Which only re-enforces what I've been saying about powermac prices.
If Apple wants to build a monster quad machine and charge $5000US or more for it, fine, but the single and dual configs need to get faster and cheaper simultaneously. Any increase in price will be as good as a speed drop on the CPU front, both things that Apple happens to be familiar with
I agree with you here. Apple MUST lower the price of single processor machines in order to remain competetive with the offerings from the x86 camp.
However, there is no economic competition that would motivate the pricing of a dual 970's to be marked at a lower price. Apple no longer needs to have dual processors to maintain "pairity" with the x86 camp because one 970 is as good as the latest offerings from Intel or AMD. Dual 970's are like what quad G4's would have been- a machine to dominate the wintel world with.
As much as I want Apple to release quad 970's, I don't see how this economically makes sense for Apple to do so on the first production run. It is in Apple's best interests to spread their available CPU's to as many customers as possible (this is also an argument against duals, although there is something to be said for having a dual 970 that just beats up on P4's). I would love to see a quad 970, but honestly, such a machine would be bandwidth starved and not the best bang for the buck when compared to dual 970's.
I agree with you here. Apple MUST lower the price of single processor machines in order to remain competetive with the offerings from the x86 camp.
However, there is no economic competition that would motivate the pricing of a dual 970's to be marked at a lower price. Apple no longer needs to have dual processors to maintain "pairity" with the x86 camp because one 970 is as good as the latest offerings from Intel or AMD. Dual 970's are like what quad G4's would have been- a machine to dominate the wintel world with.
As much as I want Apple to release quad 970's, I don't see how this economically makes sense for Apple to do so on the first production run. It is in Apple's best interests to spread their available CPU's to as many customers as possible (this is also an argument against duals, although there is something to be said for having a dual 970 that just beats up on P4's). I would love to see a quad 970, but honestly, such a machine would be bandwidth starved and not the best bang for the buck when compared to dual 970's.
I don't think a dual PPC970 will dominate dual Xeon systems from x86 makers. Basically, the PPC970 will bring us to closer parity with x86 offerings per CPU. A dual PPC970 would directly compete with dual Xeon systems while single PPC970 machines would compete with Prescott P4's and Athlon 64's. Frankly, I don't see much "ass kicking" with the PPC970, only "catching-up".
Unless Apple is agressive, x86 will still have the price/performance edge. Let's just hope the Apple premium isn't 200-400% as it is right now for equivalent systems.
umm why do people leave out what _may_ be included with the new systems.
kick ass system bus, super altivec, possible vector dsp do dadds- GPU - all of these things have to no real parity in machines on the intel side (prebuilt PCs anyway)
If there are apps that are going to run like a race car with a rocket attached due to some of these add ons then 50 spec points aint going to matter much.
all have been mentioned in rumor so its not a fact-- yet but then again we don't have the final specs on the 970 either
I don't think a dual PPC970 will dominate dual Xeon systems from x86 makers. Basically, the PPC970 will bring us to closer parity with x86 offerings per CPU. A dual PPC970 would directly compete with dual Xeon systems while single PPC970 machines would compete with Prescott P4's and Athlon 64's. Frankly, I don't see much "ass kicking" with the PPC970, only "catching-up".
Unless Apple is aggressive, x86 will still have the price/performance edge. Let's just hope the Apple premium isn't 200-400% as it is right now for equivalent systems.
I totally agree with the above
Quote:
Originally posted by keyboardf12
umm why do people leave out what _may_ be included with the new systems.
kick ass system bus, super altivec, possible vector dsp do dadds- GPU - all of these things have to no real parity in machines on the intel side (prebuilt PCs anyway)
If there are apps that are going to run like a race car with a rocket attached due to some of these add ons then 50 spec points aisn't going to matter much.
all have been mentioned in rumor so its not a fact-- yet but then again we don't have the final specs on the 970 either
I guess 2 more months and we will all know.
But we don't have that hyper-Whatever intel technology either. Tom's hardware has the Dual Xeon ahead of the Opertron, (or whatever it is 64-bit AMD), and the PC side has all the goodies. 8X AGP, Serial ATA, DDR 400, 4xQuad pumped FSB on Dual Xeon is hard to ignore.
Those 50 spec points will be more in range of 300 to 500 spec
points if the motherboard design is not state of the art. Apple used to innovate, and be an innovative Hardware manufacturer. We were the first to use USB for cripe sakes. wintel didn't adopt it for a long time. Now look at the picture. I just hope Apple realizes their message has gotten across to the other side, and can compete if they deliver the real goods like they used to.
agreed. but it seems all of those things will also be included in the 970 system.
less than 2 months to go
It seems that we would get that stuff, but I have not seen anything about the motherboard other than the 2 motherboards story that said one was single, and the other was a dual processor configuration. There was no word on any of the formentioned goodies, so I'm not going to think we are getting it until I see it.
I was just wondering how you guys are coming to these conclusions about the speed comparisons to the intel and AMD offerings. I don't really know enough about the specs of the 970 to fully understand. Are these truly informed guesses or wishfull thinking
I was just wondering how you guys are coming to these conclusions about the speed comparisons to the intel and AMD offerings. I don't really know enough about the specs of the 970 to fully understand. Are these truly informed guesses or wishfull thinking
Nothing wishful about what I said. Just call it an informed opinion, possibly "sand-bagging" to allow room for "Murphy's Law"
I was just wondering how you guys are coming to these conclusions about the speed comparisons...
IBM's provided some benchmarks of the 970. We don't know that Apple will ship anything remotely approaching whatever the hypothetical 'test machine' was. But they at least give you an idea of how much is 'nearly the best work' you can get out of the 970.
Here's a quote:
Quote:
Performance-wise, IBM believes the chip can record a benchmark of 932 on SPECint 2000 and a score of 1051 on SPECfp2000, both at 1.8-GHz. Peak SIMD GFLOPs should be about 14.4, Sandon said. Using Dhrystone MIPS, the chip should output a score of 5,220. or 2.9 DMIPS/MHz/. IBM expects the chip should test 18 million RC5 keys per second.
from Extremetech who got the numbers from an IBM presentation.
Benchmarks aren't worth the paper they're printed on... but these do show the 970 _creaming_ the G4. Cross-platform tests are tougher, as there's no 'SpecSIMD' to compare Altivec/SSE/whatever.
But the last time Apple's chip's spec marks were _this_ close to the competition, they were using Altivec to their advantage & running 'pentium-on-snail' commercials.
FYI, SPEC is a highly respected benchmark and the premier cross-platform benchmark in the microprocessor industry.
Here's some SPEC scores to compare to those of the PPC970 at 1.8GHz (SPECfp: 1050, SPECint: 932).
Also, 5220 score for Drystone isn't very impressive, specially at 1.8GHz. Intel's Pentium-M (aka Centrino) scores higher in both raw score and score/MHz.
FYI, SPEC is a highly respected benchmark and the premier cross-platform benchmark in the microprocessor industry.
But it doesn't do SIMD (understandably, given that SIMD is new and the implementations vary wildly). In practice, this severely limits its ability to measure the effectiveness of the G4, and also the 970. Especially if you're going to compare a G4 to a P4, "all else being equal" is an absurd assumption.
Basically, SPEC isn't going to tell you that FCP can do a lot of its rendering and effects in real time, or give you any indication of iDVD's encoding efficiency, etc.
Quote:
Here's some SPEC scores to compare to those of the PPC970 at 1.8GHz (SPECfp: 1050, SPECint: 932).
Also, 5220 score for Drystone isn't very impressive. Intel's Pentium-M scores higher in both raw score and score/MHz.
All the scores for the 970 are estimated, and none of them really matter until the production CPU is on a shipping board - you can't use theoretical power to do anything. Also, if the pulled IBM documents were right, and the 970 really is going to start at speeds up to 2.5GHz, the estimated scores have to be scaled up from 1.8GHz, which is the speed that IBM released the estimates for.
The 970 doesn't have to beat all comers in single CPU performance, though, because it's engineered to support SMP easily (unlike any kind of Pentium except for the pricey Xeon), and to be inexpensive and efficient. If a Pentium 4 is 20% faster than a given 970, but you can get two of those 970s for the price (and power consumption) of the P4, which will be faster in practice? And which one are you more likely to find in a notebook, or a small, silent workstation?
Again, while it's academically interesting to run "all else being equal" tests, they don't really matter that much in real terms outside of the PC commodity market. Going by SPEC, the fastest G4 should be 1/2 to 1/5 as fast as the fastest P4, but in practice, for a lot of reasons that SPEC can't account for, the speed advantage is rarely that dramatic.
Also, 5220 score for Drystone isn't very impressive, specially at 1.8GHz. Intel's Pentium-M (aka Centrino) scores higher in both raw score and score/MHz.
The Dhrystone benchmark is, in particular, quite useless and misleading. The Intel score is using a carefully optimized library with a compiler that knows about the benchmark and does things specifically to make it faster. This is not the normal IBM practice. The benchmark also has several versions which perform quite differently, and the source code is written to be CISC-friendly not RISC-friendly.
The SPECmark scores are probably more indicative, but I suspect IBM's numbers are rather conservative and these don't really portray a complete picture either.
The Dhrystone benchmark is, in particular, quite useless and misleading. The Intel score is using a carefully optimized library with a compiler that knows about the benchmark and does things specifically to make it faster. This is not the normal IBM practice. The benchmark also has several versions which perform quite differently, and the source code is written to be CISC-friendly not RISC-friendly.
The SPECmark scores are probably more indicative, but I suspect IBM's numbers are rather conservative and these don't really portray a complete picture either.
Huh? Most PC benchmark sites use SiSoft Sandra 2002 to benchmark Dhrystone, not some abscure binary to be manipulated with to achieve a particular outcome. If you'd have actually clicked on the link for the Pentium-M Dhrystone score, you'd have seen this.
3000+ XP Athlons also do quite a bit better than 5220. There isn't some Intel conspiracy for Dhrystone.
Actually the Dhrystone benchmark is a particularly poor one and easily manipulated. In fact virtually all major benchmarking software is manipulated. Graphics card manufacturers pull similar stunts with drivers, particularly nVidia who at one stage was recommending one set of drivers for benchmarking and one set for actual use.
SiSoft makes one of the worst and most overly theoretical benchmarking suites available too so don't try using their name to give it credence.
Anyway direct quote from the maker:
Quote:
Although the Dhrystone benchmark that I published in 1984 was useful at the time it cannot claim to be useful for modern workloads and CPUs because it is so short, it fits in on-chip caches, and fails to stress the memory system. Also, because it is so short and does not read from an input file, special compiler optimizations can benefit Dhrystone performance more than normal program performance.
There's really no question the only decent benchmark is actual program use.
You'll also find 2 things. Dhrystone is, like so many others, a near useless benchmark and is in fact largely dependent on clock speed alone, a rarity in the real world. Certain benchmarks IBM released are about 10 - 15% less than what actual launch values should be.
The PPC970 will compete fine with other offerings of the time. It's particularly aimed at high density and low cost MP configurations though and trust me when I say if you're in the market for a blade server you should seriously consider it.
Edit: Read this if you're interested in the Dhrystone benchmark and an idea of what confronts any benchmark in general.
Huh? Most PC benchmark sites use SiSoft Sandra 2002 to benchmark Dhrystone, not some abscure binary to be manipulated with to achieve a particular outcome. If you'd have actually clicked on the link for the Pentium-M Dhrystone score, you'd have seen this.
3000+ XP Athlons also do quite a bit better than 5220. There isn't some Intel conspiracy for Dhrystone.
Yes but that particular binary was compiled using the Intel or Microsoft compiler both of which include some benchmark optimizations. I did click on the link and I do know what executable they are using. I can also guarantee 100% that this is not exactly the same program used by IBM to test the 970 -- because its an x86 executable not a PowerPC executable. That means comparisons between its Dhrystone number and any Dhrystone number produced by IBM are completely meaningless. The fact that AMD's managed to out-do Intel's on that particular executable probably have more to do with the compiler options used to build it than anything else... and that too makes it pretty much meaningless on x86 since most software these days is being Pentium4 optimized which will tend to slow down the Athlons relative to the Intel design.
Is there any possibility at all that we could have at least one thread in this forum not devolve into this argument? It's not without merit, but it's not topical in this thread either.
I can assure everyone here that Dell's pricing structure has nothing to do with the clock speeds the 970 will be able to acheive initially.
no sh*t! we are not talking about dell h3ll, or anyone but apple, and the 970 in this thread.
Comments
Originally posted by Matsu
Which only re-enforces what I've been saying about powermac prices.
If Apple wants to build a monster quad machine and charge $5000US or more for it, fine, but the single and dual configs need to get faster and cheaper simultaneously. Any increase in price will be as good as a speed drop on the CPU front, both things that Apple happens to be familiar with
I agree with you here. Apple MUST lower the price of single processor machines in order to remain competetive with the offerings from the x86 camp.
However, there is no economic competition that would motivate the pricing of a dual 970's to be marked at a lower price. Apple no longer needs to have dual processors to maintain "pairity" with the x86 camp because one 970 is as good as the latest offerings from Intel or AMD. Dual 970's are like what quad G4's would have been- a machine to dominate the wintel world with.
As much as I want Apple to release quad 970's, I don't see how this economically makes sense for Apple to do so on the first production run. It is in Apple's best interests to spread their available CPU's to as many customers as possible (this is also an argument against duals, although there is something to be said for having a dual 970 that just beats up on P4's). I would love to see a quad 970, but honestly, such a machine would be bandwidth starved and not the best bang for the buck when compared to dual 970's.
Originally posted by Yevgeny
I agree with you here. Apple MUST lower the price of single processor machines in order to remain competetive with the offerings from the x86 camp.
However, there is no economic competition that would motivate the pricing of a dual 970's to be marked at a lower price. Apple no longer needs to have dual processors to maintain "pairity" with the x86 camp because one 970 is as good as the latest offerings from Intel or AMD. Dual 970's are like what quad G4's would have been- a machine to dominate the wintel world with.
As much as I want Apple to release quad 970's, I don't see how this economically makes sense for Apple to do so on the first production run. It is in Apple's best interests to spread their available CPU's to as many customers as possible (this is also an argument against duals, although there is something to be said for having a dual 970 that just beats up on P4's). I would love to see a quad 970, but honestly, such a machine would be bandwidth starved and not the best bang for the buck when compared to dual 970's.
I don't think a dual PPC970 will dominate dual Xeon systems from x86 makers. Basically, the PPC970 will bring us to closer parity with x86 offerings per CPU. A dual PPC970 would directly compete with dual Xeon systems while single PPC970 machines would compete with Prescott P4's and Athlon 64's. Frankly, I don't see much "ass kicking" with the PPC970, only "catching-up".
Unless Apple is agressive, x86 will still have the price/performance edge. Let's just hope the Apple premium isn't 200-400% as it is right now for equivalent systems.
kick ass system bus, super altivec, possible vector dsp do dadds- GPU - all of these things have to no real parity in machines on the intel side (prebuilt PCs anyway)
If there are apps that are going to run like a race car with a rocket attached due to some of these add ons then 50 spec points aint going to matter much.
all have been mentioned in rumor so its not a fact-- yet but then again we don't have the final specs on the 970 either
I guess 2 more months and we will all know.
Originally posted by Existence
I don't think a dual PPC970 will dominate dual Xeon systems from x86 makers. Basically, the PPC970 will bring us to closer parity with x86 offerings per CPU. A dual PPC970 would directly compete with dual Xeon systems while single PPC970 machines would compete with Prescott P4's and Athlon 64's. Frankly, I don't see much "ass kicking" with the PPC970, only "catching-up".
Unless Apple is aggressive, x86 will still have the price/performance edge. Let's just hope the Apple premium isn't 200-400% as it is right now for equivalent systems.
I totally agree with the above
Originally posted by keyboardf12
umm why do people leave out what _may_ be included with the new systems.
kick ass system bus, super altivec, possible vector dsp do dadds- GPU - all of these things have to no real parity in machines on the intel side (prebuilt PCs anyway)
If there are apps that are going to run like a race car with a rocket attached due to some of these add ons then 50 spec points aisn't going to matter much.
all have been mentioned in rumor so its not a fact-- yet but then again we don't have the final specs on the 970 either
I guess 2 more months and we will all know.
But we don't have that hyper-Whatever intel technology either. Tom's hardware has the Dual Xeon ahead of the Opertron, (or whatever it is 64-bit AMD), and the PC side has all the goodies. 8X AGP, Serial ATA, DDR 400, 4xQuad pumped FSB on Dual Xeon is hard to ignore.
Those 50 spec points will be more in range of 300 to 500 spec
points if the motherboard design is not state of the art. Apple used to innovate, and be an innovative Hardware manufacturer. We were the first to use USB for cripe sakes. wintel didn't adopt it for a long time. Now look at the picture. I just hope Apple realizes their message has gotten across to the other side, and can compete if they deliver the real goods like they used to.
you mean hyperthreading?
>and the PC side has all the goodies. 8X AGP, Serial ATA, DDR 400, 4xQuad pumped FSB on Dual Xeon is hard to ignore.
agreed. but it seems all of those things will also be included in the 970 system.
I'm talking about other things I mentioned not the stuff you can add today to PC which i agree can't be added to a mac.
I take it back... less than 2 months to go
Originally posted by keyboardf12
agreed. but it seems all of those things will also be included in the 970 system.
less than 2 months to go
It seems that we would get that stuff, but I have not seen anything about the motherboard other than the 2 motherboards story that said one was single, and the other was a dual processor configuration. There was no word on any of the formentioned goodies, so I'm not going to think we are getting it until I see it.
As for Quads, I'd really like to see a 4 CPU "ProMac" for the high end workstation market, with a Quad xServe for the clustering model in that line.
Apple needs to exceed, not match current Wintel performance and this is their chance to do it.
...
Originally posted by Aphelion
While I voted with the majority here for 1.4 to 1.8 with some duals, I'd not be suprised to see the top end break 2 GHz.
As for Quads, I'd really like to see a 4 CPU "ProMac" for the high end workstation market, with a Quad xServe for the clustering model in that line.
Apple needs to exceed, not match current Wintel performance and this is their chance to do it.
...
Apologies guys, Im very late to this party - its really rocking!.
I too am expecting the 970's to be SP 1.4 DP 1.6 and DP 1.8 for the top end model.
But like my learned colleague above, I would like Apple to release a 2GHz machine - pure psychological impact!.
Im working on an xserve solution and a quad cluster would be VERY welcome.
Great thread
But just to not letting myself being too disappointed I would bet no higher than 2GHz
Worst comes to worse....1.6GHz
Originally posted by chilleymac
I was just wondering how you guys are coming to these conclusions about the speed comparisons to the intel and AMD offerings. I don't really know enough about the specs of the 970 to fully understand. Are these truly informed guesses or wishfull thinking
Nothing wishful about what I said. Just call it an informed opinion, possibly "sand-bagging" to allow room for "Murphy's Law"
Originally posted by chilleymac
I was just wondering how you guys are coming to these conclusions about the speed comparisons...
IBM's provided some benchmarks of the 970. We don't know that Apple will ship anything remotely approaching whatever the hypothetical 'test machine' was. But they at least give you an idea of how much is 'nearly the best work' you can get out of the 970.
Here's a quote:
Performance-wise, IBM believes the chip can record a benchmark of 932 on SPECint 2000 and a score of 1051 on SPECfp2000, both at 1.8-GHz. Peak SIMD GFLOPs should be about 14.4, Sandon said. Using Dhrystone MIPS, the chip should output a score of 5,220. or 2.9 DMIPS/MHz/. IBM expects the chip should test 18 million RC5 keys per second.
from Extremetech who got the numbers from an IBM presentation.
Benchmarks aren't worth the paper they're printed on... but these do show the 970 _creaming_ the G4. Cross-platform tests are tougher, as there's no 'SpecSIMD' to compare Altivec/SSE/whatever.
But the last time Apple's chip's spec marks were _this_ close to the competition, they were using Altivec to their advantage & running 'pentium-on-snail' commercials.
Here's some SPEC scores to compare to those of the PPC970 at 1.8GHz (SPECfp: 1050, SPECint: 932).
Also, 5220 score for Drystone isn't very impressive, specially at 1.8GHz. Intel's Pentium-M (aka Centrino) scores higher in both raw score and score/MHz.
Originally posted by Existence
FYI, SPEC is a highly respected benchmark and the premier cross-platform benchmark in the microprocessor industry.
But it doesn't do SIMD (understandably, given that SIMD is new and the implementations vary wildly). In practice, this severely limits its ability to measure the effectiveness of the G4, and also the 970. Especially if you're going to compare a G4 to a P4, "all else being equal" is an absurd assumption.
Basically, SPEC isn't going to tell you that FCP can do a lot of its rendering and effects in real time, or give you any indication of iDVD's encoding efficiency, etc.
Here's some SPEC scores to compare to those of the PPC970 at 1.8GHz (SPECfp: 1050, SPECint: 932).
Also, 5220 score for Drystone isn't very impressive. Intel's Pentium-M scores higher in both raw score and score/MHz.
All the scores for the 970 are estimated, and none of them really matter until the production CPU is on a shipping board - you can't use theoretical power to do anything. Also, if the pulled IBM documents were right, and the 970 really is going to start at speeds up to 2.5GHz, the estimated scores have to be scaled up from 1.8GHz, which is the speed that IBM released the estimates for.
The 970 doesn't have to beat all comers in single CPU performance, though, because it's engineered to support SMP easily (unlike any kind of Pentium except for the pricey Xeon), and to be inexpensive and efficient. If a Pentium 4 is 20% faster than a given 970, but you can get two of those 970s for the price (and power consumption) of the P4, which will be faster in practice? And which one are you more likely to find in a notebook, or a small, silent workstation?
Again, while it's academically interesting to run "all else being equal" tests, they don't really matter that much in real terms outside of the PC commodity market. Going by SPEC, the fastest G4 should be 1/2 to 1/5 as fast as the fastest P4, but in practice, for a lot of reasons that SPEC can't account for, the speed advantage is rarely that dramatic.
Originally posted by Existence
Also, 5220 score for Drystone isn't very impressive, specially at 1.8GHz. Intel's Pentium-M (aka Centrino) scores higher in both raw score and score/MHz.
The Dhrystone benchmark is, in particular, quite useless and misleading. The Intel score is using a carefully optimized library with a compiler that knows about the benchmark and does things specifically to make it faster. This is not the normal IBM practice. The benchmark also has several versions which perform quite differently, and the source code is written to be CISC-friendly not RISC-friendly.
The SPECmark scores are probably more indicative, but I suspect IBM's numbers are rather conservative and these don't really portray a complete picture either.
Originally posted by Programmer
The Dhrystone benchmark is, in particular, quite useless and misleading. The Intel score is using a carefully optimized library with a compiler that knows about the benchmark and does things specifically to make it faster. This is not the normal IBM practice. The benchmark also has several versions which perform quite differently, and the source code is written to be CISC-friendly not RISC-friendly.
The SPECmark scores are probably more indicative, but I suspect IBM's numbers are rather conservative and these don't really portray a complete picture either.
Huh? Most PC benchmark sites use SiSoft Sandra 2002 to benchmark Dhrystone, not some abscure binary to be manipulated with to achieve a particular outcome. If you'd have actually clicked on the link for the Pentium-M Dhrystone score, you'd have seen this.
3000+ XP Athlons also do quite a bit better than 5220. There isn't some Intel conspiracy for Dhrystone.
SiSoft makes one of the worst and most overly theoretical benchmarking suites available too so don't try using their name to give it credence.
Anyway direct quote from the maker:
Although the Dhrystone benchmark that I published in 1984 was useful at the time it cannot claim to be useful for modern workloads and CPUs because it is so short, it fits in on-chip caches, and fails to stress the memory system. Also, because it is so short and does not read from an input file, special compiler optimizations can benefit Dhrystone performance more than normal program performance.
There's really no question the only decent benchmark is actual program use.
You'll also find 2 things. Dhrystone is, like so many others, a near useless benchmark and is in fact largely dependent on clock speed alone, a rarity in the real world. Certain benchmarks IBM released are about 10 - 15% less than what actual launch values should be.
The PPC970 will compete fine with other offerings of the time. It's particularly aimed at high density and low cost MP configurations though and trust me when I say if you're in the market for a blade server you should seriously consider it.
Edit: Read this if you're interested in the Dhrystone benchmark and an idea of what confronts any benchmark in general.
Originally posted by Existence
Huh? Most PC benchmark sites use SiSoft Sandra 2002 to benchmark Dhrystone, not some abscure binary to be manipulated with to achieve a particular outcome. If you'd have actually clicked on the link for the Pentium-M Dhrystone score, you'd have seen this.
3000+ XP Athlons also do quite a bit better than 5220. There isn't some Intel conspiracy for Dhrystone.
Yes but that particular binary was compiled using the Intel or Microsoft compiler both of which include some benchmark optimizations. I did click on the link and I do know what executable they are using. I can also guarantee 100% that this is not exactly the same program used by IBM to test the 970 -- because its an x86 executable not a PowerPC executable. That means comparisons between its Dhrystone number and any Dhrystone number produced by IBM are completely meaningless. The fact that AMD's managed to out-do Intel's on that particular executable probably have more to do with the compiler options used to build it than anything else... and that too makes it pretty much meaningless on x86 since most software these days is being Pentium4 optimized which will tend to slow down the Athlons relative to the Intel design.
Originally posted by Amorph
Is there any possibility at all that we could have at least one thread in this forum not devolve into this argument? It's not without merit, but it's not topical in this thread either.
I can assure everyone here that Dell's pricing structure has nothing to do with the clock speeds the 970 will be able to acheive initially.
no sh*t! we are not talking about dell h3ll, or anyone but apple, and the 970 in this thread.