Obviously this is not Cell-architecture: the Cell alliance is not called STIMS
So if the Xbox2 processor is only made of PPE-dérivative (three of them, stripped-down in-order PPC cores) but without any SPE ; and since the great force of Cell lays in its multiple SPEs and not really in its PPE... it seems at this point that MS made a "mistake" to some extent, and that PS3 will smoke Xbox2? (I know we don't have much more information, it is just a quick assumption)
Whoever said the PPEs were cell unique architecture? The PPEs are most likely IBMs contribution from its own PPC portfolio and just included in the Cell fabric as the master processing unit. No MS interaction required to use the PPE and no STI permission either.
Cell is what every PS2 engineer has dreamed of getting their hands on for the past five years and where most of us knew Sony was heading. Large numbers of autonomous and fleshed out VUs. There is nothing to deal with.
We are ready.
PS2 engineers aren't mass market software engineers, they are wildly successful niche market engineers. You may be ready but the unwashed Xbox programmers are mainly PC programmers who had to learn a few restrictive concepts, not think totally differently from the outset.
Yes, but the PS3 is supposed to have 4 Cells... at 4 ~ 4.6 Ghz each, with 32 SPEs...
If Xbox2 has the same type of PPC core, I dont' see 3 PPEs only @ 3 GHz each compete with that!
Sony hasn't said what the PS3 will look like, but you can't seriously believe that a game console will have 4 50-80 watt processors (according to ISSCC) in it, can you?
Cell is what every PS2 engineer has dreamed of getting their hands on for the past five years and where most of us knew Sony was heading. Large numbers of autonomous and fleshed out VUs. There is nothing to deal with.
We are ready.
Time will tell. There is a big difference between knowing where Sony was going, and dealing with the reality of it. I'll be surprised if many game developers have it that together right off the bat, but perhaps I'm just an old skeptic. And I hope you're not expecting just to throw more polygons at it, that would represent a huge lack of imagination and a waste of horsepower.
Are you people fn serious here? if apple is struggling with 2.5 ghz and intel cant hit 4. Do you honestly think a $250 machine (because that's what it will be not a friggin supercomputer for children) will be have multicore prcessors running at 4.6 ghz and
3 ghz? ARE YOU ALL SMOKING CRACK? A single 3.8 ghz pentium chip cost say $1,250 now you wanna say a game console is gonna have technology a computer won't? Right so just slam a $2,000 dollar processor in there and everything else ram harddrive gpu and sell it for $250? For what your enjoyment? Come on people get real here you sound like you can't understand basic business concepts let alone where computing will be in the next few years.
Are you people fn serious here? if apple is struggling with 2.5 ghz and intel cant hit 4. Do you honestly think a $250 machine (because that's what it will be not a friggin supercomputer for children) will be have multicore prcessors running at 4.6 ghz and
3 ghz? ARE YOU ALL SMOKING CRACK? A single 3.8 ghz pentium chip cost say $1,250 now you wanna say a game console is gonna have technology a computer won't? Right so just slam a $2,000 dollar processor in there and everything else ram harddrive gpu and sell it for $250? For what your enjoyment? Come on people get real here you sound like you can't understand basic business concepts let alone where computing will be in the next few years.
First, welcome to AppleInsider.
Second, read up on Cell. It's running at 4.6 GHz.
How? Well, think of a CPU as a water channel, and clock speed as current. For any given amount of water per second, you can have a broad channel with a slow current, or a narrow channel with a rapid current, or anything in between.
The P4 and the 970 are both pretty broad, which means it takes a lot of water to get them going pretty fast. The Cell is pretty narrow, so it's relatively easy to get the current going really fast. Of course, there's a tradeoff inherent in that: The current has to be faster in order to move the same amount of water.
In other words, clockspeed is not some invariant absolute. It's one parameter among many that engineers balance with energy consumption, circuit complexity, instruction cracking, out-of-order execution resources, etc., etc. The particular balance that IBM settled on allows (because it requires) very high clock speeds. There's nothing magical about this, it's just that they decided to break away from the trend toward wider and more complex processors, and build a relatively simple, narrow one. They also made use of aggressive hand optimization of the core circuitry, and dynamic circuitry, to allow the clock to be set unusually high.
I hope that clarifies things. Nothing here is impossible, it's just a consequence of some unusual design decisions by IBM.
First of all thanks for the welcome, although i have read these forums for about a year as a guest. I do know about cell. One thing about cell though is it is at least 2 years out. Never does someone come up with something and release it as soon as the public gets wind of it. Let me tell you dual core and multicore are not quite cell processing. Cell will be after those become the norm. As much as it would be nice for technology to just jump laps and bounds the reality of it in the consumer world is this just doesn't happen. You have to R&D first and there are things that have been R&D'd that have not graced us yet. Those must come first because of the investment required in them they have to in turn make a profit from it. It has worked like this since the beginning of technology. No one, Let alone the whole industry just jumps years ahead of where they are now. Business can't be done like that. It is what restricts the technology available to us. Until such time as they have made money the previous venture and prices are reasonable to move to the next venture. How many times have you seen on here this is going to come out this year simply because someone made it possible. Well the sad new is it always takes a while. Second, you would not see technology like this pop up in a video game console before a computer. Especially since Sony is in the computer business if playstation where to have this it would have needed to be inside a computer 2 years ago to work with price and supply. Seeing as how their computers are showing no signs of this it may be some time before this great new technology gets to us. I think you'll find the next generaton game console's a cut above the previous ones, but they simply can't sell something like that in a video game console till it is in computers and has had time for prices to fall. I could be wrong and hopefully I am but this is just my 2 cents.
CPU - Xenon's CPU has three 3.0 GHz PowerPC cores. Each core is capable of two instructions per cycle and has an L1 cache with 32 KB for data and 32 KB for instructions. The three cores share 1 MB of L2 cache. Alpha 2 developer kits currently have two cores instead of three.
This is certainly not the ppc970.The G5 has a 64kb instruction cache and issues like 10 instruction per cycle.Either it is complete bunk or these are some console specific simplified cpu cores just for xbox2.Does anyone really believe a triple core G5 machine at 3 ghz and of course console priceing will be out next year? I got a feeling some bubbles will burst when the real specs are released.
This is certainly not the ppc970.The G5 has a 64kb instruction cache and issues like 10 instruction per cycle.Either it is complete bunk or these are some console specific simplified cpu cores just for xbox2.Does anyone really believe a triple core G5 machine at 3 ghz and of course console priceing will be out next year? I got a feeling some bubbles will burst when the real specs are released.
The ppc970 can consume up to 5 instructions per clock, although it rarely achieves that (3-4 if you're lucky). The complexity in the chip seems to arise from its out of order execution capabilities, and this limits its speed. This new core that keeps popping up is an in-order dual issue, which means it is limited to 2 instructions per clock. Other things being equal (and they're not entirely) that means a 4 GHz new core should be roughly (very roughly) the same speed as a 2 GHz 970. It may also consume less power at that speed.
Let me tell you dual core and multicore are not quite cell processing.
No, of course not. Cell, a token ring bus connecting a PowerPC core and a number of "synergistic processing units." The XBox CPU is not a Cell processor.
What's under discussion here is not Cell, but the PPE, the PowerPC core that happens to be used in the first Cell, revealed at ISSCC. It looks like XBox will use three PPEs on a core. That's not Cell, it's three PPEs on a core. This is nothing shocking. The whole industry is going multi-core, starting this year.
The PPE is capable of running at 4.6GHz now, at least in IBM's labs. It's designed to run at high clock speeds, and its design is much more aggressive than the P4's in that regard, so it's no surprise that it does.
Quote:
Second, you would not see technology like this pop up in a video game console before a computer. Especially since Sony is in the computer business if playstation where to have this it would have needed to be inside a computer 2 years ago to work with price and supply.
The much-ballyhooed Emotion Engine never saw the inside of a PC before it appeared in the PS2. The simple fact is that consoles have used lots of dedicated hardware for as long as consoles have existed (the XBox is an exception; the XBox 2 is not), and you're far more likely to see unconventional configurations in consoles than in PCs for the simple reason that consoles don't have the same kind of application legacy or variety that PCs do. The console manufacturers can also assume that their programmers will do a better job of writing to the hardware than PC application programmers tend to do (one can only hope...).
Consoles also are only required to be marginally profitable, and only then over a period of years. Console vendors usually start out well into the red, and lose money on the hardware for the first year. The usual goal is break-even over two or three or four years. Most of the money is made from game sales and licenses. So the economics of consoles are totally different from the economics of PCs, especially Windows PCs.
Quote:
Seeing as how their computers are showing no signs of this it may be some time before this great new technology gets to us. I think you'll find the next generaton game console's a cut above the previous ones, but they simply can't sell something like that in a video game console till it is in computers and has had time for prices to fall. I could be wrong and hopefully I am but this is just my 2 cents.
Technology goes at its own pace. There is no provable relationship between the money and time you put in and the results you get back. The G3 was a tremendous leap forward in performance, which shocked everyone involved, because it was designed to be the cheap, budget chip that replaced the 603e. As a result, the project to design the much more expensive and elaborate chip that was to replace the 604e was canned. Conversely, all kinds of money and time and talent has been poured into 90nm fabrication, and the results have been sobering.
What's more likely to happen is that the early PS3 games are nothing really astonishing, simply because the early games will be early encounters with the hardware, and with the tools. Once developers are familiar with the ins and outs of the PS3, games should just fly.
Consoles also are only required to be marginally profitable, and only then over a period of years. Console vendors usually start out well into the red, and lose money on the hardware for the first year. The usual goal is break-even over two or three or four years. Most of the money is made from game sales and licenses. So the economics of consoles are totally different from the economics of PCs, especially Windows PCs.
While true for Sony, Microsoft, and Sega (when they were in the hardware business) Nintendo generally makes a profit from consoles even in the beginning. Albeit a fairly small one most of the time.
Otherwise I agree with you, the goal is not to make money on the console.
While true for Sony, Microsoft, and Sega (when they were in the hardware business) Nintendo generally makes a profit from consoles even in the beginning. Albeit a fairly small one most of the time.
Otherwise I agree with you, the goal is not to make money on the console.
Even nintendo makes most of their money on the games themselves. I don't think their last cube thing was profitable from the start either.
Well yeah, at the end of it's life cycle - it;s made a few million - after bleeding 5 or 6 billion in the last 3 years.
I realize the box itself isn't supposed to make money, but neither is the rest of the division.
And your numbers come from where, exactly?
X-Boxes have had a positive profit margin for a while now.
X-Box Live alone brought in a few hundred million.
Games from the X-Box have made it to the top 10 sellers several times. If people own an X-Box and a PS2, they'll buy games for the X-Box (if they're available) because they know the graphics are better.
X-Boxes have had a positive profit margin for a while now.
X-Box Live alone brought in a few hundred million.
Games from the X-Box have made it to the top 10 sellers several times. If people own an X-Box and a PS2, they'll buy games for the X-Box (if they're available) because they know the graphics are better.
Actually when I buy games I now buy games for the XBOX instead of PS2 just because of XBOX live. No other reason.
X-Boxes have had a positive profit margin for a while now
They've only had a single positive quarter, this last one, and it was announced in the conference call it was an anomaly brought about by high demand for certain games and wasn't expected to last. Microsoft has long known they weren't going to run at a profit for this generation. This generation was all about getting into the market.
Comments
Originally posted by Cosmos 1999
Obviously this is not Cell-architecture: the Cell alliance is not called STIMS
So if the Xbox2 processor is only made of PPE-dérivative (three of them, stripped-down in-order PPC cores) but without any SPE ; and since the great force of Cell lays in its multiple SPEs and not really in its PPE... it seems at this point that MS made a "mistake" to some extent, and that PS3 will smoke Xbox2? (I know we don't have much more information, it is just a quick assumption)
Whoever said the PPEs were cell unique architecture? The PPEs are most likely IBMs contribution from its own PPC portfolio and just included in the Cell fabric as the master processing unit. No MS interaction required to use the PPE and no STI permission either.
Originally posted by Tuttle
Huh?
Cell is what every PS2 engineer has dreamed of getting their hands on for the past five years and where most of us knew Sony was heading. Large numbers of autonomous and fleshed out VUs. There is nothing to deal with.
We are ready.
PS2 engineers aren't mass market software engineers, they are wildly successful niche market engineers. You may be ready but the unwashed Xbox programmers are mainly PC programmers who had to learn a few restrictive concepts, not think totally differently from the outset.
Then again, Im thinking G5. Maybe they wont be G5s?
Originally posted by Cosmos 1999
Yes, but the PS3 is supposed to have 4 Cells... at 4 ~ 4.6 Ghz each, with 32 SPEs...
If Xbox2 has the same type of PPC core, I dont' see 3 PPEs only @ 3 GHz each compete with that!
Sony hasn't said what the PS3 will look like, but you can't seriously believe that a game console will have 4 50-80 watt processors (according to ISSCC) in it, can you?
Originally posted by Tuttle
Huh?
Cell is what every PS2 engineer has dreamed of getting their hands on for the past five years and where most of us knew Sony was heading. Large numbers of autonomous and fleshed out VUs. There is nothing to deal with.
We are ready.
Time will tell. There is a big difference between knowing where Sony was going, and dealing with the reality of it. I'll be surprised if many game developers have it that together right off the bat, but perhaps I'm just an old skeptic. And I hope you're not expecting just to throw more polygons at it, that would represent a huge lack of imagination and a waste of horsepower.
I'm looking forward to being impressed.
3 ghz? ARE YOU ALL SMOKING CRACK? A single 3.8 ghz pentium chip cost say $1,250 now you wanna say a game console is gonna have technology a computer won't? Right so just slam a $2,000 dollar processor in there and everything else ram harddrive gpu and sell it for $250? For what your enjoyment? Come on people get real here you sound like you can't understand basic business concepts let alone where computing will be in the next few years.
Originally posted by knowledgebase
Are you people fn serious here? if apple is struggling with 2.5 ghz and intel cant hit 4. Do you honestly think a $250 machine (because that's what it will be not a friggin supercomputer for children) will be have multicore prcessors running at 4.6 ghz and
3 ghz? ARE YOU ALL SMOKING CRACK? A single 3.8 ghz pentium chip cost say $1,250 now you wanna say a game console is gonna have technology a computer won't? Right so just slam a $2,000 dollar processor in there and everything else ram harddrive gpu and sell it for $250? For what your enjoyment? Come on people get real here you sound like you can't understand basic business concepts let alone where computing will be in the next few years.
First, welcome to AppleInsider.
Second, read up on Cell. It's running at 4.6 GHz.
How? Well, think of a CPU as a water channel, and clock speed as current. For any given amount of water per second, you can have a broad channel with a slow current, or a narrow channel with a rapid current, or anything in between.
The P4 and the 970 are both pretty broad, which means it takes a lot of water to get them going pretty fast. The Cell is pretty narrow, so it's relatively easy to get the current going really fast. Of course, there's a tradeoff inherent in that: The current has to be faster in order to move the same amount of water.
In other words, clockspeed is not some invariant absolute. It's one parameter among many that engineers balance with energy consumption, circuit complexity, instruction cracking, out-of-order execution resources, etc., etc. The particular balance that IBM settled on allows (because it requires) very high clock speeds. There's nothing magical about this, it's just that they decided to break away from the trend toward wider and more complex processors, and build a relatively simple, narrow one. They also made use of aggressive hand optimization of the core circuitry, and dynamic circuitry, to allow the clock to be set unusually high.
I hope that clarifies things. Nothing here is impossible, it's just a consequence of some unusual design decisions by IBM.
...I could be wrong...
CPU - Xenon's CPU has three 3.0 GHz PowerPC cores. Each core is capable of two instructions per cycle and has an L1 cache with 32 KB for data and 32 KB for instructions. The three cores share 1 MB of L2 cache. Alpha 2 developer kits currently have two cores instead of three.
This is certainly not the ppc970.The G5 has a 64kb instruction cache and issues like 10 instruction per cycle.Either it is complete bunk or these are some console specific simplified cpu cores just for xbox2.Does anyone really believe a triple core G5 machine at 3 ghz and of course console priceing will be out next year? I got a feeling some bubbles will burst when the real specs are released.
Originally posted by cuneglasus
This is certainly not the ppc970.The G5 has a 64kb instruction cache and issues like 10 instruction per cycle.Either it is complete bunk or these are some console specific simplified cpu cores just for xbox2.Does anyone really believe a triple core G5 machine at 3 ghz and of course console priceing will be out next year? I got a feeling some bubbles will burst when the real specs are released.
The ppc970 can consume up to 5 instructions per clock, although it rarely achieves that (3-4 if you're lucky). The complexity in the chip seems to arise from its out of order execution capabilities, and this limits its speed. This new core that keeps popping up is an in-order dual issue, which means it is limited to 2 instructions per clock. Other things being equal (and they're not entirely) that means a 4 GHz new core should be roughly (very roughly) the same speed as a 2 GHz 970. It may also consume less power at that speed.
Originally posted by knowledgebase
Let me tell you dual core and multicore are not quite cell processing.
No, of course not. Cell, a token ring bus connecting a PowerPC core and a number of "synergistic processing units." The XBox CPU is not a Cell processor.
What's under discussion here is not Cell, but the PPE, the PowerPC core that happens to be used in the first Cell, revealed at ISSCC. It looks like XBox will use three PPEs on a core. That's not Cell, it's three PPEs on a core. This is nothing shocking. The whole industry is going multi-core, starting this year.
The PPE is capable of running at 4.6GHz now, at least in IBM's labs. It's designed to run at high clock speeds, and its design is much more aggressive than the P4's in that regard, so it's no surprise that it does.
Second, you would not see technology like this pop up in a video game console before a computer. Especially since Sony is in the computer business if playstation where to have this it would have needed to be inside a computer 2 years ago to work with price and supply.
The much-ballyhooed Emotion Engine never saw the inside of a PC before it appeared in the PS2. The simple fact is that consoles have used lots of dedicated hardware for as long as consoles have existed (the XBox is an exception; the XBox 2 is not), and you're far more likely to see unconventional configurations in consoles than in PCs for the simple reason that consoles don't have the same kind of application legacy or variety that PCs do. The console manufacturers can also assume that their programmers will do a better job of writing to the hardware than PC application programmers tend to do (one can only hope...).
Consoles also are only required to be marginally profitable, and only then over a period of years. Console vendors usually start out well into the red, and lose money on the hardware for the first year. The usual goal is break-even over two or three or four years. Most of the money is made from game sales and licenses. So the economics of consoles are totally different from the economics of PCs, especially Windows PCs.
Seeing as how their computers are showing no signs of this it may be some time before this great new technology gets to us. I think you'll find the next generaton game console's a cut above the previous ones, but they simply can't sell something like that in a video game console till it is in computers and has had time for prices to fall. I could be wrong and hopefully I am but this is just my 2 cents.
Technology goes at its own pace. There is no provable relationship between the money and time you put in and the results you get back. The G3 was a tremendous leap forward in performance, which shocked everyone involved, because it was designed to be the cheap, budget chip that replaced the 603e. As a result, the project to design the much more expensive and elaborate chip that was to replace the 604e was canned. Conversely, all kinds of money and time and talent has been poured into 90nm fabrication, and the results have been sobering.
What's more likely to happen is that the early PS3 games are nothing really astonishing, simply because the early games will be early encounters with the hardware, and with the tools. Once developers are familiar with the ins and outs of the PS3, games should just fly.
Originally posted by Amorph
Consoles also are only required to be marginally profitable, and only then over a period of years. Console vendors usually start out well into the red, and lose money on the hardware for the first year. The usual goal is break-even over two or three or four years. Most of the money is made from game sales and licenses. So the economics of consoles are totally different from the economics of PCs, especially Windows PCs.
While true for Sony, Microsoft, and Sega (when they were in the hardware business) Nintendo generally makes a profit from consoles even in the beginning. Albeit a fairly small one most of the time.
Otherwise I agree with you, the goal is not to make money on the console.
Originally posted by Electric Monk
While true for Sony, Microsoft, and Sega (when they were in the hardware business) Nintendo generally makes a profit from consoles even in the beginning. Albeit a fairly small one most of the time.
Otherwise I agree with you, the goal is not to make money on the console.
Even nintendo makes most of their money on the games themselves. I don't think their last cube thing was profitable from the start either.
Originally posted by the cool gut
Well yeah, at the end of it's life cycle - it;s made a few million - after bleeding 5 or 6 billion in the last 3 years.
I realize the box itself isn't supposed to make money, but neither is the rest of the division.
And your numbers come from where, exactly?
X-Boxes have had a positive profit margin for a while now.
X-Box Live alone brought in a few hundred million.
Games from the X-Box have made it to the top 10 sellers several times. If people own an X-Box and a PS2, they'll buy games for the X-Box (if they're available) because they know the graphics are better.
Originally posted by slughead
And your numbers come from where, exactly?
X-Boxes have had a positive profit margin for a while now.
X-Box Live alone brought in a few hundred million.
Games from the X-Box have made it to the top 10 sellers several times. If people own an X-Box and a PS2, they'll buy games for the X-Box (if they're available) because they know the graphics are better.
Actually when I buy games I now buy games for the XBOX instead of PS2 just because of XBOX live. No other reason.
Originally posted by slughead
X-Boxes have had a positive profit margin for a while now
They've only had a single positive quarter, this last one, and it was announced in the conference call it was an anomaly brought about by high demand for certain games and wasn't expected to last. Microsoft has long known they weren't going to run at a profit for this generation. This generation was all about getting into the market.
Originally posted by Telomar This generation was all about getting into the market. [/B]
Certainly sounds better than the more accurate this generation was all about getting destroyed by Sony in the market.