IBM really has pulled off a miracle with the 970. The miracle is that it looks like it will have 80% of the performance of a P4 . . .
Even if your estimate is correct for standard, non-AltiVec operations, the new dual PowerMacs will exceed the single P4 Windows PCs. We don't see many dual P4 systems, but dual 970 PowerMacs should be common.
With AltiVec enabled code, I believe the single 970 will likely beat the P4 too. The 970 should also scale up in clock rate more quickly than the P4 can, from its present clock rate. This should give the 970 a greater advantage in the future.
200mhz PCI bus unless i am mistaken. the speed in which the cards talk to the motherboard.
But then why in that same sentence would they mention PC3200 RAM? They definatly meant the speed of the RAM or the FSB. Since they mention BUS, it leads me to believe they are talking about the FSB, not the Ram "bus" (is there such a thing?)
Now here is a question for someone more technical. Would it make sene to have the RAM running at half the FSB speed? With Intels 800Mhz (4x200MHz) FSB, what type of RAM are they using, and at what speed?
200mhz PCI bus unless i am mistaken. the speed in which the cards talk to the motherboard.
Yes that was my understanding too. The CPU/memory bus will be as indicated. 1:4 of the CPU speed, double-pumped. i.e 1400 will be 350x4 for total of 700mhx.
What are the current PCI?
Previous Macs were crippled in bus speeds because they had to be, not because Apple wanted them to be. The Motorola processors couldn't support anything better. Now with something well beyond that available, I am confident that Apple will exploit its capabilities to the max.
But then why in that same sentence would they mention PC3200 RAM? They definatly meant the speed of the RAM or the FSB. Since they mention BUS, it leads me to believe they are talking about the FSB, not the Ram "bus" (is there such a thing?)
Now here is a question for someone more technical. Would it make sene to have the RAM running at half the FSB speed? With Intels 800Mhz (4x200MHz) FSB, what type of RAM are they using, and at what speed?
Sometimes translation between languages can be a bit rough, as certain words and phrasing do not translate easily. Without someone who is truly fluent in both French and American English, things can get a bit confusing.
I think they are trying to say the PCI bus is 200mhz, and that the RAM is PC3200. You have to be a little bit flexible with translations until you can get a real pro on it who is completely familiar with the idioms of both languages.
It would be just as confusing trying go from American English to French or any other language.
Yes that was my understanding too. The CPU/memory bus will be as indicated. 1:4 of the CPU speed, double-pumped. i.e 1400 will be 350x4 for total of 700mhx.
What are the current PCI?
Previous Macs were crippled in bus speeds because they had to be, not because Apple wanted them to be. The Motorola processors couldn't support anything better. Now with something well beyond that available, I am confident that Apple will exploit its capabilities to the max.
Due to me being technically challenged, can you explain how the bus is 1/4 of the CPU speed but is only double pumped? I thought the bus was 800mhz, that would be quad pumped? Or is it because of duals? I'm missing 400mhz somewhere.
Sometimes translation between languages can be a bit rough, as certain words and phrasing do not translate easily. Without someone who is truly fluent in both French and American English, things can get a bit confusing.
Ok, but I thought someone here did a translation on there own (not using a translator program).
Quote:
I think they are trying to say the PCI bus is 200mhz, and that the RAM is PC3200.
What is PCI-X? The Xserve has 66MHz PCI slot (right?), so does a jump to 200MHz sound right?
But lets look at the PC3200 Memory. What speed does it run at? 200MHz. So this sounds more and more like they did mean the bus was 200MHz.
Due to me being technically challenged, can you explain how the bus is 1/4 of the CPU speed but is only double pumped? I thought the bus was 800mhz, that would be quad pumped? Or is it because of duals? I'm missing 400mhz somewhere.
The 970 bus is based on clock speed at a ratio of 1:4. So a 1.8GHz processor would have a bus running at 450MHz which is 1/4 or 1800. However, since the bus is double pumped, it is running at 900MHz effective.
So if you had a 1.4GHz 970, the bus would be 350MHz, or 700MHz effective.
Now here is a question for someone more technical. Would it make sene to have the RAM running at half the FSB speed? With Intels 800Mhz (4x200MHz) FSB, what type of RAM are they using, and at what speed?
Yes, synchronous bus speeds are best. An athlon with a 333 MHz (166 DDR) FSB runs better with DDR333 (PC2700) RAM than with DDR400 (PC3200) on most things.
Intel's 800 MHz (200 QDR) FSB chipset Canterwood (& I think Springdale) uses a 2 channel DDR400 memory controller.
"Standard" PCI (1.0?) is 33MHz and 32 bit giving 133 MB/s.
There are also 66 MHz and 66MHz/64 bit PCI flavours giving 266MB/s and 533 MB/s respectively.
Quote:
Originally posted by kupan787
What is PCI-X?
Going to www.pcisig.org and checking the news page it seems that PCI-X used to be known as 3GIO. It has an "initial bit rate of 2.5 Gigabits per second per lane per direction" and a "16-lane PCI Express interconnect can provide data transfer rates of more than 8 Gigabytes per second". (Here: http://www.pcisig.com/news_room/news...ses/2002_07_23 )
PCI-X 2.0 has 2 flavours: PCI-X 266 and PCI-X 533 which give 2133 MB/s and 4267 MB/s respectively. (These come from here: http://www.pcisig.com/news_room/news...es/2002_07_23b ) I think these must be DDR rates at 64 bit each since they use the phrase "PCI-X 266, runs at speeds up to 266 Mega transfers per second".
Due to me being technically challenged, can you explain how the bus is 1/4 of the CPU speed but is only double pumped? I thought the bus was 800mhz, that would be quad pumped? Or is it because of duals? I'm missing 400mhz somewhere.
Don't feel bad. All this stuff still has my head swimming most days.
The PCI bus speed is seperate from the System/CPU bus speed. That is what is mentioned as 200mhz, (the speed at which the PCI can communicate with the motherboard) which is quite a jump up from current PCI bus apparently.
System/CPU bus speed will be vary with the processor speed. A 1400 processor will have a bus ratio of 1:4 or 350 mhz. doublepump that and you have an actual speed of 700mhz, or 1/2 half of the CPU speed. A faster CPU will equate to a faster system/CPU bus. Hope that helps.
PC3200 I believe is 400mhz. Somebody correct me if I am wrong.
I am wondering if the 1800mhz chip will require PC3500 RAM? Is it possible that the system performance may soon outpace the speed of affordable RAM?
"Standard" PCI (1.0?) is 33MHz and 32 bit giving 133 MB/s.
There are also 66 MHz and 66MHz/64 bit PCI flavours giving 266MB/s and 533 MB/s respectively.
Going to www.pcisig.org and checking the news page it seems that PCI-X used to be known as 3GIO. It has an "initial bit rate of 2.5 Gigabits per second per lane per direction" and a "16-lane PCI Express interconnect can provide data transfer rates of more than 8 Gigabytes per second". (Here: http://www.pcisig.com/news_room/news...ses/2002_07_23 )
PCI-X 2.0 has 2 flavours: PCI-X 266 and PCI-X 533 which give 2133 MB/s and 4267 MB/s respectively. (These come from here: http://www.pcisig.com/news_room/news...es/2002_07_23b ) I think these must be DDR rates at 64 bit each since they use the phrase "PCI-X 266, runs at speeds up to 266 Mega transfers per second".
MM
NO
Do not confuse PCI-X and PCI express, they are completely different beasts. What used to be called 3GIO (amongst many other names) is PCI express is a new physical interface, essentially serial, and very fast, with small connectors. PCI express is not yet available.
PCI-X is an extension to PCI that enables it to run at higher speeds, and much higher bandwidths. PCI-X is widely available already (at least at it's slower speeds).
I am wondering if the 1800mhz chip will require PC3500 RAM? Is it possible that the system performance may soon outpace the speed of affordable RAM?
RAM does not have to run at the same speed as the FSB (current PowerMacs being an example). I'd hope that Apple would use two channels of DDR SDRAM in the faster 970s' case rather than have RAM with bandwidth lower than the FSB.
PC3200 I believe is 400mhz. Somebody correct me if I am wrong.
It is 200MHz, but double pumped so it is 400MHz effective. So if you were thinking the 400MHz was doubled to 800, you were wrong. However if you were quoting the effective figure, you were right
Since nobody else has pointed it out, I will: the clock rate of the 970's FSB is completely independent of the speed of memory, AGP, PCI, etc. It is locked to half the processor clock rate, and that's it. You can't even directly compute the bandwidth it has at a given clock rate without knowing more about the protocol because it does not have seperate address/data lines -- it uses a packet protocol on the seperate read and write busses. This is a very different animal, don't apply what you know from MPX (or even the AMD & Intel busses) to the 970's FSB. It is more akin to HyperTransport or RapidIO.
My guess is that the rumour is bogus. The benchmarks could be bogus too, we simply don't know at this point.
Since nobody else has pointed it out, I will: the clock rate of the 970's FSB is completely independent of the speed of memory, AGP, PCI, etc. It is locked to half the processor clock rate, and that's it. You can't even directly compute the bandwidth it has at a given clock rate without knowing more about the protocol because it does not have seperate address/data lines -- it uses a packet protocol on the seperate read and write busses. This is a very different animal, don't apply what you know from MPX (or even the AMD & Intel busses) to the 970's FSB. It is more akin to HyperTransport or RapidIO.
My guess is that the rumour is bogus. The benchmarks could be bogus too, we simply don't know at this point.
Thanks programmer. A simple straightforward explanation that actually almost makes sense to me. Sounds like when the 970 does come out, we will all have to get used to a new way of thinking about how it all works together.
It is 200MHz, but double pumped so it is 400MHz effective. So if you were thinking the 400MHz was doubled to 800, you were wrong. However if you were quoting the effective figure, you were right
Maybe I'm misunderstanding you but PC3200 memory IS 400mhz. DDR400. Double pumped (as in the new P4 chipset) makes it 800mhz effective.
Perhaps the most obvious problem with the benchmarks was the fact that Bryce, which does not use multiple processors, sped up substantially using a dual. There are numerous other somewhat questionable aspects to the benchmarks.
The 80% figure is an educated guess based on the speculative SPEC benchmarks that IBM posted.
There are hundreds of different metrics that can be used to compare processors. IBM has almost certainly advertised its strongest rating, so I'm assuming that it's probably somewhat weaker in the non-advertised aspects and in a real world situation (i.e. Apple is not going to supply unlimited cache, fastest possible memory, etc.). The p4 with 800 MHz front side bus is a pretty hefty creature and is significantly faster than the previous P4.
I'll show where I got the 1% figure (although it is a guesstimate based on relative expenditures, since IBM doesn't break out microprocessor research by chip type) in another post when I have the time. Suffice it to say, the 970 was done very cheaply.
This is a good thing, by the way. If IBM had spent a billion on research, just amortizing the R&D would make the chip cost $500 before even getting to production expenses. If Apple is going to use a few million a year of these, they'd better not have spent more than $100 million on chip R+D. Compare this to the 10 billion that Intel has spent over the last few years (roughly $4 billion/year).
Perhaps the most obvious problem with the benchmarks was the fact that Bryce, which does not use multiple processors, sped up substantially using a dual. There are numerous other somewhat questionable aspects to the benchmarks.
I'm not saying I believe the benchmarks but I haven't seen a really compelling reason why they are blatantly false. Most of the reasons are just as weak as the benchmarks. They offered the explanation that a more recent (beta) version of Bryce was used -- given that the rumour stated that Apple was doing these benchmarks this is entirely plausible since Apple commonly has pre-release versions of software for testing on new hardware.
Quote:
The 80% figure is an educated guess based on the speculative SPEC benchmarks that IBM posted.
There are hundreds of different metrics that can be used to compare processors. IBM has almost certainly advertised its strongest rating, so I'm assuming that it's probably somewhat weaker in the non-advertised aspects and in a real world situation (i.e. Apple is not going to supply unlimited cache, fastest possible memory, etc.). The p4 with 800 MHz front side bus is a pretty hefty creature and is significantly faster than the previous P4.
IBM is usually fairly conservative with things like estimated SPECmarks for a processor they don't yet have silicon for. They've gotten in trouble with things like this before so I wouldn't necessarily discount their estimates, nor would I assume that Apple will come to market with a significantly weaker system. Looking at the processor's specifications, clock rate, and the performance of the POWER4 I'd say that the estimate were entirely reasonable. There were also Moki's comments about some sandbagging going on, and the various rumours of IBM being pleased with how the 970 has turned out.
On the other hand you're right that the new P4 is a beast. The 970 might only be 80% of its performance, but its possible that Apple will beat that number. We just don't have enough information to make a reasonable guess, but even 80% of the new Intel hardware is going to be a lot better than we've got right now!
Quote:
I'll show where I got the 1% figure (although it is a guesstimate based on relative expenditures, since IBM doesn't break out microprocessor research by chip type) in another post when I have the time. Suffice it to say, the 970 was done very cheaply.
This is a good thing, by the way. If IBM had spent a billion on research, just amortizing the R&D would make the chip cost $500 before even getting to production expenses. If Apple is going to use a few million a year of these, they'd better not have spent more than $100 million on chip R+D. Compare this to the 10 billion that Intel has spent over the last few years (roughly $4 billion/year).
I think you're pretty much right on the money here. IBM is using some advanced circuit design tools to build the POWER series, and the 970 directly leverages the POWER4 work. This has already served IBM very well and it will allow them to continue to compete with Intel into the future on a far more cost effective basis. It may cost them some potential performance, but it might also allow them to focus on high level improvements rather than getting bogged down in detail improvements. If the tools they are using can stay on the bleeding edge it'll allow IBM to take advantage of new techniques & processes faster, but if the tools don't evolve quickly this could turn into a liability. The single biggest advantage in IBM's approach is that they should be able to build massive processors much more quickly than Intel can -- when chips reach into the hundreds of millions of transistors, IBM will be able to do more interesting things with that sooner. Unless, of course, Intel adopts the same strategy.
Comments
Originally posted by Tom West
IBM really has pulled off a miracle with the 970. The miracle is that it looks like it will have 80% of the performance of a P4 . . .
Even if your estimate is correct for standard, non-AltiVec operations, the new dual PowerMacs will exceed the single P4 Windows PCs. We don't see many dual P4 systems, but dual 970 PowerMacs should be common.
With AltiVec enabled code, I believe the single 970 will likely beat the P4 too. The 970 should also scale up in clock rate more quickly than the P4 can, from its present clock rate. This should give the 970 a greater advantage in the future.
Originally posted by keyboardf12
200mhz PCI bus unless i am mistaken. the speed in which the cards talk to the motherboard.
But then why in that same sentence would they mention PC3200 RAM? They definatly meant the speed of the RAM or the FSB. Since they mention BUS, it leads me to believe they are talking about the FSB, not the Ram "bus" (is there such a thing?)
Now here is a question for someone more technical. Would it make sene to have the RAM running at half the FSB speed? With Intels 800Mhz (4x200MHz) FSB, what type of RAM are they using, and at what speed?
Originally posted by keyboardf12
200mhz PCI bus unless i am mistaken. the speed in which the cards talk to the motherboard.
Yes that was my understanding too. The CPU/memory bus will be as indicated. 1:4 of the CPU speed, double-pumped. i.e 1400 will be 350x4 for total of 700mhx.
What are the current PCI?
Previous Macs were crippled in bus speeds because they had to be, not because Apple wanted them to be. The Motorola processors couldn't support anything better. Now with something well beyond that available, I am confident that Apple will exploit its capabilities to the max.
Originally posted by kupan787
But then why in that same sentence would they mention PC3200 RAM? They definatly meant the speed of the RAM or the FSB. Since they mention BUS, it leads me to believe they are talking about the FSB, not the Ram "bus" (is there such a thing?)
Now here is a question for someone more technical. Would it make sene to have the RAM running at half the FSB speed? With Intels 800Mhz (4x200MHz) FSB, what type of RAM are they using, and at what speed?
Sometimes translation between languages can be a bit rough, as certain words and phrasing do not translate easily. Without someone who is truly fluent in both French and American English, things can get a bit confusing.
I think they are trying to say the PCI bus is 200mhz, and that the RAM is PC3200. You have to be a little bit flexible with translations until you can get a real pro on it who is completely familiar with the idioms of both languages.
It would be just as confusing trying go from American English to French or any other language.
Originally posted by Shaktai
Yes that was my understanding too. The CPU/memory bus will be as indicated. 1:4 of the CPU speed, double-pumped. i.e 1400 will be 350x4 for total of 700mhx.
What are the current PCI?
Previous Macs were crippled in bus speeds because they had to be, not because Apple wanted them to be. The Motorola processors couldn't support anything better. Now with something well beyond that available, I am confident that Apple will exploit its capabilities to the max.
Due to me being technically challenged, can you explain how the bus is 1/4 of the CPU speed but is only double pumped? I thought the bus was 800mhz, that would be quad pumped? Or is it because of duals? I'm missing 400mhz somewhere.
Originally posted by Shaktai
Sometimes translation between languages can be a bit rough, as certain words and phrasing do not translate easily. Without someone who is truly fluent in both French and American English, things can get a bit confusing.
Ok, but I thought someone here did a translation on there own (not using a translator program).
I think they are trying to say the PCI bus is 200mhz, and that the RAM is PC3200.
What is PCI-X? The Xserve has 66MHz PCI slot (right?), so does a jump to 200MHz sound right?
But lets look at the PC3200 Memory. What speed does it run at? 200MHz. So this sounds more and more like they did mean the bus was 200MHz.
Originally posted by KidRed
Due to me being technically challenged, can you explain how the bus is 1/4 of the CPU speed but is only double pumped? I thought the bus was 800mhz, that would be quad pumped? Or is it because of duals? I'm missing 400mhz somewhere.
The 970 bus is based on clock speed at a ratio of 1:4. So a 1.8GHz processor would have a bus running at 450MHz which is 1/4 or 1800. However, since the bus is double pumped, it is running at 900MHz effective.
So if you had a 1.4GHz 970, the bus would be 350MHz, or 700MHz effective.
Originally posted by Tom West
Not to rain on anybody's parade, ...
That's not rain, that's a monsoon.
Just curious, why do you say 80%?
Originally posted by kupan787
Now here is a question for someone more technical. Would it make sene to have the RAM running at half the FSB speed? With Intels 800Mhz (4x200MHz) FSB, what type of RAM are they using, and at what speed?
Yes, synchronous bus speeds are best. An athlon with a 333 MHz (166 DDR) FSB runs better with DDR333 (PC2700) RAM than with DDR400 (PC3200) on most things.
Intel's 800 MHz (200 QDR) FSB chipset Canterwood (& I think Springdale) uses a 2 channel DDR400 memory controller.
MM
Originally posted by Shaktai
What are the current PCI?
"Standard" PCI (1.0?) is 33MHz and 32 bit giving 133 MB/s.
There are also 66 MHz and 66MHz/64 bit PCI flavours giving 266MB/s and 533 MB/s respectively.
Originally posted by kupan787
What is PCI-X?
Going to www.pcisig.org and checking the news page it seems that PCI-X used to be known as 3GIO. It has an "initial bit rate of 2.5 Gigabits per second per lane per direction" and a "16-lane PCI Express interconnect can provide data transfer rates of more than 8 Gigabytes per second". (Here: http://www.pcisig.com/news_room/news...ses/2002_07_23 )
PCI-X 2.0 has 2 flavours: PCI-X 266 and PCI-X 533 which give 2133 MB/s and 4267 MB/s respectively. (These come from here: http://www.pcisig.com/news_room/news...es/2002_07_23b ) I think these must be DDR rates at 64 bit each since they use the phrase "PCI-X 266, runs at speeds up to 266 Mega transfers per second".
MM
Originally posted by KidRed
Due to me being technically challenged, can you explain how the bus is 1/4 of the CPU speed but is only double pumped? I thought the bus was 800mhz, that would be quad pumped? Or is it because of duals? I'm missing 400mhz somewhere.
Don't feel bad. All this stuff still has my head swimming most days.
The PCI bus speed is seperate from the System/CPU bus speed. That is what is mentioned as 200mhz, (the speed at which the PCI can communicate with the motherboard) which is quite a jump up from current PCI bus apparently.
System/CPU bus speed will be vary with the processor speed. A 1400 processor will have a bus ratio of 1:4 or 350 mhz. doublepump that and you have an actual speed of 700mhz, or 1/2 half of the CPU speed. A faster CPU will equate to a faster system/CPU bus. Hope that helps.
PC3200 I believe is 400mhz. Somebody correct me if I am wrong.
I am wondering if the 1800mhz chip will require PC3500 RAM? Is it possible that the system performance may soon outpace the speed of affordable RAM?
Originally posted by MartianMatt
"Standard" PCI (1.0?) is 33MHz and 32 bit giving 133 MB/s.
There are also 66 MHz and 66MHz/64 bit PCI flavours giving 266MB/s and 533 MB/s respectively.
Going to www.pcisig.org and checking the news page it seems that PCI-X used to be known as 3GIO. It has an "initial bit rate of 2.5 Gigabits per second per lane per direction" and a "16-lane PCI Express interconnect can provide data transfer rates of more than 8 Gigabytes per second". (Here: http://www.pcisig.com/news_room/news...ses/2002_07_23 )
PCI-X 2.0 has 2 flavours: PCI-X 266 and PCI-X 533 which give 2133 MB/s and 4267 MB/s respectively. (These come from here: http://www.pcisig.com/news_room/news...es/2002_07_23b ) I think these must be DDR rates at 64 bit each since they use the phrase "PCI-X 266, runs at speeds up to 266 Mega transfers per second".
MM
NO
Do not confuse PCI-X and PCI express, they are completely different beasts. What used to be called 3GIO (amongst many other names) is PCI express is a new physical interface, essentially serial, and very fast, with small connectors. PCI express is not yet available.
PCI-X is an extension to PCI that enables it to run at higher speeds, and much higher bandwidths. PCI-X is widely available already (at least at it's slower speeds).
michael
I am wondering if the 1800mhz chip will require PC3500 RAM? Is it possible that the system performance may soon outpace the speed of affordable RAM?
RAM does not have to run at the same speed as the FSB (current PowerMacs being an example). I'd hope that Apple would use two channels of DDR SDRAM in the faster 970s' case rather than have RAM with bandwidth lower than the FSB.
Originally posted by Shaktai
PC3200 I believe is 400mhz. Somebody correct me if I am wrong.
It is 200MHz, but double pumped so it is 400MHz effective. So if you were thinking the 400MHz was doubled to 800, you were wrong. However if you were quoting the effective figure, you were right
My guess is that the rumour is bogus. The benchmarks could be bogus too, we simply don't know at this point.
Originally posted by Programmer
Since nobody else has pointed it out, I will: the clock rate of the 970's FSB is completely independent of the speed of memory, AGP, PCI, etc. It is locked to half the processor clock rate, and that's it. You can't even directly compute the bandwidth it has at a given clock rate without knowing more about the protocol because it does not have seperate address/data lines -- it uses a packet protocol on the seperate read and write busses. This is a very different animal, don't apply what you know from MPX (or even the AMD & Intel busses) to the 970's FSB. It is more akin to HyperTransport or RapidIO.
My guess is that the rumour is bogus. The benchmarks could be bogus too, we simply don't know at this point.
Thanks programmer. A simple straightforward explanation that actually almost makes sense to me. Sounds like when the 970 does come out, we will all have to get used to a new way of thinking about how it all works together.
Originally posted by kupan787
It is 200MHz, but double pumped so it is 400MHz effective. So if you were thinking the 400MHz was doubled to 800, you were wrong. However if you were quoting the effective figure, you were right
Maybe I'm misunderstanding you but PC3200 memory IS 400mhz. DDR400. Double pumped (as in the new P4 chipset) makes it 800mhz effective.
Perhaps the most obvious problem with the benchmarks was the fact that Bryce, which does not use multiple processors, sped up substantially using a dual. There are numerous other somewhat questionable aspects to the benchmarks.
The 80% figure is an educated guess based on the speculative SPEC benchmarks that IBM posted.
There are hundreds of different metrics that can be used to compare processors. IBM has almost certainly advertised its strongest rating, so I'm assuming that it's probably somewhat weaker in the non-advertised aspects and in a real world situation (i.e. Apple is not going to supply unlimited cache, fastest possible memory, etc.). The p4 with 800 MHz front side bus is a pretty hefty creature and is significantly faster than the previous P4.
I'll show where I got the 1% figure (although it is a guesstimate based on relative expenditures, since IBM doesn't break out microprocessor research by chip type) in another post when I have the time. Suffice it to say, the 970 was done very cheaply.
This is a good thing, by the way. If IBM had spent a billion on research, just amortizing the R&D would make the chip cost $500 before even getting to production expenses. If Apple is going to use a few million a year of these, they'd better not have spent more than $100 million on chip R+D. Compare this to the 10 billion that Intel has spent over the last few years (roughly $4 billion/year).
Originally posted by Tom West
How do we know it the benchmarks are fake?
Perhaps the most obvious problem with the benchmarks was the fact that Bryce, which does not use multiple processors, sped up substantially using a dual. There are numerous other somewhat questionable aspects to the benchmarks.
I'm not saying I believe the benchmarks but I haven't seen a really compelling reason why they are blatantly false. Most of the reasons are just as weak as the benchmarks. They offered the explanation that a more recent (beta) version of Bryce was used -- given that the rumour stated that Apple was doing these benchmarks this is entirely plausible since Apple commonly has pre-release versions of software for testing on new hardware.
The 80% figure is an educated guess based on the speculative SPEC benchmarks that IBM posted.
There are hundreds of different metrics that can be used to compare processors. IBM has almost certainly advertised its strongest rating, so I'm assuming that it's probably somewhat weaker in the non-advertised aspects and in a real world situation (i.e. Apple is not going to supply unlimited cache, fastest possible memory, etc.). The p4 with 800 MHz front side bus is a pretty hefty creature and is significantly faster than the previous P4.
IBM is usually fairly conservative with things like estimated SPECmarks for a processor they don't yet have silicon for. They've gotten in trouble with things like this before so I wouldn't necessarily discount their estimates, nor would I assume that Apple will come to market with a significantly weaker system. Looking at the processor's specifications, clock rate, and the performance of the POWER4 I'd say that the estimate were entirely reasonable. There were also Moki's comments about some sandbagging going on, and the various rumours of IBM being pleased with how the 970 has turned out.
On the other hand you're right that the new P4 is a beast. The 970 might only be 80% of its performance, but its possible that Apple will beat that number. We just don't have enough information to make a reasonable guess, but even 80% of the new Intel hardware is going to be a lot better than we've got right now!
I'll show where I got the 1% figure (although it is a guesstimate based on relative expenditures, since IBM doesn't break out microprocessor research by chip type) in another post when I have the time. Suffice it to say, the 970 was done very cheaply.
This is a good thing, by the way. If IBM had spent a billion on research, just amortizing the R&D would make the chip cost $500 before even getting to production expenses. If Apple is going to use a few million a year of these, they'd better not have spent more than $100 million on chip R+D. Compare this to the 10 billion that Intel has spent over the last few years (roughly $4 billion/year).
I think you're pretty much right on the money here. IBM is using some advanced circuit design tools to build the POWER series, and the 970 directly leverages the POWER4 work. This has already served IBM very well and it will allow them to continue to compete with Intel into the future on a far more cost effective basis. It may cost them some potential performance, but it might also allow them to focus on high level improvements rather than getting bogged down in detail improvements. If the tools they are using can stay on the bleeding edge it'll allow IBM to take advantage of new techniques & processes faster, but if the tools don't evolve quickly this could turn into a liability. The single biggest advantage in IBM's approach is that they should be able to build massive processors much more quickly than Intel can -- when chips reach into the hundreds of millions of transistors, IBM will be able to do more interesting things with that sooner. Unless, of course, Intel adopts the same strategy.