link below helps dispell some of the misconceptions about Cell. It doesn't sound like Cell is really as applicable to the PC as we thought or at least according to this guy.
P.S. I don't know that much about computers so be please be easy on me.
He is quite right, it will not become the next PC. It cannot and will not ever run x86 software in a reasonable fashion. It could, however, run Mac software just fine -- and if Apple chose to do so it could support CoreAudio/Image/Video on the Cell just like it is supporting it on VMX and the GPU hardware. Plus the Cell could be exposed directly to the developers so that anybody can take advantage of all that computing power. Not all problems will do well on it, but most of today's common media processing algorithms would do very well and many problems might have new solutions made possible by the amount of computing power in this chip.
And no, I don't see why this chip would not fit nicely into a Mac. Sometime in 2006.
In a multimedia workstation using the SPEs as sort of a super Altivec, the cell could have a huge boost over current hardware. The average consumer (whose programs usually don't even use altivec) would see very little difference as only the main CPU would be used.
Currently important part of "real world" performance of a PC is related to Word/Excel (replacing a the letter "e" in a 1 GB file ), internet browsing, scrolling, gaming. But the battle for the home/entertaiment market already started. It is too early to tell which KIND OF DEVICE is going to be a winner, yet the best positioned company. It is very likely, however, that in 5 years the most important "real world" tests will include a bit different set of applications. So I am not sure all arguments regarding the suitability of the Cell processor for the "General purpose" computer are valid - the "General purpose" will change.
Exactly! So-called "typical consumers" will start using iLife (and other similar apps) more and more, and they will become the mainstream apps, moreso than things like MS Office. And all those media apps could REALLY take advantage of the Cell. Especially if Apple makes all of the Core frameworks use them.
And hey, even if the main apps a person uses doesn't get a lot of use out of the Cell, eye-candy can get a huge boost from it. And "pretty" is definitely something Apple is going for these days.
The average consumer (whose programs usually don't even use altivec) would see very little difference as only the main CPU would be used.
This is simply not true -- all applications use system services that now use VMX/AltiVec extensively. There are complaints about the slow interface all the time, and Cell could definitely fix that. Quartz2D (i.e. all apps with a GUI), OpenGL (i.e. all games), CoreAudio, iPhoto, iMovie, iDVD, Safari (image decoding & network stack acceleration), screen savers, desktop pictures, etc etc would all benefit enormously from SPE acceleration and even simply from the stupendous bandwidth this chip provides.
Most operations which are slow on a consumer's machine are things which can be accelerated with VMX/SPE. The majority of other stuff is already fast enough that the Cell's 4 GHz PPE would deal with it at least as well as a 2 GHz G5... probably better because the SPEs will be taking the expensive stuff off of its shoulders.
And if Apple machines had this capability in them waiting to be used, I'm sure there's a programmer around somewhere that would think of something to use it for.
We'll probably see 2 to 3 PowerMac updates before we see a cell processor in any Mac if you ask me, and that's if we ever even see one. It's not like the PowerPC doesn't have a future, and a roadmap to gain significant ground in the future. PPC could be a 10x better alternative to cell by the time a cell is ready to be in todays Macintosh.
. . . It's not like the PowerPC doesn't have a future, and a roadmap to gain significant ground in the future. PPC could be a 10x better alternative to cell by the time a cell is ready to be in todays Macintosh.
Maybe in a few years a standard PPC will have inherited so much Cell technology that the two will be indistinguishable. There have been discussions about a new super AltiVec 2 for a long time, I'd say it is here in the Cell. The possibilities seem unlimited for what can be done with Cell's modular architecture.
I tend to be a dreamer, but realistically it will boil down to performance and economics. The best, however Apple defines best, will win.
Does someone keep Steven Milunovich's track record. Steven Milunovich is the Merrill Lynch analyst who speculated about Apple-Sony partnership and Apple's use of Cell processor. I remember that he also speculated about the mac-mini and iPod shuffle (before they were announced, the rumors were circulating on these boards, however). Everybody sad that he is speculating on rumors but the guy may have some internal info.
Some of you seem to have been caught up in our own dreaming scheme in here. Cell is not going to be in the NEXT PowerMac no matter how much you bullshit yourself. Cell may wind up in a Future PowerMac, but chances are it wont be in 2005. Thinking anything else is far too optimistic.
It may not be in the next PowerMac, infact I suspect that it would be atleast another 6 months before Apple would have anything ready. Further I would not be surprised if Apple goes with a different variant of Cell possibly with a diefferent memory controller. Rambus would be difficult to push on to the desktop market.
Even given all of the above I wouln't be surprised to see Apples first Cell machine be a PowerBook with a Cell variant optimized for portable use.
Quote:
snoopy, SPE's are not going to replace a GPU. It could conceivably help, but it hardly seems as wise as to use it for something else, and use Core Image to utilize a real GPU that has been designed by the best graphics engineers to do just that.
High end graphics cards will still be needed.
I tend to agree here, this informaiton released so far does not indicate that Cell and the SPE are optimized at all for replacement of the Graphic card. It may replace some functionality but that is only an issue of distribution of capability.
The reference article is very interesting and accurate based on the information I've seen. However I think the conclusion is simply wrong. That however is based partly on where you think computers are going in the future and just what is the average desktop computer used for.
In the case of Apple I could see them making good use of Cell right off the bat. Take audio for example an area that many will admit that Apple has been weak with. The SPE's in Cell, likely one or two, could provide a state of the art sound facility for Apple hardware. Would Apple take this approach, that is adopt a new hardware architecture to enhance it systems this way - well I don't know but it is obvious that the posibility is there.
The other issue that I see with respect to implementation of PC based on Cell is the PPE. If the admititedly thin information on power usage is any indication, then we could very well see the PPE implemented in Apple hardware in the future. Attaching all the SPE's may or may not be required. It is not advisable to come to conclusions about Cell based on the one chip released at ISSCC. Everyone involved in Cell have gone to great lengths to indicate that the design is modular and will be targetted at a broad spectrum of applications.
None of this is speculation. However Apples plans are not known. Apple may very well have a alternative low power 64 bit processor in the pipe line, that would eliminate one need for Cell. They may also take a different approach to powering CORE. Apple has indicated that future machine will be able to take advantage of CORE with a significant increase in performance over current hardware. That could be nothing more than a dual core processor as a base machine. Or it could mean a processor optimized for those sorts of tasks - Cell. Sure what Apple is up to is speculation, however what Cell is capable of is slowly becoming public.
I look at it this way, if somebody gave me a choice of a G4 based computer and a Cell based computer both running OS/X or Linux which would I choose given that each is built for maxiumum performance from that chip. At this point I'd have to opt for Cell with the idea that performance increase as the system better makes us of Cell. Not that Cell would be all that bad right now from the looks of it. And yes good solid performance figures would change my mind, but it does look like Cell would build an excellent low cost machine.
Dave
Quote:
Originally posted by onlooker
Finally! This is what I've been saying, but people keep jumping to conclusions with their speculative hypotheses, and I'm in (Fill in the Blank) field of work so I know what I'm saying.
s. I'm just a pixel pusher, but it seems to me that this would be feasible with current apps working fine with the core PPC 970 + Altivec and then the developer implementing over time Core Audio, Core Image, and Core Data API's without much of a hiccup?
Could this be part of Apple's plan all along or am I way of base?
Yep! I suspect that Apple had Cell or something like it in mind when they developed some of the CORE faciliites. One does have to understand though that now all CORE facilities would take advantage of Cell. Many could though and would move Apple significnatly forward with respect some facilities such as Audio.
Quote:
Sorry, if someone down stream has already commented about this.
Sorry, but I don't see the analogy. IBM could have chosen to mate their 4+ GHz design to an MPX bus, or even a 1 Hz binary blinking light -- that is easily technically achieveable. Building a short pipeline 4 GHz design is not technically achieveable.
Boy is that a negative attitude or what. The whole approach to Cell is different than the sychronous logic of the past, isn't a bit of a jump to conclude that it is totally impossible. I've seen indications (that I don't trust) that the pipeline is maybe only a couple of stages longer than a G4. IF so that is not to bad.
Further if the SMT faciltiy can occupy the processor while another thread is suffering a branch mispredict then the PPE should do pretty well with the given pipelines.
Of course solid information that is non NDA would be fantastic and would quickly put this issue to rest.
Quote:
It is also unlikely that Freescale has the technical wherewithal to achieve what STI has. They simply don't have the brain trust and process capabilities.
Well they aren't Intel size by any means but that doesn't mean they aren't capable men. In any event Freescale has already debuted their solution to MaxBus.
Boy is that a negative attitude or what. The whole approach to Cell is different than the sychronous logic of the past, isn't a bit of a jump to conclude that it is totally impossible. I've seen indications (that I don't trust) that the pipeline is maybe only a couple of stages longer than a G4. IF so that is not to bad.
Further if the SMT faciltiy can occupy the processor while another thread is suffering a branch mispredict then the PPE should do pretty well with the given pipelines.
Of course solid information that is non NDA would be fantastic and would quickly put this issue to rest.
I give up. Lets just wait until they publish more information.
Anyone remember the rumors about Apple putting a dsp/simd core on the north bridge?
It would certainly fit with the technology of the time ( too big to go on the cpu core - lots of room on the northbridge ). It could be that Apple have been working on cell for a long time already.
I think it is unlikely that the chip that Sony is going to use would be the chip that Apple will use, but if they have licensed/participated in cell then IBM could easily be building a custom cpu for Apple.
Indeed, with the Cell being its own cpu, and the 970 not having an on chip memory controller, putting the cell on the northbridge makes a weird sort of sense.
Does anyone know any details about the internal databus of the cell chip? Could they be using the PI of the 970?
Comments
Originally posted by cben
link below helps dispell some of the misconceptions about Cell. It doesn't sound like Cell is really as applicable to the PC as we thought or at least according to this guy.
http://www.jonpeddie.com/index.shtml
P.S. I don't know that much about computers so be please be easy on me.
He is quite right, it will not become the next PC. It cannot and will not ever run x86 software in a reasonable fashion. It could, however, run Mac software just fine -- and if Apple chose to do so it could support CoreAudio/Image/Video on the Cell just like it is supporting it on VMX and the GPU hardware. Plus the Cell could be exposed directly to the developers so that anybody can take advantage of all that computing power. Not all problems will do well on it, but most of today's common media processing algorithms would do very well and many problems might have new solutions made possible by the amount of computing power in this chip.
And no, I don't see why this chip would not fit nicely into a Mac. Sometime in 2006.
Originally posted by Programmer
And no, I don't see why this chip would not fit nicely into a Mac. Sometime in 2006.
So when you spoke earlier about waiting for the big WWDC announcement, you were referring to WWDC '06?
And hey, even if the main apps a person uses doesn't get a lot of use out of the Cell, eye-candy can get a huge boost from it. And "pretty" is definitely something Apple is going for these days.
Originally posted by BenRoethig
The average consumer (whose programs usually don't even use altivec) would see very little difference as only the main CPU would be used.
This is simply not true -- all applications use system services that now use VMX/AltiVec extensively. There are complaints about the slow interface all the time, and Cell could definitely fix that. Quartz2D (i.e. all apps with a GUI), OpenGL (i.e. all games), CoreAudio, iPhoto, iMovie, iDVD, Safari (image decoding & network stack acceleration), screen savers, desktop pictures, etc etc would all benefit enormously from SPE acceleration and even simply from the stupendous bandwidth this chip provides.
Most operations which are slow on a consumer's machine are things which can be accelerated with VMX/SPE. The majority of other stuff is already fast enough that the Cell's 4 GHz PPE would deal with it at least as well as a 2 GHz G5... probably better because the SPEs will be taking the expensive stuff off of its shoulders.
And if Apple machines had this capability in them waiting to be used, I'm sure there's a programmer around somewhere that would think of something to use it for.
I think cell is ready for it's own forum.
It's infested every thread in this forum.
We'll probably see 2 to 3 PowerMac updates before we see a cell processor in any Mac if you ask me, and that's if we ever even see one. It's not like the PowerPC doesn't have a future, and a roadmap to gain significant ground in the future. PPC could be a 10x better alternative to cell by the time a cell is ready to be in todays Macintosh.
Originally posted by onlooker
. . . It's not like the PowerPC doesn't have a future, and a roadmap to gain significant ground in the future. PPC could be a 10x better alternative to cell by the time a cell is ready to be in todays Macintosh.
Maybe in a few years a standard PPC will have inherited so much Cell technology that the two will be indistinguishable. There have been discussions about a new super AltiVec 2 for a long time, I'd say it is here in the Cell. The possibilities seem unlimited for what can be done with Cell's modular architecture.
I tend to be a dreamer, but realistically it will boil down to performance and economics. The best, however Apple defines best, will win.
Originally posted by onlooker
Some of you seem to have been caught up in our own dreaming scheme in here. Cell is not going to be in the NEXT PowerMac no matter how much you bullshit yourself. Cell may wind up in a Future PowerMac, but chances are it wont be in 2005. Thinking anything else is far too optimistic.
It may not be in the next PowerMac, infact I suspect that it would be atleast another 6 months before Apple would have anything ready. Further I would not be surprised if Apple goes with a different variant of Cell possibly with a diefferent memory controller. Rambus would be difficult to push on to the desktop market.
Even given all of the above I wouln't be surprised to see Apples first Cell machine be a PowerBook with a Cell variant optimized for portable use.
Quote:
snoopy, SPE's are not going to replace a GPU. It could conceivably help, but it hardly seems as wise as to use it for something else, and use Core Image to utilize a real GPU that has been designed by the best graphics engineers to do just that.
High end graphics cards will still be needed.
I tend to agree here, this informaiton released so far does not indicate that Cell and the SPE are optimized at all for replacement of the Graphic card. It may replace some functionality but that is only an issue of distribution of capability.
Dave
Originally posted by onlooker
And the speculation (in every thread) goes on.
I think cell is ready for it's own forum.
It's infested every thread in this forum.
Here's one:
Cell Processor
In the case of Apple I could see them making good use of Cell right off the bat. Take audio for example an area that many will admit that Apple has been weak with. The SPE's in Cell, likely one or two, could provide a state of the art sound facility for Apple hardware. Would Apple take this approach, that is adopt a new hardware architecture to enhance it systems this way - well I don't know but it is obvious that the posibility is there.
The other issue that I see with respect to implementation of PC based on Cell is the PPE. If the admititedly thin information on power usage is any indication, then we could very well see the PPE implemented in Apple hardware in the future. Attaching all the SPE's may or may not be required. It is not advisable to come to conclusions about Cell based on the one chip released at ISSCC. Everyone involved in Cell have gone to great lengths to indicate that the design is modular and will be targetted at a broad spectrum of applications.
None of this is speculation. However Apples plans are not known. Apple may very well have a alternative low power 64 bit processor in the pipe line, that would eliminate one need for Cell. They may also take a different approach to powering CORE. Apple has indicated that future machine will be able to take advantage of CORE with a significant increase in performance over current hardware. That could be nothing more than a dual core processor as a base machine. Or it could mean a processor optimized for those sorts of tasks - Cell. Sure what Apple is up to is speculation, however what Cell is capable of is slowly becoming public.
I look at it this way, if somebody gave me a choice of a G4 based computer and a Cell based computer both running OS/X or Linux which would I choose given that each is built for maxiumum performance from that chip. At this point I'd have to opt for Cell with the idea that performance increase as the system better makes us of Cell. Not that Cell would be all that bad right now from the looks of it. And yes good solid performance figures would change my mind, but it does look like Cell would build an excellent low cost machine.
Dave
Originally posted by onlooker
Finally! This is what I've been saying, but people keep jumping to conclusions with their speculative hypotheses, and I'm in (Fill in the Blank) field of work so I know what I'm saying.
Originally posted by tink
s. I'm just a pixel pusher, but it seems to me that this would be feasible with current apps working fine with the core PPC 970 + Altivec and then the developer implementing over time Core Audio, Core Image, and Core Data API's without much of a hiccup?
Could this be part of Apple's plan all along or am I way of base?
Yep! I suspect that Apple had Cell or something like it in mind when they developed some of the CORE faciliites. One does have to understand though that now all CORE facilities would take advantage of Cell. Many could though and would move Apple significnatly forward with respect some facilities such as Audio.
Sorry, if someone down stream has already commented about this.
Originally posted by Programmer
Sorry, but I don't see the analogy. IBM could have chosen to mate their 4+ GHz design to an MPX bus, or even a 1 Hz binary blinking light -- that is easily technically achieveable. Building a short pipeline 4 GHz design is not technically achieveable.
Boy is that a negative attitude or what. The whole approach to Cell is different than the sychronous logic of the past, isn't a bit of a jump to conclude that it is totally impossible. I've seen indications (that I don't trust) that the pipeline is maybe only a couple of stages longer than a G4. IF so that is not to bad.
Further if the SMT faciltiy can occupy the processor while another thread is suffering a branch mispredict then the PPE should do pretty well with the given pipelines.
Of course solid information that is non NDA would be fantastic and would quickly put this issue to rest.
It is also unlikely that Freescale has the technical wherewithal to achieve what STI has. They simply don't have the brain trust and process capabilities.
Well they aren't Intel size by any means but that doesn't mean they aren't capable men. In any event Freescale has already debuted their solution to MaxBus.
Originally posted by wizard69
Boy is that a negative attitude or what. The whole approach to Cell is different than the sychronous logic of the past, isn't a bit of a jump to conclude that it is totally impossible. I've seen indications (that I don't trust) that the pipeline is maybe only a couple of stages longer than a G4. IF so that is not to bad.
Further if the SMT faciltiy can occupy the processor while another thread is suffering a branch mispredict then the PPE should do pretty well with the given pipelines.
Of course solid information that is non NDA would be fantastic and would quickly put this issue to rest.
It would certainly fit with the technology of the time ( too big to go on the cpu core - lots of room on the northbridge ). It could be that Apple have been working on cell for a long time already.
I think it is unlikely that the chip that Sony is going to use would be the chip that Apple will use, but if they have licensed/participated in cell then IBM could easily be building a custom cpu for Apple.
Indeed, with the Cell being its own cpu, and the 970 not having an on chip memory controller, putting the cell on the northbridge makes a weird sort of sense.
Does anyone know any details about the internal databus of the cell chip? Could they be using the PI of the 970?
Originally posted by mmmpie
Does anyone know any details about the internal databus of the cell chip? Could they be using the PI of the 970?
No, it is completely different.
;^p
Originally posted by MacRonin
I say that Cell becomes the default 'CPU' for the Mac mini sometime around mid-2006...
... and makes the Antares Powermacs we all bought in 2005 look lame.