You know what this reminds me of? Microsoft's Smart Displays. They were too expensive and dog slow when released, but were basically PocketPCs with huge displays that ran as terminals to your CPU over WiFi.
If Apple has solved the too slow/too expensive issue (or more likely just the passage of time), then I could see this as a great product, your on the couch web surfing solution, iTunes remote control, email viewer, etc.
It just has to be like a $250-$500 unit to really get decent market acceptance I think, not the $1,200 "accessory" MS land was peddling.
But most importantly, as essentially a dumb terminal, or some minor variation thereof, it could essentially use any CPU architecture, and really the XScale CPU line from Intel/ARM would be ideal.
Oh, and I discount the "Virtualization" CPU speculation, since after all Apple wouldn't voluntarily take a 20% performance hit just to escape a processor architecture that isn't ideal for peroformance anymore. Just seems to me to be self defeating logic, no strategic gain there.
Oh, and I discount the "Virtualization" CPU speculation, since after all Apple wouldn't voluntarily take a 20% performance hit just to escape a processor architecture that isn't ideal for peroformance anymore. Just seems to me to be self defeating logic, no strategic gain there.
Don't discount something you don't know much about.
I will, and happily. I know enough to understand that historically hardware with this level of OS abstraction takes massive efficiency hits. I consider speculation to the contrary wishful thinking at best, and unprintable stupidity at worst.
To consider it an option for pro level machines falls deep into the worst category.
It's a totally different chip. It requires a very different programming model.
I can't see Apple being able to utilize this chip without some rather large changes to the Mach core which is not designed to operate with a chip of this type.
Remember this chip uses a much simplified PPC chip as a controller, though it does some calculations of it's own. There are no out of order execution units in it.
I can see Apple using it for a rendering engine.
Find me a G4 that has out of order execution, none do. The MS chip certainly isn't that vastly different from what is already out there except it might have a decent chunk of logic cut out. If the software is already multithreaded it'd run quite well on it I expect compared to a G4. Even Cell isn't that vastly different from current technology, although software would take a bit more tweaking on it and it'd either need multiple PPEs or a beefed up one or it might suffer a bit on some day to day tasks.
Find me a G4 that has out of order execution, none do.
All G4s from the 7450 forward have some out of order execution, limited to the integer units. This makes sense, since the 745x series has four(!) of them.
Quote:
The MS chip certainly isn't that vastly different from what is already out there except it might have a decent chunk of logic cut out.
Define "vastly." A 21+-stage, in-order, SMT, three-core CPU is significantly different from anything else out there. The small L2 cache and tweaked AltiVec are probably console-specific design decisions, but neither is as significant as the design of the core itself, never mind the decision to put three of them together.
Getting things to run on it might not be so hard, but you'd suffer from all the AltiVec code breaking (in all likelihood, since the easiest way to deal with the incompatibilities is to conditionally compile all the AV code out and revert to generic PPC code). Optimization could be tricky, too, because Xenon is like and unlike both the G4+ and the 970.
Quote:
If the software is already multithreaded it'd run quite well on it I expect compared to a G4. Even Cell isn't that vastly different from current technology, although software would take a bit more tweaking on it and it'd either need multiple PPEs or a beefed up one or it might suffer a bit on some day to day tasks.
You're welcome to look for one other design that has a deep, narrow, in-order, SMT core with a full VMX unit hooked to eight SIMD units of a via an on-chip token ring bus.
It's not so new in the sense that Cell is exactly the sort of weird hardware that you'd expect to find in a Sony console. The difference is that IBM has plans for the architecture well beyond the PS3.
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
All G4s from the 7450 forward have some out of order execution, limited to the integer units. This makes sense, since the 745x series has four(!) of them.
Define "vastly." A 21+-stage, in-order, SMT, three-core CPU is significantly different from anything else out there. The small L2 cache and tweaked AltiVec are probably console-specific design decisions, but neither is as significant as the design of the core itself, never mind the decision to put three of them together.
Getting things to run on it might not be so hard, but you'd suffer from all the AltiVec code breaking (in all likelihood, since the easiest way to deal with the incompatibilities is to conditionally compile all the AV code out and revert to generic PPC code). Optimization could be tricky, too, because Xenon is like and unlike both the G4+ and the 970.
You're welcome to look for one other design that has a deep, narrow, in-order, SMT core with a full VMX unit hooked to eight SIMD units of a via an on-chip token ring bus.
It's not so new in the sense that Cell is exactly the sort of weird hardware that you'd expect to find in a Sony console. The difference is that IBM has plans for the architecture well beyond the PS3.
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
Yes.
Thanks for the support on this one. I couldn't get back earlier (family, (sigh)).
All G4s from the 7450 forward have some out of order execution, limited to the integer units. This makes sense, since the 745x series has four(!) of them.
Although that is true it is very limited. Add to that 3 of those integer units are very simple integer units. They serve their purpose but they are somewhat specific in their design.
Quote:
Define "vastly." A 21+-stage, in-order, SMT, three-core CPU is significantly different from anything else out there. The small L2 cache and tweaked AltiVec are probably console-specific design decisions, but neither is as significant as the design of the core itself, never mind the decision to put three of them together.
Getting things to run on it might not be so hard, but you'd suffer from all the AltiVec code breaking (in all likelihood, since the easiest way to deal with the incompatibilities is to conditionally compile all the AV code out and revert to generic PPC code). Optimization could be tricky, too, because Xenon is like and unlike both the G4+ and the 970.
Current multithreaded programs would run quite happily on MS' design after a recompile. There is nothing particularly spectacular about their core though. They are very stripped down cores aimed at streamed processing in an environment where there is relatively tight control on the code. That said they would handle general purpose code about as well as the G4+. There is nothing startingly new or different about anything MS has put out except they packed 3 cores together, which in and of itself isn't particularly special at this stage.
The debate over processor stages and in order and out of order design is somewhat moot given at the end of the day it is designed to do a specific task and decent frameworks and compilers will hide most of that unless you want to hand optimise it.
Optimisation quirks are there for any new processor design so provide a weak argument. Simple fact is current software could run quite happily on it following a recompile to better utilise its design. It'd quite likely run many programs very happily without one too with a small system update.
Quote:
You're welcome to look for one other design that has a deep, narrow, in-order, SMT core with a full VMX unit hooked to eight SIMD units of a via an on-chip token ring bus.
It's not so new in the sense that Cell is exactly the sort of weird hardware that you'd expect to find in a Sony console. The difference is that IBM has plans for the architecture well beyond the PS3.
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
Cell is unique for the SPEs but again the PPE isn't new technology. Just a redesign of what was already out there. I would be very surprised to ever see Apple adopt cell. It'd be absolutely atrocious in anything outside of multimedia and rendering in its current state. For a general CPU MS has the right idea and in fact I expect they chose that route precisely for that reason, so they can run more general apps on their console rather than focus on multimedia. In the end I doubt they will ever use either since they both produce too much heat.
As an aside the idea of specific silicon to speed up certain processes is hardly new. Taking it on chip was a new slant but is the sort of thing you'd expect in tvs, consoles, set top boxes and other areas where cell might appear. General purpose computers have tended to use entirely separate add in cards.
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
I suspect they wrote all those frameworks for two reasons:
1. Easy for developers to simply call the function when needed.
Although that is true it is very limited. Add to that 3 of those integer units are very simple integer units. They serve their purpose but they are somewhat specific in their design.
Current multithreaded programs would run quite happily on MS' design after a recompile. There is nothing particularly spectacular about their core though. They are very stripped down cores aimed at streamed processing in an environment where there is relatively tight control on the code. That said they would handle general purpose code about as well as the G4+. There is nothing startingly new or different about anything MS has put out except they packed 3 cores together, which in and of itself isn't particularly special at this stage.
The debate over processor stages and in order and out of order design is somewhat moot given at the end of the day it is designed to do a specific task and decent frameworks and compilers will hide most of that unless you want to hand optimise it.
Optimisation quirks are there for any new processor design so provide a weak argument. Simple fact is current software could run quite happily on it following a recompile to better utilise its design. It'd quite likely run many programs very happily without one too with a small system update.
Cell is unique for the SPEs but again the PPE isn't new technology. Just a redesign of what was already out there. I would be very surprised to ever see Apple adopt cell. It'd be absolutely atrocious in anything outside of multimedia and rendering in its current state. For a general CPU MS has the right idea and in fact I expect they chose that route precisely for that reason, so they can run more general apps on their console rather than focus on multimedia. In the end I doubt they will ever use either since they both produce too much heat.
As an aside the idea of specific silicon to speed up certain processes is hardly new. Taking it on chip was a new slant but is the sort of thing you'd expect in tvs, consoles, set top boxes and other areas where cell might appear. General purpose computers have tended to use entirely separate add in cards.
What you're forgetting here is that the OS is more and more being optimised for the G5, as are the programs. There are a fair number of instructions that are not supported in the G5 that have been important performance enhancers in the G3-4's. The opposite is also true.
Not supporting out of order execution is not as trivial as you assume. While it's true that an intensively hand-tuned re-compile would allow the OS and the programs for it to run on the chip, they would be far from optimal.
The performance hits would be enough to derail any advantage gained from there being three cores running at 3.2Ghz.
Also, while there must be some reason for IBM and MS using three cores for the XBox, it is sub-optimal for general computing use. Three cores would give only a 35% or so boost to a two core system. Using two dual core chips would give a 70% boost.
Any Mac using this chip could expect to se a significant loss in power from the current dual 2.7 G5. It would be a pointless exercise.
Discussing G4 chips isn't relevent at this point for desktop usage, and we all know how Apple is trying to leave the G4's behind in the portable space as well. With no evidence that these chips consume a level of power that would be suitable for a portable, we should let this idea rest in peace.
Current multithreaded programs would run quite happily on MS' design after a recompile. There is nothing particularly spectacular about their core though.
It's not that the cores are "spectacular," whatever that means. It's that the design decisions they made are pretty novel. The trend had been toward wide cores with massive scheduling flexibility, and now this.
Quote:
That said they would handle general purpose code about as well as the G4+.
Two major differences: 1) The G4 has lower latency access to RAM (better for lots of random access), while Cell has (significantly) higher bandwidth (better for streaming). 2) The G4 recovers quickly from a failed branch prediction, because of its 7-stage pipeline. The same scenario will make Cell, with its over-20-stage pipeline, unhappy. So code with lots of conditional execution will probably run better on the G4 (which can currently outrun the 970 on such code). Of course, Cell will crush the G4 on floating point, any kind of filtering or blitting code, etc. And that's before you bring the SPEs into the equation.
Once you bring the SPEs into the equation, Cell gets a whole huge gob of performance, but the PPE spends more of its time managing the SPEs (the simple integer units on the G4 have a roughly analogous role, freeing the complex unit to focus on actual work while they handle the simple arithmetic that frequently accompanies FP and AltiVec code).
Quote:
The debate over processor stages and in order and out of order design is somewhat moot given at the end of the day it is designed to do a specific task and decent frameworks and compilers will hide most of that unless you want to hand optimise it.
It becomes an issue once your code is running on a G4 and a G4+ and a 970 and Cell—in single and dual and single-threaded and SMT configurations. Apple is trying to make it so that most of those issues will be handled by their frameworks, but at the end of the day, Cell on Macs will leave developers (Apple's and others') with three discrete performance profiles to target. Every new profile doubles the amount of testing necessary.
Quote:
Cell is unique for the SPEs but again the PPE isn't new technology. Just a redesign of what was already out there.
No, Cell and Xenon are the first products in a discrete development line. They do not inherit from the 400, 700 or 900 series, and their philosophy is different from any of those three lines. I'm not sure what qualifications you require for "difference," but I suspect that it would be something alien enough that it couldn't implement POWER with anything like native performance. Obviously, the need to implement POWER constrains them from coming up with something really freaky.
Quote:
I would be very surprised to ever see Apple adopt cell. It'd be absolutely atrocious in anything outside of multimedia and rendering in its current state. For a general CPU MS has the right idea and in fact I expect they chose that route precisely for that reason, so they can run more general apps on their console rather than focus on multimedia. In the end I doubt they will ever use either since they both produce too much heat.
I'd be surprised if either produces much more heat than the first 2.5GHz 970s. Even if they do, the Power Mac case is big and overdesigned enough that it had damn well better be able to a accommodate a few more watts.
Cell is automatically more attractive than Xenon simply for having standard AltiVec.
It's more attractive, too, because it's not a fixed design. That token ring bus is there for a reason. Sony asked for one PPE and 8 SPEs. Apple can ask for something else. I imagine they'd be keenly interested in something that boosted multimedia performance as much as the SPEs can (Apple coined the term "multimedia," after all, and they just hitched their wagon to the computationally intensive combination of H.264 and HD).
Quote:
As an aside the idea of specific silicon to speed up certain processes is hardly new. Taking it on chip was a new slant but is the sort of thing you'd expect in tvs, consoles, set top boxes and other areas where cell might appear. General purpose computers have tended to use entirely separate add in cards.
SoC designs are nothing new, but high-end SoC implementations are. Most SoC designs are adopted for cost reasons.
The main advantage to bringing the SPEs on chip is the same as the advantage of bringing AltiVec on chip: If you add something new that programmers have to be aware of, it had better be common or (preferably) standard, or it will be ignored. The old AV series Macs had a separate DSP chip. Hardly anything used it. AltiVec adoption has just started to really take off in the last year. (That's why it's so important that the implementation not change much from chip to chip—if it's not a consistent and well-supported platform, developers will stick to the general-purpose code that they know works.) If Apple does use Cell with the SPEs, they'll be making a commitment. But that commitment will pay off in the form of massive performance in exactly the sort of areas where Apple is strongest. The fallback—generic PPC code running on the PPE—won't be so bad, either, although as with all implementations there will be edge cases where it doesn't perform so well.
They did this solely because of power issues. The PCI connection was not designed to deliver huge power to a 30 inch LCD - plane and simple.
Someone had their triple espresso this morning. I think you had better worry more about your own grudge, it might lower your blood pressure and save your life.
Let the engineers do their job. Companies rarely ever over promise in a public way on purpose and never deliver. It depresses stock values way to fast - just not smart business.
So what's your point? That I need meds? Your second point only validates mine. As far as the DVI switch - source your comments, please.
Very good, thanks murch! Using Cell chips in laptops and in the PM looks to me like it would be a good thing, even if integer performance on old apps was reduced a little. We get higher speed, better performance for new apps, esp. if the CoreX apps run on SPEs, and cooler, lower-power operation. What's not to like?
Very good, thanks murch! Using Cell chips in laptops and in the PM looks to me like it would be a good thing, even if integer performance on old apps was reduced a little. We get higher speed, better performance for new apps, esp. if the CoreX apps run on SPEs, and cooler, lower-power operation. What's not to like?
yeah after going through the slides a little now i know why hmurchinson is on about Cell so much
my bet is that it's not an either or, what apple is either going to announce in a few weeks, or sort out by end of the year, is a strategy where the next-gen G5 is kind of some kind of hybrid between G5 and Cell, or a dual-coreG5 with cell co-processor (i loathe to use this example but kinda like Cell to the G5 what the MMX was to the Pentium2)
i think apple and ibm have been working very very hard to make sure the Mac pipeline looks solid for the next few years... i have faith. WWDC or Paris or Macworld, within the next 6-8 months, Apple's weakest link (CPU "attractiveness") should be shored up.
at least $50 by the end of the year
pity i got no money for shares... but i do have a part-time job now
teaching flash
"now kids, pointless long flash intros with no skip button is bad... mmmmkay?...."
Comments
If Apple has solved the too slow/too expensive issue (or more likely just the passage of time), then I could see this as a great product, your on the couch web surfing solution, iTunes remote control, email viewer, etc.
It just has to be like a $250-$500 unit to really get decent market acceptance I think, not the $1,200 "accessory" MS land was peddling.
But most importantly, as essentially a dumb terminal, or some minor variation thereof, it could essentially use any CPU architecture, and really the XScale CPU line from Intel/ARM would be ideal.
Originally posted by ChevalierMalFet
Oh, and I discount the "Virtualization" CPU speculation, since after all Apple wouldn't voluntarily take a 20% performance hit just to escape a processor architecture that isn't ideal for peroformance anymore. Just seems to me to be self defeating logic, no strategic gain there.
Don't discount something you don't know much about.
To consider it an option for pro level machines falls deep into the worst category.
Originally posted by melgross
It's a totally different chip. It requires a very different programming model.
I can't see Apple being able to utilize this chip without some rather large changes to the Mach core which is not designed to operate with a chip of this type.
Remember this chip uses a much simplified PPC chip as a controller, though it does some calculations of it's own. There are no out of order execution units in it.
I can see Apple using it for a rendering engine.
Find me a G4 that has out of order execution, none do. The MS chip certainly isn't that vastly different from what is already out there except it might have a decent chunk of logic cut out. If the software is already multithreaded it'd run quite well on it I expect compared to a G4. Even Cell isn't that vastly different from current technology, although software would take a bit more tweaking on it and it'd either need multiple PPEs or a beefed up one or it might suffer a bit on some day to day tasks.
Originally posted by Telomar
Find me a G4 that has out of order execution, none do.
All G4s from the 7450 forward have some out of order execution, limited to the integer units. This makes sense, since the 745x series has four(!) of them.
The MS chip certainly isn't that vastly different from what is already out there except it might have a decent chunk of logic cut out.
Define "vastly." A 21+-stage, in-order, SMT, three-core CPU is significantly different from anything else out there. The small L2 cache and tweaked AltiVec are probably console-specific design decisions, but neither is as significant as the design of the core itself, never mind the decision to put three of them together.
Getting things to run on it might not be so hard, but you'd suffer from all the AltiVec code breaking (in all likelihood, since the easiest way to deal with the incompatibilities is to conditionally compile all the AV code out and revert to generic PPC code). Optimization could be tricky, too, because Xenon is like and unlike both the G4+ and the 970.
If the software is already multithreaded it'd run quite well on it I expect compared to a G4. Even Cell isn't that vastly different from current technology, although software would take a bit more tweaking on it and it'd either need multiple PPEs or a beefed up one or it might suffer a bit on some day to day tasks.
You're welcome to look for one other design that has a deep, narrow, in-order, SMT core with a full VMX unit hooked to eight SIMD units of a via an on-chip token ring bus.
It's not so new in the sense that Cell is exactly the sort of weird hardware that you'd expect to find in a Sony console. The difference is that IBM has plans for the architecture well beyond the PS3.
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
Originally posted by JLL
And on a desktop system (which has an external display) it's right there without pressing anything.
true. i found it a while back and messed around with it. sorta interesting but not that useful for me.
Originally posted by Amorph
All G4s from the 7450 forward have some out of order execution, limited to the integer units. This makes sense, since the 745x series has four(!) of them.
Define "vastly." A 21+-stage, in-order, SMT, three-core CPU is significantly different from anything else out there. The small L2 cache and tweaked AltiVec are probably console-specific design decisions, but neither is as significant as the design of the core itself, never mind the decision to put three of them together.
Getting things to run on it might not be so hard, but you'd suffer from all the AltiVec code breaking (in all likelihood, since the easiest way to deal with the incompatibilities is to conditionally compile all the AV code out and revert to generic PPC code). Optimization could be tricky, too, because Xenon is like and unlike both the G4+ and the 970.
You're welcome to look for one other design that has a deep, narrow, in-order, SMT core with a full VMX unit hooked to eight SIMD units of a via an on-chip token ring bus.
It's not so new in the sense that Cell is exactly the sort of weird hardware that you'd expect to find in a Sony console. The difference is that IBM has plans for the architecture well beyond the PS3.
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
Yes.
Thanks for the support on this one. I couldn't get back earlier (family, (sigh)).
Originally posted by Amorph
All G4s from the 7450 forward have some out of order execution, limited to the integer units. This makes sense, since the 745x series has four(!) of them.
Although that is true it is very limited. Add to that 3 of those integer units are very simple integer units. They serve their purpose but they are somewhat specific in their design.
Define "vastly." A 21+-stage, in-order, SMT, three-core CPU is significantly different from anything else out there. The small L2 cache and tweaked AltiVec are probably console-specific design decisions, but neither is as significant as the design of the core itself, never mind the decision to put three of them together.
Getting things to run on it might not be so hard, but you'd suffer from all the AltiVec code breaking (in all likelihood, since the easiest way to deal with the incompatibilities is to conditionally compile all the AV code out and revert to generic PPC code). Optimization could be tricky, too, because Xenon is like and unlike both the G4+ and the 970.
Current multithreaded programs would run quite happily on MS' design after a recompile. There is nothing particularly spectacular about their core though. They are very stripped down cores aimed at streamed processing in an environment where there is relatively tight control on the code. That said they would handle general purpose code about as well as the G4+. There is nothing startingly new or different about anything MS has put out except they packed 3 cores together, which in and of itself isn't particularly special at this stage.
The debate over processor stages and in order and out of order design is somewhat moot given at the end of the day it is designed to do a specific task and decent frameworks and compilers will hide most of that unless you want to hand optimise it.
Optimisation quirks are there for any new processor design so provide a weak argument. Simple fact is current software could run quite happily on it following a recompile to better utilise its design. It'd quite likely run many programs very happily without one too with a small system update.
You're welcome to look for one other design that has a deep, narrow, in-order, SMT core with a full VMX unit hooked to eight SIMD units of a via an on-chip token ring bus.
It's not so new in the sense that Cell is exactly the sort of weird hardware that you'd expect to find in a Sony console. The difference is that IBM has plans for the architecture well beyond the PS3.
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
Cell is unique for the SPEs but again the PPE isn't new technology. Just a redesign of what was already out there. I would be very surprised to ever see Apple adopt cell. It'd be absolutely atrocious in anything outside of multimedia and rendering in its current state. For a general CPU MS has the right idea and in fact I expect they chose that route precisely for that reason, so they can run more general apps on their console rather than focus on multimedia. In the end I doubt they will ever use either since they both produce too much heat.
As an aside the idea of specific silicon to speed up certain processes is hardly new. Taking it on chip was a new slant but is the sort of thing you'd expect in tvs, consoles, set top boxes and other areas where cell might appear. General purpose computers have tended to use entirely separate add in cards.
Originally posted by Amorph
Of the two, I'd expect Apple to adopt Cell first. The whole point of all those nifty frameworks is to let the system decide what to run where, so any application written to the frameworks as of Tiger could get significant support for the Cell architecture simply by virtue of calling into OpenGL and Core Image and Quartz and mathlib implementations that have been ported to Cell by Apple. That would be enough for most applications. Games could be ported from the PS3 codebase if necessary.
I suspect they wrote all those frameworks for two reasons:
1. Easy for developers to simply call the function when needed.
2. Portability
Originally posted by Telomar
Although that is true it is very limited. Add to that 3 of those integer units are very simple integer units. They serve their purpose but they are somewhat specific in their design.
Current multithreaded programs would run quite happily on MS' design after a recompile. There is nothing particularly spectacular about their core though. They are very stripped down cores aimed at streamed processing in an environment where there is relatively tight control on the code. That said they would handle general purpose code about as well as the G4+. There is nothing startingly new or different about anything MS has put out except they packed 3 cores together, which in and of itself isn't particularly special at this stage.
The debate over processor stages and in order and out of order design is somewhat moot given at the end of the day it is designed to do a specific task and decent frameworks and compilers will hide most of that unless you want to hand optimise it.
Optimisation quirks are there for any new processor design so provide a weak argument. Simple fact is current software could run quite happily on it following a recompile to better utilise its design. It'd quite likely run many programs very happily without one too with a small system update.
Cell is unique for the SPEs but again the PPE isn't new technology. Just a redesign of what was already out there. I would be very surprised to ever see Apple adopt cell. It'd be absolutely atrocious in anything outside of multimedia and rendering in its current state. For a general CPU MS has the right idea and in fact I expect they chose that route precisely for that reason, so they can run more general apps on their console rather than focus on multimedia. In the end I doubt they will ever use either since they both produce too much heat.
As an aside the idea of specific silicon to speed up certain processes is hardly new. Taking it on chip was a new slant but is the sort of thing you'd expect in tvs, consoles, set top boxes and other areas where cell might appear. General purpose computers have tended to use entirely separate add in cards.
What you're forgetting here is that the OS is more and more being optimised for the G5, as are the programs. There are a fair number of instructions that are not supported in the G5 that have been important performance enhancers in the G3-4's. The opposite is also true.
Not supporting out of order execution is not as trivial as you assume. While it's true that an intensively hand-tuned re-compile would allow the OS and the programs for it to run on the chip, they would be far from optimal.
The performance hits would be enough to derail any advantage gained from there being three cores running at 3.2Ghz.
Also, while there must be some reason for IBM and MS using three cores for the XBox, it is sub-optimal for general computing use. Three cores would give only a 35% or so boost to a two core system. Using two dual core chips would give a 70% boost.
Any Mac using this chip could expect to se a significant loss in power from the current dual 2.7 G5. It would be a pointless exercise.
Discussing G4 chips isn't relevent at this point for desktop usage, and we all know how Apple is trying to leave the G4's behind in the portable space as well. With no evidence that these chips consume a level of power that would be suitable for a portable, we should let this idea rest in peace.
Originally posted by Telomar
Current multithreaded programs would run quite happily on MS' design after a recompile. There is nothing particularly spectacular about their core though.
It's not that the cores are "spectacular," whatever that means. It's that the design decisions they made are pretty novel. The trend had been toward wide cores with massive scheduling flexibility, and now this.
That said they would handle general purpose code about as well as the G4+.
Two major differences: 1) The G4 has lower latency access to RAM (better for lots of random access), while Cell has (significantly) higher bandwidth (better for streaming). 2) The G4 recovers quickly from a failed branch prediction, because of its 7-stage pipeline. The same scenario will make Cell, with its over-20-stage pipeline, unhappy. So code with lots of conditional execution will probably run better on the G4 (which can currently outrun the 970 on such code). Of course, Cell will crush the G4 on floating point, any kind of filtering or blitting code, etc. And that's before you bring the SPEs into the equation.
Once you bring the SPEs into the equation, Cell gets a whole huge gob of performance, but the PPE spends more of its time managing the SPEs (the simple integer units on the G4 have a roughly analogous role, freeing the complex unit to focus on actual work while they handle the simple arithmetic that frequently accompanies FP and AltiVec code).
The debate over processor stages and in order and out of order design is somewhat moot given at the end of the day it is designed to do a specific task and decent frameworks and compilers will hide most of that unless you want to hand optimise it.
It becomes an issue once your code is running on a G4 and a G4+ and a 970 and Cell—in single and dual and single-threaded and SMT configurations. Apple is trying to make it so that most of those issues will be handled by their frameworks, but at the end of the day, Cell on Macs will leave developers (Apple's and others') with three discrete performance profiles to target. Every new profile doubles the amount of testing necessary.
Cell is unique for the SPEs but again the PPE isn't new technology. Just a redesign of what was already out there.
No, Cell and Xenon are the first products in a discrete development line. They do not inherit from the 400, 700 or 900 series, and their philosophy is different from any of those three lines. I'm not sure what qualifications you require for "difference," but I suspect that it would be something alien enough that it couldn't implement POWER with anything like native performance. Obviously, the need to implement POWER constrains them from coming up with something really freaky.
I would be very surprised to ever see Apple adopt cell. It'd be absolutely atrocious in anything outside of multimedia and rendering in its current state. For a general CPU MS has the right idea and in fact I expect they chose that route precisely for that reason, so they can run more general apps on their console rather than focus on multimedia. In the end I doubt they will ever use either since they both produce too much heat.
I'd be surprised if either produces much more heat than the first 2.5GHz 970s. Even if they do, the Power Mac case is big and overdesigned enough that it had damn well better be able to a accommodate a few more watts.
Cell is automatically more attractive than Xenon simply for having standard AltiVec.
It's more attractive, too, because it's not a fixed design. That token ring bus is there for a reason. Sony asked for one PPE and 8 SPEs. Apple can ask for something else. I imagine they'd be keenly interested in something that boosted multimedia performance as much as the SPEs can (Apple coined the term "multimedia," after all, and they just hitched their wagon to the computationally intensive combination of H.264 and HD).
As an aside the idea of specific silicon to speed up certain processes is hardly new. Taking it on chip was a new slant but is the sort of thing you'd expect in tvs, consoles, set top boxes and other areas where cell might appear. General purpose computers have tended to use entirely separate add in cards.
SoC designs are nothing new, but high-end SoC implementations are. Most SoC designs are adopted for cost reasons.
The main advantage to bringing the SPEs on chip is the same as the advantage of bringing AltiVec on chip: If you add something new that programmers have to be aware of, it had better be common or (preferably) standard, or it will be ignored. The old AV series Macs had a separate DSP chip. Hardly anything used it. AltiVec adoption has just started to really take off in the last year. (That's why it's so important that the implementation not change much from chip to chip—if it's not a consistent and well-supported platform, developers will stick to the general-purpose code that they know works.) If Apple does use Cell with the SPEs, they'll be making a commitment. But that commitment will pay off in the form of massive performance in exactly the sort of areas where Apple is strongest. The fallback—generic PPC code running on the PPE—won't be so bad, either, although as with all implementations there will be edge cases where it doesn't perform so well.
Originally posted by Outsider
Question # 345
Which one of these does not belong?
Xbox moves from x86 to PowerPC.
Playstation moves from proprietary to PowerPC.
Gamecube goes from embedded PowerPC to PowerPC.
Apple goes from PowerPC to Intel.
The first one, because it is the only one which is true.
Playstation is using a NEC MIPS processor, not proprietary.
Originally posted by msconvert
They did this solely because of power issues. The PCI connection was not designed to deliver huge power to a 30 inch LCD - plane and simple.
Someone had their triple espresso this morning. I think you had better worry more about your own grudge, it might lower your blood pressure and save your life.
Let the engineers do their job. Companies rarely ever over promise in a public way on purpose and never deliver. It depresses stock values way to fast - just not smart business.
So what's your point? That I need meds? Your second point only validates mine. As far as the DVI switch - source your comments, please.
Nice slides of the cell architecture.
talks it creates a great deal of
BUZZ and what not. But take it as it is
Buzz.
Nothing will change the beloved platform.
Anyone?
Originally posted by Vox Barbara
If apple and intel have their casual
talks it creates a great deal of
BUZZ and what not. But take it as it is
Buzz.
Nothing will change the beloved platform.
Anyone?
voxy speaks!! welcome back
Originally posted by hmurchison
http://www.research.scea.com/researc...lGDC05/02.html
Nice slides of the cell architecture.
Very good, thanks murch! Using Cell chips in laptops and in the PM looks to me like it would be a good thing, even if integer performance on old apps was reduced a little. We get higher speed, better performance for new apps, esp. if the CoreX apps run on SPEs, and cooler, lower-power operation. What's not to like?
Originally posted by cubist
Very good, thanks murch! Using Cell chips in laptops and in the PM looks to me like it would be a good thing, even if integer performance on old apps was reduced a little. We get higher speed, better performance for new apps, esp. if the CoreX apps run on SPEs, and cooler, lower-power operation. What's not to like?
yeah after going through the slides a little now i know why hmurchinson is on about Cell so much
my bet is that it's not an either or, what apple is either going to announce in a few weeks, or sort out by end of the year, is a strategy where the next-gen G5 is kind of some kind of hybrid between G5 and Cell, or a dual-coreG5 with cell co-processor (i loathe to use this example but kinda like Cell to the G5 what the MMX was to the Pentium2)
i think apple and ibm have been working very very hard to make sure the Mac pipeline looks solid for the next few years... i have faith. WWDC or Paris or Macworld, within the next 6-8 months, Apple's weakest link (CPU "attractiveness") should be shored up.
at least $50 by the end of the year
pity i got no money for shares... but i do have a part-time job now
teaching flash
"now kids, pointless long flash intros with no skip button is bad... mmmmkay?...."
Originally posted by Vox Barbara
If apple and intel have their casual
talks it creates a great deal of
BUZZ and what not. But take it as it is
Buzz.
Nothing will change the beloved platform.
Anyone?
voxy you on Tiger yet? a few kinks, but time to fall in love all over again