Ya, but doesn't moore's law also say that processor speed would double every 18 months? I could be wrong, I don't really know much about the law, I just heard that.</strong><hr></blockquote>
Yup, that's one way to look at it (see my original post for an edit to that effect). If we do look at it in this particular way (and ignore all the other bits and pieces) then we are right on track with Moore's Law at 1.25GHz.
I do believe that the original Moore's Law was simply an observance that the # of transistors on a chip was doubling every 18 months (correct me if I'm wrong).
I find it amusing that the generally accepted Moore's Law we have today is neither a) by Moore nor b) a law (it's just an observation)...
Yup, that's one way to look at it (see my original post for an edit to that effect). If we do look at it in this particular way (and ignore all the other bits and pieces) then we are right on track with Moore's Law at 1.25GHz.
I do believe that the original Moore's Law was simply an observance that the # of transistors on a chip was doubling every 18 months (correct me if I'm wrong).
I find it amusing that the generally accepted Moore's Law we have today is neither a) by Moore nor b) a law (it's just an observation)...
Ya, but doesn't moore's law also say that processor speed would double every 18 months? I could be wrong, I don't really know much about the law, I just heard that.</strong><hr></blockquote>
Moore's Law is something like "2x transistors every 18 months."
If it took 7 months to get from 1 GHz to 1.25
If you start from 1 GHz and add 25% compounded every 7 months, after 21 months you'd almost be at 2x the clock speed.
Seriously... think about it. iMacs still only 800 mhz? And we would expect PowerMacs to get to what? 1.4 ghz? <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
1) I am buying this new PowerMac and I will like it. For me these specs are fine, if there's FireWire 2. I have an iBook 500 and B&W G3 300.
2) Dartmouth is getting rid of lots of its Macs. While I'll buy this new Mac and like it, this new machine looks bad compared to PCs. Especially when the person doing the looking is stupid and doesn't know what "ease of use" "MacOS X" or "Dual Processors" are. Either they need to tell people through the Switch ads about the megahertz myth, or they need to convey that Macs are more than Mhz. They're easier, more productive, and friendly. Again, Dartmouth is dropping Macs for staff, and DeLorme dropped Street Atlas. Some key pieces are starting to move against Apple, like Motorola sucking, developers dropping Apple, and the Switch ads SUCKING. Apple needs better marketing. They need to tell average people what we already know. Dell users are used to being beat up when using a computer. They would LOVE the MacOS, especially OS X. They just don't know it, yet....Microsoft may have been right with their pre-MWNY FUD. Paging Apple Marketing...
I am not flaming against Apple. Don't flame me. I am just trying to show everyone here how ordinary people will and do see Apple.
3) Also, Apple is not going out of business I just thought that would be important to add... Since they have been through worse times. Somehow they always pull through, and although Steve Jobs seems to be completely oblivious and/or arrogant sometimes I feel good having an iCEO at Apple. PCs are faster at some mthings sure, but all that accomplishes: They can crash Windows even faster!
"Moore's (1st) Law" is that a microprocessor's transistor count doubles every 18 to 24 months. This is an observation based on manufacturing technology advancing every 18 to 24 months or so, typically a reduction of 30% (or 50% in acreage), in gate, transistor, line length, etc.
Unfortunately, "Moore's (2nd) Law" is always ignored but is going to rear its ugly head pretty soon. This observation is that the cost of implementing the next CMOS manufacturing technology will exceed the amount of revenue the prospective chips it manufactures. That is, it so expensive to build to plant that you can't recover the costs. For the 0.10, 0.07 and 0.05 CMOS technology, it looks like only Intel, maybe IBM, and a couple of East Asian conglomerates will be able to afford the R&D. Motorola will be hitting this wall soon, or probably hit it 2 years ago when they trailed the pack in getting to 0.18u.
Moore said nothing about clock rates. Many of you are falling into the MHz = performance game. It gets very tiring. MHz is half of the performance equation. IPC (or CPI), how many instructions per cycle, is the other half. The two together makes:
G4 IPC ~= 1.3 x P3 IPC ~= 1.7 * P4 IPC ~= 1.1 x Athlon IPC (or something thereabouts)
In other words, Apple needs about 1.5 GHz G4 to be competitive. They would have it if Motorola would produce a 0.13 micron G4. Intel, AMD and IBM have been shipping 0.13 parts for a long while now...
Smart people: do two 1Ghz processors mean the speed of a 2Ghz processor? BRussel mentioned that without proper threading apps couldn't take advantage of Dual processors. I thought that was one of the wonders of OS X? It takes care of this problem right? I just want to know if ordinary apps like Office X, IE, games, iApps, would see two 1Ghz processors perform as well as one 2Ghz processor. Thanks. Programmer?
<strong>Moore's law says that transitors will grow exponentially at a base of 10 (10x) every year, I believe. We're no where near that with the G4.</strong><hr></blockquote>
Actually no. I took the effort and went to the source of all knowledge (google) and found that "Moore's Law" was the observation of "doubling of transistor density on a manufactured die every year." However, the progress has since slowed to about 2x every 18 months so the law was updated by Moore.
The source for this? Dr. Gordon Moore, so I think we can trust it
If transistor count had been going up at 10x per year since 1965 (when the law was first observed), and *if* we had 1 transistor in 1965, we'd have 100,000,000,000,000,000,000,000,000,000,000 transistors on a chip today.
<strong>Man this is gonna hurt, I just hope this transition phase isn't long.</strong><hr></blockquote>
Well, I don't think it's as bad as that, but if the rumors are accurate, it means that this isn't what Apple wanted.
Duals across the board last time meant that something got f'd up. I suspect that the same holds true today. I'm suspicious of true DDR and suspect that we'll get 166MHz FSB and Xserve style DDR and have to be content with that. We may get some gimmicks to make these perform faster that the numbers might suggest, though.
But duals across the board means that Apple is making the most of a bad situation. We won't see anything revolutionary out of a press release, we won't see true DDR.
Go back and reread moki's posts over the last month. Apple didn't pull the big system together. They're giving us something less. Tune back in March, or thereabouts.
Don't assume that I'm bitter or feel that you should be. Processor speed is highly overrated. In almost 20 years of doing this stuff, slow CPUs have hardly gotten my attention. But crap software and operating systems will suck your life away. Poor integration, systems locked up such that users can't do what they need, simply bad code. Mac OS X is the jewel here. Focus on it. If you actually do work with your Mac, it'll deliver more than a 2GHz G4 will.
Apple will never lose the user base that realize what the true benefits of a computing experience are.
<strong>Smart people: do two 1Ghz processors mean the speed of a 2Ghz processor? BRussel mentioned that without proper threading apps couldn't take advantage of Dual processors. I thought that was one of the wonders of OS X? It takes care of this problem right? I just want to know if ordinary apps like Office X, IE, games, iApps, would see two 1Ghz processors perform as well as one 2Ghz processor. Thanks. Programmer? </strong><hr></blockquote>
Short answers: No with a but..., Long Answer: Yes with an if...
In particular: Office X: I wasn't aware this was slow in the first place? IE: ditto, window resizing is the domain of the window manager and 10.2 should fix that; games: definately not. Vast majority are single threaded; iApps: Don't know of any massively multi-threaded iApps.
Keep in mind that, even in the perfect implementation, a multi-threaded app running on a 2x1GHz system will not be 2x as fast as a 1x system. And most apps are not perfect.
Smart people: do two 1Ghz processors mean the speed of a 2Ghz processor? BRussel mentioned that without proper threading apps couldn't take advantage of Dual processors. I thought that was one of the wonders of OS X? It takes care of this problem right? I just want to know if ordinary apps like Office X, IE, games, iApps, would see two 1Ghz processors perform as well as one 2Ghz processor. Thanks. Programmer? </strong>
If you run a lot of apps concurrently, x MHz dual processors will approximate 2x MHx single processor pretty well. It may even run smoother due to load balancing issues.
If you run one app that is properly threaded, x MHz dual processors = 1.8 * 2x MHz single processor.
If you run one app that is single threaded, then the app can only run on one processor. The 2x MHz processor will run that app 2x as fast the x MHz dual processor.
So if you run a lot of apps concurrently like Office X, IE, iApps, dual processors will do ok compared to single processor with 2x MHz. If you mostly run one app at the time, 2x MHz single processor is usually the way to go.
A dual 450 or dual 500 MHz G4 with 512+ MB of RAM would be very nice machines for typical office automation and web usage. A silent dual G4 Cube would be so nice in my bedroom...
I'm suspicious of true DDR and suspect that we'll get 166MHz FSB and Xserve style DDR and have to be content with that. We may get some gimmicks to make these perform faster that the numbers might suggest, though.</strong>
You know, if RAM companies made PC166 SDRAM, a prospective 166 MHz MPX FSB + PC166 SDRAM would perform about the same as (or perhaps even better than) a 266 "MHz" FSB + PC2100 DDR SDRAM. It wouldn't be bad solution. Now, if Apple can ship a 166 MHz bus this soon, I think they are doing well.
<strong>But duals across the board means that Apple is making the most of a bad situation. We won't see anything revolutionary out of a press release, we won't see true DDR.</strong>
Nothing spectacular certainly. But Apple should be shipping duals in every desktop machine in any case.
<strong>Processor speed is highly overrated. In almost 20 years of doing this stuff, slow CPUs have hardly gotten my attention. But crap software and operating systems will suck your life away. Poor integration, systems locked up such that users can't do what they need, simply bad code.</strong>
That and the American airline industry are embarrassments to humity! No, the American airline industry is worse.
Thanks for clearing that up THT, johnsonwax, and JustAGuy. Nice posts!
So are most developers taking advantage of Duals? I would think NOT. Since most Macs out there don't have them. Too bad Apple can't invent a way for OS X to make up for the lack of good thread programming of developers! Is it hard to take advantage of Duals?
<strong>As far as PowerMacs go, all those specs are doable, we just need a G4 that does DDR - the mythical 7470.</strong><hr></blockquote>
Given that MOT was supposed to just be delivering 1.4ghz units to Apple in August (and who knows if they'll make it on time), I'd be pretty surprised if that was accurate at all. Maybe if they announce the machines, but they aren't available for a bit (especially the high end models).
actually more things are multi threaded than you think. how well it is implemented i have no idea. but go to the terminal and type top the look at what ever application running and it tells you the number of threads in one of the columns
Comments
<strong>
Ya, but doesn't moore's law also say that processor speed would double every 18 months? I could be wrong, I don't really know much about the law, I just heard that.</strong><hr></blockquote>
Yup, that's one way to look at it (see my original post for an edit to that effect). If we do look at it in this particular way (and ignore all the other bits and pieces) then we are right on track with Moore's Law at 1.25GHz.
I do believe that the original Moore's Law was simply an observance that the # of transistors on a chip was doubling every 18 months (correct me if I'm wrong).
I find it amusing that the generally accepted Moore's Law we have today is neither a) by Moore nor b) a law (it's just an observation)...
[ 08-13-2002: Message edited by: JustAGuy ]</p>
<strong>
Yup, that's one way to look at it (see my original post for an edit to that effect). If we do look at it in this particular way (and ignore all the other bits and pieces) then we are right on track with Moore's Law at 1.25GHz.
I do believe that the original Moore's Law was simply an observance that the # of transistors on a chip was doubling every 18 months (correct me if I'm wrong).
I find it amusing that the generally accepted Moore's Law we have today is neither a) by Moore nor b) a law (it's just an observation)...
[ 08-13-2002: Message edited by: JustAGuy ]</strong><hr></blockquote>
Ya, I don't really know if it is processor or transistors...
<img src="graemlins/hmmm.gif" border="0" alt="[Hmmm]" />
<strong>
Ya, but doesn't moore's law also say that processor speed would double every 18 months? I could be wrong, I don't really know much about the law, I just heard that.</strong><hr></blockquote>
Moore's Law is something like "2x transistors every 18 months."
If it took 7 months to get from 1 GHz to 1.25
If you start from 1 GHz and add 25% compounded every 7 months, after 21 months you'd almost be at 2x the clock speed.
<strong>Thanks about the Yikes info.
So what do you think they will name the new case?</strong><hr></blockquote>
Maybe the G4 Dual 867 Holy Mary Mother of God?
Depending of course on mobo inventory...
MSKR
<strong>I don't think Moore considered Moto when he made up his law
Moto is the exception that proves the rule.
1) I am buying this new PowerMac and I will like it. For me these specs are fine, if there's FireWire 2. I have an iBook 500 and B&W G3 300.
2) Dartmouth is getting rid of lots of its Macs. While I'll buy this new Mac and like it, this new machine looks bad compared to PCs. Especially when the person doing the looking is stupid and doesn't know what "ease of use" "MacOS X" or "Dual Processors" are. Either they need to tell people through the Switch ads about the megahertz myth, or they need to convey that Macs are more than Mhz. They're easier, more productive, and friendly. Again, Dartmouth is dropping Macs for staff, and DeLorme dropped Street Atlas. Some key pieces are starting to move against Apple, like Motorola sucking, developers dropping Apple, and the Switch ads SUCKING. Apple needs better marketing. They need to tell average people what we already know. Dell users are used to being beat up when using a computer. They would LOVE the MacOS, especially OS X. They just don't know it, yet....Microsoft may have been right with their pre-MWNY FUD. Paging Apple Marketing...
I am not flaming against Apple. Don't flame me. I am just trying to show everyone here how ordinary people will and do see Apple.
3) Also, Apple is not going out of business
"Moore's (1st) Law" is that a microprocessor's transistor count doubles every 18 to 24 months. This is an observation based on manufacturing technology advancing every 18 to 24 months or so, typically a reduction of 30% (or 50% in acreage), in gate, transistor, line length, etc.
Unfortunately, "Moore's (2nd) Law" is always ignored but is going to rear its ugly head pretty soon. This observation is that the cost of implementing the next CMOS manufacturing technology will exceed the amount of revenue the prospective chips it manufactures. That is, it so expensive to build to plant that you can't recover the costs. For the 0.10, 0.07 and 0.05 CMOS technology, it looks like only Intel, maybe IBM, and a couple of East Asian conglomerates will be able to afford the R&D. Motorola will be hitting this wall soon, or probably hit it 2 years ago when they trailed the pack in getting to 0.18u.
Moore said nothing about clock rates. Many of you are falling into the MHz = performance game. It gets very tiring. MHz is half of the performance equation. IPC (or CPI), how many instructions per cycle, is the other half. The two together makes:
G4 IPC ~= 1.3 x P3 IPC ~= 1.7 * P4 IPC ~= 1.1 x Athlon IPC (or something thereabouts)
or
1 GHz G4 ~= 1.3 GHz P3 ~= 1.7 GHz P4 ~= 1.2 GHz Athlon (or something thereabouts)
In other words, Apple needs about 1.5 GHz G4 to be competitive. They would have it if Motorola would produce a 0.13 micron G4. Intel, AMD and IBM have been shipping 0.13 parts for a long while now...
[ 08-13-2002: Message edited by: THT ]</p>
<strong>Moore's law says that transitors will grow exponentially at a base of 10 (10x) every year, I believe. We're no where near that with the G4.</strong><hr></blockquote>
Actually no. I took the effort and went to the source of all knowledge (google) and found that "Moore's Law" was the observation of "doubling of transistor density on a manufactured die every year." However, the progress has since slowed to about 2x every 18 months so the law was updated by Moore.
The source for this? Dr. Gordon Moore, so I think we can trust it
If transistor count had been going up at 10x per year since 1965 (when the law was first observed), and *if* we had 1 transistor in 1965, we'd have 100,000,000,000,000,000,000,000,000,000,000 transistors on a chip today.
Now, I thought the PIV's 55 million was a lot...
<strong>Man this is gonna hurt, I just hope this transition phase isn't long.</strong><hr></blockquote>
Well, I don't think it's as bad as that, but if the rumors are accurate, it means that this isn't what Apple wanted.
Duals across the board last time meant that something got f'd up. I suspect that the same holds true today. I'm suspicious of true DDR and suspect that we'll get 166MHz FSB and Xserve style DDR and have to be content with that. We may get some gimmicks to make these perform faster that the numbers might suggest, though.
But duals across the board means that Apple is making the most of a bad situation. We won't see anything revolutionary out of a press release, we won't see true DDR.
Go back and reread moki's posts over the last month. Apple didn't pull the big system together. They're giving us something less. Tune back in March, or thereabouts.
Don't assume that I'm bitter or feel that you should be. Processor speed is highly overrated. In almost 20 years of doing this stuff, slow CPUs have hardly gotten my attention. But crap software and operating systems will suck your life away. Poor integration, systems locked up such that users can't do what they need, simply bad code. Mac OS X is the jewel here. Focus on it. If you actually do work with your Mac, it'll deliver more than a 2GHz G4 will.
Apple will never lose the user base that realize what the true benefits of a computing experience are.
<strong>Smart people: do two 1Ghz processors mean the speed of a 2Ghz processor? BRussel mentioned that without proper threading apps couldn't take advantage of Dual processors. I thought that was one of the wonders of OS X? It takes care of this problem right? I just want to know if ordinary apps like Office X, IE, games, iApps, would see two 1Ghz processors perform as well as one 2Ghz processor. Thanks. Programmer?
Short answers: No with a but..., Long Answer: Yes with an if...
In particular: Office X: I wasn't aware this was slow in the first place? IE: ditto, window resizing is the domain of the window manager and 10.2 should fix that; games: definately not. Vast majority are single threaded; iApps: Don't know of any massively multi-threaded iApps.
Keep in mind that, even in the perfect implementation, a multi-threaded app running on a 2x1GHz system will not be 2x as fast as a 1x system. And most apps are not perfect.
Smart people: do two 1Ghz processors mean the speed of a 2Ghz processor? BRussel mentioned that without proper threading apps couldn't take advantage of Dual processors. I thought that was one of the wonders of OS X? It takes care of this problem right? I just want to know if ordinary apps like Office X, IE, games, iApps, would see two 1Ghz processors perform as well as one 2Ghz processor. Thanks. Programmer?
If you run a lot of apps concurrently, x MHz dual processors will approximate 2x MHx single processor pretty well. It may even run smoother due to load balancing issues.
If you run one app that is properly threaded, x MHz dual processors = 1.8 * 2x MHz single processor.
If you run one app that is single threaded, then the app can only run on one processor. The 2x MHz processor will run that app 2x as fast the x MHz dual processor.
So if you run a lot of apps concurrently like Office X, IE, iApps, dual processors will do ok compared to single processor with 2x MHz. If you mostly run one app at the time, 2x MHz single processor is usually the way to go.
A dual 450 or dual 500 MHz G4 with 512+ MB of RAM would be very nice machines for typical office automation and web usage. A silent dual G4 Cube would be so nice in my bedroom...
I'm suspicious of true DDR and suspect that we'll get 166MHz FSB and Xserve style DDR and have to be content with that. We may get some gimmicks to make these perform faster that the numbers might suggest, though.</strong>
You know, if RAM companies made PC166 SDRAM, a prospective 166 MHz MPX FSB + PC166 SDRAM would perform about the same as (or perhaps even better than) a 266 "MHz" FSB + PC2100 DDR SDRAM. It wouldn't be bad solution. Now, if Apple can ship a 166 MHz bus this soon, I think they are doing well.
<strong>But duals across the board means that Apple is making the most of a bad situation. We won't see anything revolutionary out of a press release, we won't see true DDR.</strong>
Nothing spectacular certainly. But Apple should be shipping duals in every desktop machine in any case.
<strong>Processor speed is highly overrated. In almost 20 years of doing this stuff, slow CPUs have hardly gotten my attention. But crap software and operating systems will suck your life away. Poor integration, systems locked up such that users can't do what they need, simply bad code.</strong>
That and the American airline industry are embarrassments to humity!
So are most developers taking advantage of Duals? I would think NOT. Since most Macs out there don't have them. Too bad Apple can't invent a way for OS X to make up for the lack of good thread programming of developers! Is it hard to take advantage of Duals?
<strong>As far as PowerMacs go, all those specs are doable, we just need a G4 that does DDR - the mythical 7470.</strong><hr></blockquote>
Given that MOT was supposed to just be delivering 1.4ghz units to Apple in August (and who knows if they'll make it on time), I'd be pretty surprised if that was accurate at all. Maybe if they announce the machines, but they aren't available for a bit (especially the high end models).