Sounds more like a tech-geeks wet dream, than a consumer priced living-room appliance...
Not sure what you are saying. The XBox2 will definately be PPC based. Whether it has 3 of the things is more of the question. And it's the ATI card that will be doing all of the Direct X stuff anyway.
True but the original Xbox cost almost $400 to make and still sold for under $200. MS told its shareholders that it expected to lose a billion dollars...that's billion with a 'b'...in the first two years. Of course their plan is to get people off other gaming platforms and dominate that market the same way they do with PCs. Only MS has the resources -- and the complete lack of ethics -- to do this.
There strategy has a big hole in it. When xBox was released there was 20Million PS/2's out there. Now there are 13 Million Xbox'x with demand decelerating and 70 Million PS/2's. I think MSFT's reasons are more covert. They are working on their own PC. They are adopting the Apple model, as they have exhausted the licensing fee strategy. The only way for MSFT to succeed in this endeavor is to make the next MSFT OS incompatible with x86 hardware and develop their own proprietary system.
It is impossible for MSFT to innovate while having to support millions of potential combinations of hardware and software. They see how fast Apple is innovating and moving OS X and they know every year that goes by is another 2 years Apple adds to its already four year lead it has over Windows.
And what or where does Apple have this lead in? General market share? Gaming console market share? I'm afraid you're comparing apples, oranges and badgers. (OS X, Windows, PS2)
Any concept that Microsoft is doomed is purely delusional. Microsoft doesn't have to innovate to "win," just make an overall profit every quarter (but that's the trick and the trap Microsoft is in -- but that's another thread). Oh sure, the Xbox endeavor might fail, but not soon and not without sapping Sony's market share and crushing all others.
As for the "Microsoft PC?" Right now, Microsoft is profiting on Windows sales by a greater factor than they are losing on Xbox sales.
Microsoft does not need, (or most likely) want their own hardware to run Windows on. They didn't become the most wealthy company on the planet selling hardware. I think Microsoft is just having fun with some of it's money while trying to make products that people are asking for. I'm sure they would like to turn a profit, and doubt they will ever fold the XBOX no matter what happens, but I certainly don't see them trying to do their own hardware, and trying to kick Apple off the it's little niche.
Apple finally has something very profitable for them with iTunes, and the iPod, and will probably get up to %10 market share with Mac OS, and Macintosh within 10 years, but I don't see it as a war between them any more.
I don't think MS is trying to push Sony out either. I think with all the games MS has made over the years it's just didn't seem like a total mislead venture to them. Gaming is big business anyway, and having a piece of that in the future is a wise decision for a company of their size that can't really do much more with software than they already have. They need to diversify, but making their own hardware for the windows platform would be a mistake, and I'm sure they know that.
The original Mercury article stated Microsoft chose to reduce costs by eschewing a hard disk. How much sense does that make, when, in the same article, they say the company will include three ultra-expensive PPC's in the same unit, especially since 10G hard disks will be manufactured at the cost of somewhere in the range of $10-20 in 2005. The article is pretty damn inconsistent in that regard.
Microsoft's vision for the Xbox has always been for it to be a closed unit ? that all of its components are internal and not removable. They don't want to make it a PC (they already own that market). Furthermore, the console's biggest strength this generation has actually been its strength ? and this is strength is not just from its CPU or GPU, but from its hard disk, and its ability to be used as a data spool for unlimited virtual memory, keeping games performing great. Xbox has also gotten tremendous support from the PC games community, because of the familiarity a hard disk provides (and also because of similar architectures of DirectX and X86, but the inclusion of the HD has made development much easier).
Let's forget for a second that the entire reason for the console's existence is Xbox Live, Microsoft's online service. How feasible is this service without a hard disk on which to download levels, modes, missions, characters, etc.? A lot of the functionality of even Xbox 1, such as the ability to use custom soundtracks ripped from game CD's, disappears when you remove the hard disk. It will be just one huge step backwards from the standard Microsoft set with their own console, four years earlier. It also makes backwards compatibility impossible.
Just how much good will two ? let alone three ? PPC's be to developers? For a development community that can't get over PS2's architectural hurdles after three years, do you think they'll invest in all these proprietary multiprocessing and advanced threading engines? You forget, there's no OS overhead like OS X to provide the automatic resource shifting present on a game console (IIRC, you can't even saturate two processors with the same process in Windows, the SMP code is so primitive), so all of that has to be in the developer's own code, and requires more time and money.
Sony's plan with CELL is a bit different from the traditional game console, since it will handle stuff traditionally done by the GPU. Everything on the PS3 will be done through software rendering, which makes for scalability in graphics prowess and performance. Theoretically, one can just plug in another CELL unit and boom, added graphical robustness (increased textures, poly-count, etc.). I assume Sony's API's will automatically make games adjust.
There's just so much in this article that doesn't make sense. It's so backwards from the vision from what people have expected. Between removing a hard disk and blowing backwards compatibility to hell, MS would be losing far more repeat customers than they'd be gaining with the lower price and misplaced performance.
Is there any technical obstacle for doing three way systems? Why should it always be about power of two? This rumored Xbox Next is a six way system.
And now to something completely different:
Comparison:
Two singe core processors fabbed at 130 nm, 1.3 V
vs.
Three dual core processors fabbed at 65 nm, .9 V
I we assume that the transistor count per core is the same and the cost of a processor is directly related to the size of the die, then the latter alternative would cost two thirds of the former.
The cost of the CPUs in Xbox Next would be considerable cheaper than those in current two way Power Macs.
If we assume that IBM manages to reduce wattage when moving to 65 nm, in the same ammount that they did when moving to 90 nm, then the latter alternative would draw a bit less power than the former.
The Xbox Next wouldn't need as much cooling as the current Power Mac, and they would be more heat resistant too.
If we assume that Microsoft will be requiring 40 times the ammount of processor cores as Apple use today (10 million Xbox Next, 6 cores in each = 60 million cores, compared to 1 million Power Macs with 1.5 cores each = 1,5 million) then Microsoft will be in a better spot for volume discount.
If we assume that MS will use the "slowest" version of a processor-family then we're talking about parts running at 3 GHz. These will be cheapest and draw the least ammount of power of all the processors in that generation comming of the production line.
970 is the smallest high end procesor out there today, less then half the size of a Prescott. If Intel can make a fat margin out of selling cheap Pentiums ($80?) then so can IBM. Why do we assume that these processors will cost a fortune for Microsoft? Is it because Apple demands $3000 for a complete two way Power Mac? Is there a way for Microsoft to make a Xbox Next with the suggested specs for let's say $500? I think so. Especially considering that they are developing just one machine, and then making tens of millions of identical boxes.
That's a lot to assume, but I think it's doable. With that said, I really don't think the report is accurate. Seems a bit to extreme for me. But, either way, Apple will be paying a lot less for processors in the future. That's good!
I agree with whomever suggested the leak mistook having two virtual processors a la hyperthreading (SMT) for a CPU being "dual-core."
That would make more sense, wouldn't it? Its also clearly silly to think that MS would go from 1 processor chip in the XBox to 3 in the XBox2 as a cost and heat reduction measure. Assuming for a moment that the rumour has some validity, surely that implies something about the one chip that they are going to use...
And as for having 3 processors, what is wrong with that? Why is everybody so fixated on having powers-of-2 processors? If your processor is, say, 100 million transistors and your chip can hold 300 million transistors (for optimal power/cost/heat/yield reasons) do you just waste a third of your space? That's like saying you can build a 2-car garage, or a 4-car garage, but not a 3-car garage.
This probably comes from processor word sizes always being a power of 2... but that is driven because that multiple needs to be quickly calcuable in hardware using shifts instead of multiplies. The number of processor cores is under no such restrictions.
That's a lot to assume, but I think it's doable. With that said, I really don't think the report is accurate. Seems a bit to extreme for me. But, either way, Apple will be paying a lot less for processors in the future. That's good!
Why do you think Apple will be paying less for processors? If MS is getting this much bang for the buck from IBM at the low price point, imagine how much bang Apple gets for more bucks? Apple's current price point gets them the margins they need to survive in the PC business, they cannot compete at the price points of mass produced fixed-design closed box consumer hardware. Completely different business model, and Microsoft has figured this out.
That would make more sense, wouldn't it? Its also clearly silly to think that MS would go from 1 processor chip in the XBox to 3 in the XBox2 as a cost and heat reduction measure. Assuming for a moment that the rumour has some validity, surely that implies something about the one chip that they are going to use...
Ah. So either a triple-threaded core(!!!) or three cores on one die, with space the size of a 4th core free for an ASIC. Then you have a system-on-a-chip design, which wins on cost efficiency.
Re: Multithreading. The games don't have to explicitly support SMP for this to be useful. If you have concurrency, you can have system functions (sound, networking) running concurrently, which means the game can just tell another thread to "do this" and go right back to rendering frames. It gives the system vendor the ability to provide something other than bare metal to code to, which means easier ports and easier development, without the system getting in the way of the game. And, of course, if a game does decide to use the concurrent processing capabilities, they have a whole world of power to tap into for physics, AI, pre-rendering for instant transitions between levels, etc.
Ah. So either a triple-threaded core(!!!) or three cores on one die, with space the size of a 4th core free for an ASIC. Then you have a system-on-a-chip design, which wins on cost efficiency.
And what space would that be? What 4th core?
Quote:
Re: Multithreading. The games don't have to explicitly support SMP for this to be useful. If you have concurrency, you can have system functions (sound, networking) running concurrently, which means the game can just tell another thread to "do this" and go right back to rendering frames. It gives the system vendor the ability to provide something other than bare metal to code to, which means easier ports and easier development, without the system getting in the way of the game. And, of course, if a game does decide to use the concurrent processing capabilities, they have a whole world of power to tap into for physics, AI, pre-rendering for instant transitions between levels, etc.
Games use around 99.5% of the available CPU, so you pretty much have to code the game to use it. Everything else that is expensive is in the GPU or DMA controllers.
Comments
Originally posted by rok
so basically what you're saying here is that a "mini-cluster" central computing hub could become as necessary as a water heater?
The 'mini-cluster' could actually become useful *as* a water heater.
Seems highly unlikely to me.
Sounds more like a tech-geeks wet dream, than a consumer priced living-room appliance...
Originally posted by G-News
direct X 10, ported onto a PPC platform?
Seems highly unlikely to me.
Sounds more like a tech-geeks wet dream, than a consumer priced living-room appliance...
Not sure what you are saying. The XBox2 will definately be PPC based. Whether it has 3 of the things is more of the question. And it's the ATI card that will be doing all of the Direct X stuff anyway.
Originally posted by onlooker
Those are total BS specs.
There are accuracies, and there are inaccuracies.
direct X 10, ported onto a PPC platform?
Seems highly unlikely to me.
Sounds more like a tech-geeks wet dream, than a consumer priced living-room appliance...
It doesn't matter how unlikely it seems to you, does it?
Originally posted by Programmer
There are accuracies, and there are inaccuracies.
It doesn't matter how unlikely it seems to you, does it?
Programmer,
You seem to be having a little fun tonight. I just came from ars and you had a post on this topic there too.
Originally posted by Michael Grey
True but the original Xbox cost almost $400 to make and still sold for under $200. MS told its shareholders that it expected to lose a billion dollars...that's billion with a 'b'...in the first two years. Of course their plan is to get people off other gaming platforms and dominate that market the same way they do with PCs. Only MS has the resources -- and the complete lack of ethics -- to do this.
There strategy has a big hole in it. When xBox was released there was 20Million PS/2's out there. Now there are 13 Million Xbox'x with demand decelerating and 70 Million PS/2's. I think MSFT's reasons are more covert. They are working on their own PC. They are adopting the Apple model, as they have exhausted the licensing fee strategy. The only way for MSFT to succeed in this endeavor is to make the next MSFT OS incompatible with x86 hardware and develop their own proprietary system.
It is impossible for MSFT to innovate while having to support millions of potential combinations of hardware and software. They see how fast Apple is innovating and moving OS X and they know every year that goes by is another 2 years Apple adds to its already four year lead it has over Windows.
Any concept that Microsoft is doomed is purely delusional. Microsoft doesn't have to innovate to "win," just make an overall profit every quarter (but that's the trick and the trap Microsoft is in -- but that's another thread). Oh sure, the Xbox endeavor might fail, but not soon and not without sapping Sony's market share and crushing all others.
As for the "Microsoft PC?"
Does. Not. Compute.
Screed
apples, oranges and badgers
badger badger badger badger...
http://www.companymen.net/stuff/badger.swf
edit damn link wont work for some reason, just copy, paste, and enjoy
Apple finally has something very profitable for them with iTunes, and the iPod, and will probably get up to %10 market share with Mac OS, and Macintosh within 10 years, but I don't see it as a war between them any more.
I don't think MS is trying to push Sony out either. I think with all the games MS has made over the years it's just didn't seem like a total mislead venture to them. Gaming is big business anyway, and having a piece of that in the future is a wise decision for a company of their size that can't really do much more with software than they already have. They need to diversify, but making their own hardware for the windows platform would be a mistake, and I'm sure they know that.
And it's the ATI card that will be doing all of the Direct X stuff anyway.
That's why we have DirectX support on the Mac for all machines with either an ATI or an nVidia card, right?
Ummm, google for DirectX...
Microsoft's vision for the Xbox has always been for it to be a closed unit ? that all of its components are internal and not removable. They don't want to make it a PC (they already own that market). Furthermore, the console's biggest strength this generation has actually been its strength ? and this is strength is not just from its CPU or GPU, but from its hard disk, and its ability to be used as a data spool for unlimited virtual memory, keeping games performing great. Xbox has also gotten tremendous support from the PC games community, because of the familiarity a hard disk provides (and also because of similar architectures of DirectX and X86, but the inclusion of the HD has made development much easier).
Let's forget for a second that the entire reason for the console's existence is Xbox Live, Microsoft's online service. How feasible is this service without a hard disk on which to download levels, modes, missions, characters, etc.? A lot of the functionality of even Xbox 1, such as the ability to use custom soundtracks ripped from game CD's, disappears when you remove the hard disk. It will be just one huge step backwards from the standard Microsoft set with their own console, four years earlier. It also makes backwards compatibility impossible.
Just how much good will two ? let alone three ? PPC's be to developers? For a development community that can't get over PS2's architectural hurdles after three years, do you think they'll invest in all these proprietary multiprocessing and advanced threading engines? You forget, there's no OS overhead like OS X to provide the automatic resource shifting present on a game console (IIRC, you can't even saturate two processors with the same process in Windows, the SMP code is so primitive), so all of that has to be in the developer's own code, and requires more time and money.
Sony's plan with CELL is a bit different from the traditional game console, since it will handle stuff traditionally done by the GPU. Everything on the PS3 will be done through software rendering, which makes for scalability in graphics prowess and performance. Theoretically, one can just plug in another CELL unit and boom, added graphical robustness (increased textures, poly-count, etc.). I assume Sony's API's will automatically make games adjust.
There's just so much in this article that doesn't make sense. It's so backwards from the vision from what people have expected. Between removing a hard disk and blowing backwards compatibility to hell, MS would be losing far more repeat customers than they'd be gaining with the lower price and misplaced performance.
Perhaps IBM will produce a dual core G5 and two chipsed chips. That would make more sense.
The power alone of a dual core G5 will more than enought to have stellar performances. (compare this to a PPC at 400 mhz or the pentium 700 ...)
Originally posted by Powerdoc
These story of 3 G5 is bs. There is bi processor, quadri processor, octo and cluster, but tri ?
There are in fact strings in AppleSystemInfo.strings that suggests that Darwin is ready to be put on a three way system. Link to Hardmac.com.
// String used to describe a triple processor configuration.
//
// IMPORTANT: Make sure the right hand side value contains the substring "%@".
// "%@" will get replaced by the processor speed and type string
// (e.g. "800 MHz PowerPC G3").
//--------------------------------------------------------------------------------
"3xCPUFormat" = "3 x %@";
Is there any technical obstacle for doing three way systems? Why should it always be about power of two? This rumored Xbox Next is a six way system.
And now to something completely different:
Comparison:
Two singe core processors fabbed at 130 nm, 1.3 V
vs.
Three dual core processors fabbed at 65 nm, .9 V
I we assume that the transistor count per core is the same and the cost of a processor is directly related to the size of the die, then the latter alternative would cost two thirds of the former.
The cost of the CPUs in Xbox Next would be considerable cheaper than those in current two way Power Macs.
If we assume that IBM manages to reduce wattage when moving to 65 nm, in the same ammount that they did when moving to 90 nm, then the latter alternative would draw a bit less power than the former.
The Xbox Next wouldn't need as much cooling as the current Power Mac, and they would be more heat resistant too.
If we assume that Microsoft will be requiring 40 times the ammount of processor cores as Apple use today (10 million Xbox Next, 6 cores in each = 60 million cores, compared to 1 million Power Macs with 1.5 cores each = 1,5 million) then Microsoft will be in a better spot for volume discount.
If we assume that MS will use the "slowest" version of a processor-family then we're talking about parts running at 3 GHz. These will be cheapest and draw the least ammount of power of all the processors in that generation comming of the production line.
970 is the smallest high end procesor out there today, less then half the size of a Prescott. If Intel can make a fat margin out of selling cheap Pentiums ($80?) then so can IBM. Why do we assume that these processors will cost a fortune for Microsoft? Is it because Apple demands $3000 for a complete two way Power Mac? Is there a way for Microsoft to make a Xbox Next with the suggested specs for let's say $500? I think so. Especially considering that they are developing just one machine, and then making tens of millions of identical boxes.
That's a lot to assume, but I think it's doable. With that said, I really don't think the report is accurate. Seems a bit to extreme for me. But, either way, Apple will be paying a lot less for processors in the future. That's good!
Originally posted by Eugene
I agree with whomever suggested the leak mistook having two virtual processors a la hyperthreading (SMT) for a CPU being "dual-core."
That would make more sense, wouldn't it? Its also clearly silly to think that MS would go from 1 processor chip in the XBox to 3 in the XBox2 as a cost and heat reduction measure. Assuming for a moment that the rumour has some validity, surely that implies something about the one chip that they are going to use...
And as for having 3 processors, what is wrong with that? Why is everybody so fixated on having powers-of-2 processors? If your processor is, say, 100 million transistors and your chip can hold 300 million transistors (for optimal power/cost/heat/yield reasons) do you just waste a third of your space? That's like saying you can build a 2-car garage, or a 4-car garage, but not a 3-car garage.
This probably comes from processor word sizes always being a power of 2... but that is driven because that multiple needs to be quickly calcuable in hardware using shifts instead of multiplies. The number of processor cores is under no such restrictions.
Originally posted by Henriok
...good deductions omitted...
That's a lot to assume, but I think it's doable. With that said, I really don't think the report is accurate. Seems a bit to extreme for me. But, either way, Apple will be paying a lot less for processors in the future. That's good!
Why do you think Apple will be paying less for processors? If MS is getting this much bang for the buck from IBM at the low price point, imagine how much bang Apple gets for more bucks? Apple's current price point gets them the margins they need to survive in the PC business, they cannot compete at the price points of mass produced fixed-design closed box consumer hardware. Completely different business model, and Microsoft has figured this out.
Originally posted by Programmer
That would make more sense, wouldn't it? Its also clearly silly to think that MS would go from 1 processor chip in the XBox to 3 in the XBox2 as a cost and heat reduction measure. Assuming for a moment that the rumour has some validity, surely that implies something about the one chip that they are going to use...
Ah. So either a triple-threaded core(!!!) or three cores on one die, with space the size of a 4th core free for an ASIC. Then you have a system-on-a-chip design, which wins on cost efficiency.
Re: Multithreading. The games don't have to explicitly support SMP for this to be useful. If you have concurrency, you can have system functions (sound, networking) running concurrently, which means the game can just tell another thread to "do this" and go right back to rendering frames. It gives the system vendor the ability to provide something other than bare metal to code to, which means easier ports and easier development, without the system getting in the way of the game. And, of course, if a game does decide to use the concurrent processing capabilities, they have a whole world of power to tap into for physics, AI, pre-rendering for instant transitions between levels, etc.
Originally posted by Amorph
Ah. So either a triple-threaded core(!!!) or three cores on one die, with space the size of a 4th core free for an ASIC. Then you have a system-on-a-chip design, which wins on cost efficiency.
And what space would that be? What 4th core?
Quote:
Re: Multithreading. The games don't have to explicitly support SMP for this to be useful. If you have concurrency, you can have system functions (sound, networking) running concurrently, which means the game can just tell another thread to "do this" and go right back to rendering frames. It gives the system vendor the ability to provide something other than bare metal to code to, which means easier ports and easier development, without the system getting in the way of the game. And, of course, if a game does decide to use the concurrent processing capabilities, they have a whole world of power to tap into for physics, AI, pre-rendering for instant transitions between levels, etc.
Games use around 99.5% of the available CPU, so you pretty much have to code the game to use it. Everything else that is expensive is in the GPU or DMA controllers.