Then I guess my money is on the 975 being called the G6. From what I havae read the 975 should be very different from the 970 to the point that Apple would call it the G6, and market it towards the pro machines.
You can believe what ever you want. The G5 is perdicted to reach 4.0 Ghz. Then the G6 will be introduced as a dual core. The G3 was a transition processor. Thats why it held the spotlight for a short term and it had problems.
You can believe what ever you want. The G5 is perdicted to reach 4.0 Ghz.
That is what I was saying, the G5 was supposed to hit 4.0GHz prior to the problems that they are having with the .09 process, 3.0GHz may be all they can get out of the 970fx if they can get that.
Quote:
Then the G6 will be introduced as a dual core.
Humm, so the POWER5 derived chips cannot be dual core? News to me, I thought that they could be.
Quote:
The G3 was a transition processor. Thats why it held the spotlight for a short term and it had problems.
My money is on the 975 or the POWER5 derived chips being the G6. G3 had problems, could you be more specific. For a chip with problems it ran for a long time in the product line. I thought the problem was with the G4, great low power but Moto could not get it to scale. I never thought that IBM was having problems with the G3. In fact I thought that IBM was not pushing the G3 because Apple did not want a consumer chip clocked higher than the pro chips.
The G6 might be the POWER6 derivative, but not because the numbers happen to line up.
The POWER5 is based on the POWER4 core. The POWER6 is a whole new beast. One slide that leaked out showed that where the former two CPUs have "two cores," the POWER6 will have "many cores" — a nod to Cell, perhaps? At any rate, from that data point alone, it'll be a significantly new thing, and IBM has confirmed that in general terms.
Until the POWER6 derivative hits the streets, we'll be looking at Apple "G5s."
G3 had problems, could you be more specific. For a chip with problems it ran for a long time in the product line. I thought the problem was with the G4, great low power but Moto could not get it to scale. I never thought that IBM was having problems with the G3. In fact I thought that IBM was not pushing the G3 because Apple did not want a consumer chip clocked higher than the pro chips.
The original G3 and G4 were very similar, and both apparently had an L1 cache timing bug that didn't get serious until the clock hit 500MHz. Both the 750 and the 7400 were stalled by the same problem, and they didn't budge until Motorola and IBM revised the designs later on. Since the G3 took a while to get to 500MHz from 233MHz, the problem didn't manifest for a long time.
After that problem was licked, the G3 was still a simple little CPU with a 5 stage pipeline. That's two stages shorter than the 745x G4's, and only one stage longer than the pipelines of the original G3 and G4. Since deeper pipelines allow for higher clock speeds (because the work of each stage takes less time) the posts talking about 1GHz G3s when the G4 was still at 733MHz were nothing more than wishful thinking. The G3 was never built for (clock) speed. In fact, in the last revision (the much-ballyhooed 750fx) IBM forecast a top speed of 1.1GHz and then reneged after they couldn't get it to go faster than 1GHz in any salable quantity. But since we have better-clocking G4s and the G5 now, nobody here cared about that recent IBM "speed dump."
No politics involved there, just simple engineering tradeoffs.
The G6 might be the POWER6 derivative, but not because the numbers happen to line up.
The POWER5 is based on the POWER4 core. The POWER6 is a whole new beast. One slide that leaked out showed that where the former two CPUs have "two cores," the POWER6 will have "many cores" — a nod to Cell, perhaps? At any rate, from that data point alone, it'll be a significantly new thing, and IBM has confirmed that in general terms.
Until the POWER6 derivative hits the streets, we'll be looking at Apple "G5s."
Well if the POWER5 is that similar to the POWER4, and the POWER6 is very different from both of them, then it's only logical that the POWER6 derivative, not the POWER5 derivative, will be called the PowerPC G6 by Apple. Therefore, the only question is "will the G6 appear in 2007, 2008, or later?".
Amorph said it already, but I'll say it again -- the POWER5 is a very close derivative of the POWER4. What is typically called the 975 around here is generally held to be a "half POWER5", i.e. 1 core w/ SMT and more cache. That is a fairly minor change to the 970FX, at least compared to what IBM is talking about for the POWER5 -> POWER6 transition.
I don't expect to see the POWER6 until 2006, and a 9xx derivative of it not before 2007. Before then we might see a couple of 970 derivatives, including the 975 and a multi-cored monster. The POWER4/5 architecture has lots of life left in it for Apple, they've barely gotten started. The 90nm transition hurt and delayed things, but it doesn't reduce the lifespan of the processor lineage. It might even extend it because more ingeneous solutions are required to move the next architecture forward.
Well if the POWER5 is that similar to the POWER4, and the POWER6 is very different from both of them, then it's only logical that the POWER6 derivative, not the POWER5 derivative, will be called the PowerPC G6 by Apple. Therefore, the only question is "will the G6 appear in 2007, 2008, or later?".
Quote:
Originally posted by Programmer
Amorph said it already, but I'll say it again -- the POWER5 is a very close derivative of the POWER4. What is typically called the 975 around here is generally held to be a "half POWER5", i.e. 1 core w/ SMT and more cache. That is a fairly minor change to the 970FX, at least compared to what IBM is talking about for the POWER5 -> POWER6 transition. I don't expect to see the POWER6 until 2006, and a 9xx derivative of it not before 2007. Before then we might see a couple of 970 derivatives, including the 975 and a multi-cored monster. The POWER4/5 architecture has lots of life left in it for Apple, they've barely gotten started. The 90nm transition hurt and delayed things, but it doesn't reduce the lifespan of the processor lineage. It might even extend it because more ingeneous solutions are required to move the next architecture forward.
I believe that it will be marketing that will decide which chip will be called the G6.
It might even extend it because more ingeneous solutions are required to move the next architecture forward.
I think that these solutions will improve performance beyond what had been intended. If the 97x series can't make it much beyond 3.0GHz dual core may be the way to go. The 980 series appears to be very different. If the .09 process will not yield chips that can be clocked to the point that IBM can get 20 to 30% more clock speed per update then the performance will come from somewhere. I believe that the 970 and 980 series are going to be able to take advantage of their parents and learn some of their tricks. What I am saying is that Apple wants to transition everything to the 970 series chips as soon as they can. I am under the impression that IBM fully intends to solve the yield issues, which may say take a 10% increase in speed per update and get performance other ways. When the yield issue is solved and the power issues are delt with Apple will want these chips everywhere due to the fact that they, I believe, are much cheaper than the G4s. So dual core for pro called the G6 based on the 975 design, or there abouts, and low power 97x's for the consumer machines. But I am just guessing.
I think Apple syncs up to the POWER naming schema.
Thus
The derivative of the POWER5 will be called G5+ or G5e and will remain that way until there is a POWER6 derivative which will then be a G6.
There is little benefit in upping the name of your computer like software. Every little changes doesn't need to correspond to a name change. Marketing is marketing.
Comments
Originally posted by Brendon
Then I guess my money is on the 975 being called the G6. From what I havae read the 975 should be very different from the 970 to the point that Apple would call it the G6, and market it towards the pro machines.
You can believe what ever you want. The G5 is perdicted to reach 4.0 Ghz. Then the G6 will be introduced as a dual core. The G3 was a transition processor. Thats why it held the spotlight for a short term and it had problems.
Originally posted by quagmire
You can believe what ever you want. The G5 is perdicted to reach 4.0 Ghz.
That is what I was saying, the G5 was supposed to hit 4.0GHz prior to the problems that they are having with the .09 process, 3.0GHz may be all they can get out of the 970fx if they can get that.
Then the G6 will be introduced as a dual core.
Humm, so the POWER5 derived chips cannot be dual core? News to me, I thought that they could be.
The G3 was a transition processor. Thats why it held the spotlight for a short term and it had problems.
My money is on the 975 or the POWER5 derived chips being the G6. G3 had problems, could you be more specific. For a chip with problems it ran for a long time in the product line. I thought the problem was with the G4, great low power but Moto could not get it to scale. I never thought that IBM was having problems with the G3. In fact I thought that IBM was not pushing the G3 because Apple did not want a consumer chip clocked higher than the pro chips.
The POWER5 is based on the POWER4 core. The POWER6 is a whole new beast. One slide that leaked out showed that where the former two CPUs have "two cores," the POWER6 will have "many cores" — a nod to Cell, perhaps? At any rate, from that data point alone, it'll be a significantly new thing, and IBM has confirmed that in general terms.
Until the POWER6 derivative hits the streets, we'll be looking at Apple "G5s."
Originally posted by Brendon
G3 had problems, could you be more specific. For a chip with problems it ran for a long time in the product line. I thought the problem was with the G4, great low power but Moto could not get it to scale. I never thought that IBM was having problems with the G3. In fact I thought that IBM was not pushing the G3 because Apple did not want a consumer chip clocked higher than the pro chips.
The original G3 and G4 were very similar, and both apparently had an L1 cache timing bug that didn't get serious until the clock hit 500MHz. Both the 750 and the 7400 were stalled by the same problem, and they didn't budge until Motorola and IBM revised the designs later on. Since the G3 took a while to get to 500MHz from 233MHz, the problem didn't manifest for a long time.
After that problem was licked, the G3 was still a simple little CPU with a 5 stage pipeline. That's two stages shorter than the 745x G4's, and only one stage longer than the pipelines of the original G3 and G4. Since deeper pipelines allow for higher clock speeds (because the work of each stage takes less time) the posts talking about 1GHz G3s when the G4 was still at 733MHz were nothing more than wishful thinking. The G3 was never built for (clock) speed. In fact, in the last revision (the much-ballyhooed 750fx) IBM forecast a top speed of 1.1GHz and then reneged after they couldn't get it to go faster than 1GHz in any salable quantity. But since we have better-clocking G4s and the G5 now, nobody here cared about that recent IBM "speed dump."
No politics involved there, just simple engineering tradeoffs.
Originally posted by Amorph
The G6 might be the POWER6 derivative, but not because the numbers happen to line up.
The POWER5 is based on the POWER4 core. The POWER6 is a whole new beast. One slide that leaked out showed that where the former two CPUs have "two cores," the POWER6 will have "many cores" — a nod to Cell, perhaps? At any rate, from that data point alone, it'll be a significantly new thing, and IBM has confirmed that in general terms.
Until the POWER6 derivative hits the streets, we'll be looking at Apple "G5s."
Well if the POWER5 is that similar to the POWER4, and the POWER6 is very different from both of them, then it's only logical that the POWER6 derivative, not the POWER5 derivative, will be called the PowerPC G6 by Apple. Therefore, the only question is "will the G6 appear in 2007, 2008, or later?".
I don't expect to see the POWER6 until 2006, and a 9xx derivative of it not before 2007. Before then we might see a couple of 970 derivatives, including the 975 and a multi-cored monster. The POWER4/5 architecture has lots of life left in it for Apple, they've barely gotten started. The 90nm transition hurt and delayed things, but it doesn't reduce the lifespan of the processor lineage. It might even extend it because more ingeneous solutions are required to move the next architecture forward.
Originally posted by wrldwzrd89
Well if the POWER5 is that similar to the POWER4, and the POWER6 is very different from both of them, then it's only logical that the POWER6 derivative, not the POWER5 derivative, will be called the PowerPC G6 by Apple. Therefore, the only question is "will the G6 appear in 2007, 2008, or later?".
Originally posted by Programmer
Amorph said it already, but I'll say it again -- the POWER5 is a very close derivative of the POWER4. What is typically called the 975 around here is generally held to be a "half POWER5", i.e. 1 core w/ SMT and more cache. That is a fairly minor change to the 970FX, at least compared to what IBM is talking about for the POWER5 -> POWER6 transition. I don't expect to see the POWER6 until 2006, and a 9xx derivative of it not before 2007. Before then we might see a couple of 970 derivatives, including the 975 and a multi-cored monster. The POWER4/5 architecture has lots of life left in it for Apple, they've barely gotten started. The 90nm transition hurt and delayed things, but it doesn't reduce the lifespan of the processor lineage. It might even extend it because more ingeneous solutions are required to move the next architecture forward.
I believe that it will be marketing that will decide which chip will be called the G6.
It might even extend it because more ingeneous solutions are required to move the next architecture forward.
I think that these solutions will improve performance beyond what had been intended. If the 97x series can't make it much beyond 3.0GHz dual core may be the way to go. The 980 series appears to be very different. If the .09 process will not yield chips that can be clocked to the point that IBM can get 20 to 30% more clock speed per update then the performance will come from somewhere. I believe that the 970 and 980 series are going to be able to take advantage of their parents and learn some of their tricks. What I am saying is that Apple wants to transition everything to the 970 series chips as soon as they can. I am under the impression that IBM fully intends to solve the yield issues, which may say take a 10% increase in speed per update and get performance other ways. When the yield issue is solved and the power issues are delt with Apple will want these chips everywhere due to the fact that they, I believe, are much cheaper than the G4s. So dual core for pro called the G6 based on the 975 design, or there abouts, and low power 97x's for the consumer machines. But I am just guessing.
China has developed their own PowerPC with the help of IBM.
http://www.theinquirer.net/?article=10724
http://headlines.yahoo.co.jp/hl?a=20...000004-scn-sci
The article mentions nothing about PowerPC but the photo says it all.
This CPU supposedly costs $20 to $35 per chip. I'm not sure if it's American dollars.
If anyone knows the specs of this chip, please post.
Thus
The derivative of the POWER5 will be called G5+ or G5e and will remain that way until there is a POWER6 derivative which will then be a G6.
There is little benefit in upping the name of your computer like software. Every little changes doesn't need to correspond to a name change. Marketing is marketing.