I'm no networking expert but a little Googling around will show that IBM has been building on-chip hardware acceleration for the protocols that send data out the NIC's thus speeding up the broadband connection.
If you are on dial-up you don't get to play, it's just out of your league.
That's the point. You can't just hand someone broadband on a chip.
This is actually a series of questions for Programmer or someone of his ilk, largely because I'm a doofus who doesn't really "get" some of this stuff?
Am I right in thinking that, having read a small amount of the Cell material from today, I can't help but hear coffin nails being banged into SGI?
Could Cell be deployed as an "outboard" technology tied to either an Xserve or Power Macintosh "controller" handing out simulation/visualisation tasks? If so. what kind of Cell unit to Macintosh controller ratio could one expect? 8 to 1, 16 to 1, more? (!!!)
If the answer to these questions are largely affirmative, would it be reasonable to see Macintosh/Cell implementations being used in areas like?
geophysical modelling
signal processing
scientific simulation
HD image render farm
Could we reasonably expect to see a 64 dual-processor Xserve's feeding large data sets to (for example) 1024 Cells using Xsan as a SAN filesystem? And if we did, what sort of performance could we expect?
Looks like Cell will have to wait for 65 nanometers.
From MacCentral...
Quote:
The processor shown Monday was only a prototype, and it's likely that the high-volume shipments of the processor will come when the three companies are ready to make chips using a 65-nanometer processing technology, Glaskowsky said. That technology will allow the companies to shrink the chip and reduce their manufacturing costs, he said.
Not necessarily used outboard, but the Power Processor Element core is sans out-of-order circuitry making it hamstrung in non-carefully scheduled or sloppy code. Having a 970MP AND a Cell would make a hell of a team and obviate an independent GPU.
Mark -- the Cell has a Power core inboard, so why would it need to be used "outboard"?
In my head, I see a shelf of Cell blades attached to an Xserve by some high-speed interconnect like Infiniband.
So I guess what I mean by outboard is that it's slightly removed from the host that controls it which means that you don't have to build in the capability to drag all of that heat away into what should remain mainstream computing products.
Does that philosophy not then given you an opportunity to attach a small Cell breakout box (for audio or video processing) to a PowerBook using a FireWire umbilical.The Cell unit does all the heavy lifting (encoding, effects, whatever) and then feeds the finished data stream down FireWire to be written to disk., whilst th main 9xx or 3xx "host" processor controls less demanding tasks.
Does that not make it feasible to run hitherto deskbound applications on a laptop? Could you - I/O connections notwithstanding - land up with a 96-track recording solution running on a next-gen PowerBook?
I don't know how this thing wants to get fed, which is I'm asking the question.
I was at the Cell event today, and quoted in some of the news stories. I also have the ISSCC technical papers.
The PowerPC core in the Cell prototype chip is NOT a Power5, as speculated here. According to IBM, this core was designed from scratch for this application. One critical difference is that the new pipeline executes instructions in strict program order rather than reordering instructions to improve throughput as is done with Power5.
Also, IBM has not described the core as "simultaneous multithreaded", just "multithreaded." I presume from this that the multithreading is coarse-grained-- only one thread is active at a time, unlike Power5 which can execute instructions from two different threads in the same cycle.
The logic design for the Cell CPU was optimized for higher clock speeds in a given process than Power5 can achieve. This is a good tradeoff for more linear multimedia algorithms, but reduces effective throughput on other types of code.
I think it's reasonable to suppose that if Apple were interested in using the Cell architecture, it would prefer to use a version of the design that includes a Power5 core in place of the one in the Cell prototype.
This answers my own question pretty well. And the guy seems to be right.
I got utterly confused after reading a dozen of articles and a hundred of posts on various forums.
It is not a POWER5. It is a dual-issue, in-order core running at 4+ GHz. At that clock rate you can make up for a lot of the limitations of the core... its probably in the same performance ballpark as a 1.8-2.0 GHz 970 on non-optimized code. On optimized code it'll fly.
The other thing that they are saying about the Cell is that it is highly modular. Basically it is centered around this EIB on-chip bus and everything is a peer on this bus. If you have a different Power core then why not stick on this bus interface and make it a peer on the EIB. Or put two of them there. The main obstacle is probably that this chip is running at 4 GHz!
I don't know whether this has any implications for Apple (in using a Power5 derived processor), but I thought the following was quite interesting...
Quote:
IBM is also talking up its p5 575 box that is designed to be part of a server cluster. The system takes up just 2U of rack space but can hold 8 single-core Power5 chips each running at 1.9GHz.
It seems that IBM may well be tweaking things to allow for cooler running of all it's Power family?
One of the Power5's big features was reduced power consumption. At a very low-level in the basic circuit design of the chip they did some things which optimize power consumption. The features are part of what this new core has inherited, which is why there is any hope of this Cell chip fitting in a game console package at anything close to 4 GHz.
I posted this over in the other Cell thread and thought it might spark some fond memories here too.
Another nagging thought keeps entering my brain. A few months ago some one on these boards was touting Apple's next generation lap top chip to be some agglomeration of 440 style chips using low low power. In looking at information on Cell, the SPE's(I think this is the right term) are actually somewhat independent specialized cpu's??? Is it even remotely possible that this person's posts about multiple 440's on die for a low power lap top chip had any validity whatsoever, or at least were spawned from the Cell??
With a lot of people including me dismissing his posts as anything from dubious to out and out BS, wouldn't it be a real kick in the pants if it was Cell he was describing and it does end up in an Apple laptop in some form.
Please forgive me if I'm totally wrong as I really have no clue about this technology.
IF they will use the Cell processor we won't see it before 2006. but i could imagine that IBM and apple were working on the G5-successor (which - of course - could be based on a implementation-variant of the new Cell-design-ideas). i expect IBM to be very tight-lipped on new CPUs made for apple and bet we will all be surprised when the G5s successor will be presented...
IF they will use the Cell processor we won't see it before 2006. but i could imagine that IBM and apple were working on the G5-successor (which - of course - could be based on a implementation-variant of the new Cell-design-ideas). i expect IBM to be very tight-lipped on new CPUs made for apple and bet we will all be surprised when the G5s successor will be presented...
A thought hit me as I read your post. Sony will use the Cell in the PS3. IBM will produce a workstation based on the Cell. Maybe these are two different chips? IBM and Apple may use a variant of Cell for high end computers that is different from the one Sony will use in the PS3.
Comments
Originally posted by Aphelion
I'm no networking expert but a little Googling around will show that IBM has been building on-chip hardware acceleration for the protocols that send data out the NIC's thus speeding up the broadband connection.
If you are on dial-up you don't get to play, it's just out of your league.
That's the point. You can't just hand someone broadband on a chip.
Originally posted by onlooker
That's the point. You can't just hand someone broadband on a chip.
No, the point is you can speed up broadband on the chip.
Am I right in thinking that, having read a small amount of the Cell material from today, I can't help but hear coffin nails being banged into SGI?
Could Cell be deployed as an "outboard" technology tied to either an Xserve or Power Macintosh "controller" handing out simulation/visualisation tasks? If so. what kind of Cell unit to Macintosh controller ratio could one expect? 8 to 1, 16 to 1, more? (!!!)
If the answer to these questions are largely affirmative, would it be reasonable to see Macintosh/Cell implementations being used in areas like?
- geophysical modelling
- signal processing
- scientific simulation
- HD image render farm
Could we reasonably expect to see a 64 dual-processor Xserve's feeding large data sets to (for example) 1024 Cells using Xsan as a SAN filesystem? And if we did, what sort of performance could we expect?From MacCentral...
The processor shown Monday was only a prototype, and it's likely that the high-volume shipments of the processor will come when the three companies are ready to make chips using a 65-nanometer processing technology, Glaskowsky said. That technology will allow the companies to shrink the chip and reduce their manufacturing costs, he said.
Originally posted by murk
Looks like Cell will have to wait for 65 nanometers.
From MacCentral...
Sounds a lot like guessing to me.
Mark -- the Cell has a Power core inboard, so why would it need to be used "outboard"?
Originally posted by Aphelion
No, the point is you can speed up broadband on the chip.
Helllllooo? THe point of what I said is not about that at all. You can't just hand someone broadband on a chip. It makes no sense. That was my point.
Originally posted by Programmer
Sounds a lot like guessing to me.
Mark -- the Cell has a Power core inboard, so why would it need to be used "outboard"?
In my head, I see a shelf of Cell blades attached to an Xserve by some high-speed interconnect like Infiniband.
So I guess what I mean by outboard is that it's slightly removed from the host that controls it which means that you don't have to build in the capability to drag all of that heat away into what should remain mainstream computing products.
Does that philosophy not then given you an opportunity to attach a small Cell breakout box (for audio or video processing) to a PowerBook using a FireWire umbilical.The Cell unit does all the heavy lifting (encoding, effects, whatever) and then feeds the finished data stream down FireWire to be written to disk., whilst th main 9xx or 3xx "host" processor controls less demanding tasks.
Does that not make it feasible to run hitherto deskbound applications on a laptop? Could you - I/O connections notwithstanding - land up with a 96-track recording solution running on a next-gen PowerBook?
I don't know how this thing wants to get fed, which is I'm asking the question.
Originally posted by Existence
Cell = PPC980
I hope so, i relly really hope so
PPC980 = Multi-core Power5 + PowerPC
Originally posted by Programmer
Cell has a Power core inboard
Sorry, Power5 or PowerPC one?
I got utterly confused after reading a dozen of articles and a hundred of posts on various forums.
Posted by Ideaphile on Slashdot
I was at the Cell event today, and quoted in some of the news stories. I also have the ISSCC technical papers.
The PowerPC core in the Cell prototype chip is NOT a Power5, as speculated here. According to IBM, this core was designed from scratch for this application. One critical difference is that the new pipeline executes instructions in strict program order rather than reordering instructions to improve throughput as is done with Power5.
Also, IBM has not described the core as "simultaneous multithreaded", just "multithreaded." I presume from this that the multithreading is coarse-grained-- only one thread is active at a time, unlike Power5 which can execute instructions from two different threads in the same cycle.
The logic design for the Cell CPU was optimized for higher clock speeds in a given process than Power5 can achieve. This is a good tradeoff for more linear multimedia algorithms, but reduces effective throughput on other types of code.
I think it's reasonable to suppose that if Apple were interested in using the Cell architecture, it would prefer to use a version of the design that includes a Power5 core in place of the one in the Cell prototype.
This answers my own question pretty well. And the guy seems to be right.
Originally posted by costique
Sorry, Power5 or PowerPC one?
I got utterly confused after reading a dozen of articles and a hundred of posts on various forums.
It is not a POWER5. It is a dual-issue, in-order core running at 4+ GHz. At that clock rate you can make up for a lot of the limitations of the core... its probably in the same performance ballpark as a 1.8-2.0 GHz 970 on non-optimized code. On optimized code it'll fly.
The other thing that they are saying about the Cell is that it is highly modular. Basically it is centered around this EIB on-chip bus and everything is a peer on this bus. If you have a different Power core then why not stick on this bus interface and make it a peer on the EIB. Or put two of them there. The main obstacle is probably that this chip is running at 4 GHz!
IBM is also talking up its p5 575 box that is designed to be part of a server cluster. The system takes up just 2U of rack space but can hold 8 single-core Power5 chips each running at 1.9GHz.
From The Register.
It seems that IBM may well be tweaking things to allow for cooler running of all it's Power family?
[Edited for spelling]
Originally posted by DaveLee
It seems that IBM may well be tweaking things to allow for cooler running of all it's Power family?
One of the Power5's big features was reduced power consumption. At a very low-level in the basic circuit design of the chip they did some things which optimize power consumption. The features are part of what this new core has inherited, which is why there is any hope of this Cell chip fitting in a game console package at anything close to 4 GHz.
Another nagging thought keeps entering my brain. A few months ago some one on these boards was touting Apple's next generation lap top chip to be some agglomeration of 440 style chips using low low power. In looking at information on Cell, the SPE's(I think this is the right term) are actually somewhat independent specialized cpu's??? Is it even remotely possible that this person's posts about multiple 440's on die for a low power lap top chip had any validity whatsoever, or at least were spawned from the Cell??
With a lot of people including me dismissing his posts as anything from dubious to out and out BS, wouldn't it be a real kick in the pants if it was Cell he was describing and it does end up in an Apple laptop in some form.
Please forgive me if I'm totally wrong as I really have no clue about this technology.
Anyone remember the poster's name?
Originally posted by rickag
Anyone remember the poster's name?
It was Nr9 I think. And this one the original discussion.
Originally posted by Krassy
IF they will use the Cell processor we won't see it before 2006. but i could imagine that IBM and apple were working on the G5-successor (which - of course - could be based on a implementation-variant of the new Cell-design-ideas). i expect IBM to be very tight-lipped on new CPUs made for apple and bet we will all be surprised when the G5s successor will be presented...
A thought hit me as I read your post. Sony will use the Cell in the PS3. IBM will produce a workstation based on the Cell. Maybe these are two different chips? IBM and Apple may use a variant of Cell for high end computers that is different from the one Sony will use in the PS3.