SonOfSylvanus
2004-05-25, 05:47
ArsTechnica article (http://arstechnica.com/reviews/004/software/intel-naming/schema-1.html) about Intel's upcoming new processor numbering system.
From the conclusion:
One thing is clear to me, and that's the fact that the processor numbers aren't doing anything to really clarify, well, much of anything. There are two fundamental problems: the desktop and the mobile line have their own numbering systems, but they overlap. The Pentium M rates a 7XX number, but so does the 1066MHz FSB, 2MB L2 cache 3.8 GHz Pentium 4!
The second, related problem, is that of double encoding. The architecture names are reflected in the numbers when they are already reflected in the product name. The only thing that makes the Mobile Pentium 4 a 5-class CPU is the fact that it's not a Celeron M (3XX) or a Pentium M (7XX). In this way, the model numbers just repeat what the product name already tells us. More worrisome, the process numbers tell us nothing about the underlying features. Consider the Pentium M 740 and 745. The former is a 1.73 GHz part on a 533MHz FSB, while the later is a 1.8 GHz CPU on a 400MHz FSB. Clearly, to make an informed decision, you're going to need the "feature list," which at the end of the day obviates the need for the processor numbers.
If you do your homework, you'll know this. But then that's the original problem: numbers which are spat out as if they mean something obvious. Unfortunately, these new round of numbers seem even less helpful than the MHz ratings themselves, and if you've been watching the CPU world from more than just the x86 angle, you know that this is meant as chagrin.
Jeez... I'm thankful for G3, G4 and G5.
Funny thing is, despite the MHz comparison being useless when comparing machines of different platforms, MHz comparison is a solid indicator of relative performance in the Apple world itself. As far as I know, every new processor that Apple incorporates into a new system has a higher MHz clock-speed than the ones it supersedes. So Apple has never had the problem of a potential customer wondering if a 1.25GHz G3 is faster than a 800MHz G4, the way one might think that a 2.6GHz Celeron is a better performer than a 2.0GHz Pentium 4. We will soon be in the ironic situation in which Apple is the only manufacturer perpetuating the MHz Myth.
In retrospect, Apple has done well to devise a product line (based on processor development out of its control) that, with each new Mac, shows an improvement in performance in line with an increase in numerical specifications: this is essential for us dumb consumers.
Bring on the Dual 3GHz G5! That will be, like, loads better than the Dual 2GHz G5 (and, boy, won't it be...)
From the conclusion:
One thing is clear to me, and that's the fact that the processor numbers aren't doing anything to really clarify, well, much of anything. There are two fundamental problems: the desktop and the mobile line have their own numbering systems, but they overlap. The Pentium M rates a 7XX number, but so does the 1066MHz FSB, 2MB L2 cache 3.8 GHz Pentium 4!
The second, related problem, is that of double encoding. The architecture names are reflected in the numbers when they are already reflected in the product name. The only thing that makes the Mobile Pentium 4 a 5-class CPU is the fact that it's not a Celeron M (3XX) or a Pentium M (7XX). In this way, the model numbers just repeat what the product name already tells us. More worrisome, the process numbers tell us nothing about the underlying features. Consider the Pentium M 740 and 745. The former is a 1.73 GHz part on a 533MHz FSB, while the later is a 1.8 GHz CPU on a 400MHz FSB. Clearly, to make an informed decision, you're going to need the "feature list," which at the end of the day obviates the need for the processor numbers.
If you do your homework, you'll know this. But then that's the original problem: numbers which are spat out as if they mean something obvious. Unfortunately, these new round of numbers seem even less helpful than the MHz ratings themselves, and if you've been watching the CPU world from more than just the x86 angle, you know that this is meant as chagrin.
Jeez... I'm thankful for G3, G4 and G5.
Funny thing is, despite the MHz comparison being useless when comparing machines of different platforms, MHz comparison is a solid indicator of relative performance in the Apple world itself. As far as I know, every new processor that Apple incorporates into a new system has a higher MHz clock-speed than the ones it supersedes. So Apple has never had the problem of a potential customer wondering if a 1.25GHz G3 is faster than a 800MHz G4, the way one might think that a 2.6GHz Celeron is a better performer than a 2.0GHz Pentium 4. We will soon be in the ironic situation in which Apple is the only manufacturer perpetuating the MHz Myth.
In retrospect, Apple has done well to devise a product line (based on processor development out of its control) that, with each new Mac, shows an improvement in performance in line with an increase in numerical specifications: this is essential for us dumb consumers.
Bring on the Dual 3GHz G5! That will be, like, loads better than the Dual 2GHz G5 (and, boy, won't it be...)