefits, apart from speed, if they switched to Apple.

However, forgive me for sounding like the pessimist again, but is there any reason why this is only good news for Apple ?

Couldn't PCs use the same technique to get similar speed increases ?

 

[Evangelion]

Hmm, well I would bet that if Apple were to use it, it would be a variation. Fewer ALUs, more cache,

Why more cache? Cell as it is has more cache that G5 does! The PPC-CPU in the cell has 512KB of L2-cache (same as G5) and each SPE has 256KB of local memory at it's disposal. The design that IBM now introduced has 2.5MB of cache/local RAM, and that's quite a bit more than what the G5 has.

Further more it would probably have dual primary cores.

Would you rather have two slower cores, or one faster core with SMT? I have no definite answer to that, but dual-cores are not a necessity. Besides, this thing alrady has 9 cores .

 

[wrldwzrd89]

Hmm, well I would bet that if Apple were to use it, it would be a variation. Fewer ALUs, more cache, allowing for slower bus speeds and cheaper ram. Further more it would probably have dual primary cores. IBM has a certain agreement that they usually require on their design projects....that they can license the technology they develop. Hence once upon a time when IBM officially made intel chips (Intel used to use several companies such as IBM, AMD, and NEC, all with similar arrangements), after the arrangement was over IBM made the BlueLightning, a 486 class processor that really kicked ass.

I would it put it past IBM to redesign a desktop class processor with similar design elements but in a more-consumer grade design for lower cost parts and better mass-production capabilities. In that case, Apple would have more say.

Lets also remember that these are just MACs, Multiply-Add-Accumilator, from what we gather. While used quite often, they aren't the only functions used in floating point. We'll see if most applications really need 8 parallel MACs on a chip. If 95% don't, then Apple won't spend the money for a premium waste of CPU....

Your acronym's a bit goofed up...shouldn't it be "MAA"? Anyway, I agree with your points.

 

[Mitthrawnuruodo]

Maybe I'm missing something, but I've been reading up about this chip for the past few weeks and a couple of things have struck me:

1: IBM has bet the house on this thing, which means that either it's going to be huge or it will spell the end of IBM.

2: It will need an OS, there are 4 main options:

• A version of Windows - not likely as no one wants to be held in a Microsoft half-nelson anymore.
• Linux - not likely as it not very consumer friendly, and lacks overall control, think of the government.
• A new OS - has anyone heard any rumours, I certainly haven't.
• OS X - Already prized as the best OS out there, PowerPC experience, links to IBM, Sony etc.

IBM has pushed linux boxes for about a year now... hasn't they...?

Edit: Took a look at IBM's pages and on most of the systems the sell you can choose between Win XP and Red Hat Enterprice as preloaded OS...

 

[johnadurcan]

My 2 cents worth

A short story on where I think it is heading:

Speculation is mounting as to where the future plans of colaboration between Apple and Sony will lead. Often people suspect that it will be headed towards products that sony produce that will aid the Mac's infiltration into the home media segment, yet a deeper look into the pockets that could line both companies will undobtably show that their are bigger plans ahead.

With the recent news of the Cell Processor that will feature in a workstation and then later in the Playstation 3, the applications of the processor have been made apparent. Along with parallel processing of content their is another feature that is detailed of the processor. That of broadband shared processing. Since they seem to be flauting this at a consumer market with the PS3, why would we(the consumer) need to have grid processing?
The reason to this would seem to be that the consumer can have a richer experience of media across multiple platforms within the home, with plans already being made to place the cell processor in a multitude of applications from PC's, TV's and even fridges.

So why does this relate to the Mac, you ask? Well, since the cell processor is a collaboration between Sony, Ibm and Toshiba and the flexibility of the processor, we can assume that the three companies have plans to use it in a large array of their future products. Sony with the Playstation and TVs, Toshiba with their multitude of consumer electronics, and IBM with the Server and workstation range of computers.

Global domination here we come and move over intel and microsoft at the same time! Because, it is afterall, IBM who provides the PowerPC processor to Mac. With such large development costs placed into the Cell processor, it is almost undeniable that IBM will plan to use CELL as it's fore runner to the PowerPC. Since it is also RISC based, it provides the ideal transition processor platform for all it's previous PowerPC line.

Bottom Line: Cell processor is likely to be the new MAC G6 (or whatever they call it. Perhaps it will be called MacCell.)

So now onto the global domination plan. You can clearly see the future of all these products melded together. We all know that Sony is extremly worried about the future plans of microsoft with it's media center PC's, XBOX and variety of operating systems that support these platforms. So how can Sony compete with the Might of Microsoft while at the same time retain the look of a nice company that loves it's customers(Sony, Global Domination? Nah! Never! Or would they?)? The answer to this is already found in the preceeding paragraphs. That's right, CELL Processor. Because the Cell processor will allow a multitude of consumer products to work together in a more consumer friendly way. That is to say, 'press play and it plays'. Not 'press play, and set up this connection, then make sure this connection exists, and hold on, you need this patch, and restart this...'.

So consider this future collaboration between Sony and MAC: Sony announces the new Sony Playstation 3 around the same time that Mac announces the new G6 Cell based Mac. They both announce the collaboration of joint projects to allow media content sharing in the home allowing your MAC to be based in the office with all the media content, and the PS3 able to access that content. In addition, the MAC can access the PS3 processor to ofload processing tasks and also assist in heavy duty tasks such as video rendering and 3D rendering.

Think it sounds far fetched? Well think how microsoft are thinking. You know that they are planning to do the same with the XBOX 2 and PC.

The only two questions are: Who can do it better and who can market it better? I think I know the answer to the first question but the second could be a very big slagging match!

 

[shen]

IBM has pushed linux boxes for about a year now... hasn't they...?

Edit: Took a look at IBM's pages and on most of the systems the sell you can choose between Win XP and Red Hat Enterprice as preloaded OS...

IBM has pushed linux to big companies for well more than a year, and "under the radar" for 2. part of the IBM strat seems to be multiple OS, especially on one box (the mainframes do some cool sandboxing and self heal things) and cell will be no exception. despite a lot of talk about how hard it will be to port OS X, i doubt it will. the NeXT BSD base it is built on is pretty friendly to porting. the apps will take more work than the OS, and even then i will wager it will be minimal. IBM has worked hard to create a powerhouse CPU that is pretty OS neutral, and if anyone can, they can.

cell chips will, if IBM gets its vision, be in your cell phone, TV microwave, PDA, iPod, laptop, desktop, car, and super-computing node. why would Apple not use it? no good reason.........

 

[Booga]

Cell Scalable

I would be surprised if Apple never released a Cell-based Macintosh, but it might not happen with the chip we see today. The number of APUs, from my understanding, is not set by the architecture, nor is the full nature of the core. I would expect to see a G4+2APU embedded chip, a G5+16APU supercomputer chip, and several variations in-between. In effect, you could think of the G5 with a different memory bus as a Cell with no APUs. I wouldn't be surprised to see the next generation PowerPC chip all be based on a Cell core.

Apple could offer the APU instructions as an option to developers, and integrate them into Quicktime and their own high-performance floating point and image manipulation libraries for noticable speedups to existing applications written to the APIs.

After 2 years of hearing how 64-bit "native" code will speed up software (it won't) and such, it's nice to see an innovation in the PowerPC field that really CAN speed up software.