There is only one MDF per year so, based on the sampling frequency, you can't really drawn any conclusions from when the new processor will be ready.</strong>
Hehehe... IBM revealed the technical details of the Power4 at MPF in 1999. Power4 machines didn't ship until Fall of 2001! A mere 2 years later.
On the other hand, if this upcoming 64 bit desktop PPC from IBM is just a single core Power 4 with SIMD unit and 512KB L2 cache fabbed on CMOS9S, it could come out in a couple of months ready for a Q1 03 launch.
Cutting the Power4 down to 1 core, 512 KB L2 and fabbing on CMOS9S would reduce the die size to <100 sq mm, which would definitely be desktop die size affordable. If it's dual core and 512 KB L2, probably in the 130 sq mm range.
A chip like that may not be ready for PowerBook use but think of what a 130nm process G4 would offer. We would be looking at 1.2-1.5GHz (extrapolating from a 1.25GHz 180nm G4). For a PowerBook, this is not bad.
"Remember when Apple was offering a single-proc 733 and a dual 533, and the smart money was on the dual 533?"
I've tried to forget those dark years Amorph. The G4 was a and still is a struggling flop.
"Anyone who's seen just how silky OS X is on a DP machine"
Sorry, I'm lost here, Amorph, shouldn't 'X' be able to run 'silky' smooth on a single G4? Y'know, just 'XP' runs 'silky' and 'snappy' on any low end Athlon XP? A £50 chip tops.
"...would still be complaining if the towers had single CPUs."
What a crock of crap. A single G5, all things being vapourware, will destroy any dual G4 machine Apple puts out. Hands down. Maybe if we go quad G4...we'll get some 'snap', eh?
Mhz, smeghz. Performance is where its at. The Pentium 4 has it and the G4 doesn't. Photoshop, Lightwave, After Affects... It takes Apple's best machine costing almost twice the cost of a Pentium 2.5 to come trailing in all those apps. I don't like it. But that's the way it is. (And don't get me started on 'browser' performance...)
If 'X' was running on anything approaching the performance of a 2.8 Pentium 4...'X' would feel very silky indeed. It would have a dollop of snap too...running on a big '500' bus as opposed to a measly 167.
The mhz myth maybe had 'balls' when it was Pentium 3 and G3 at 500 and 450 mhz. The mhz and performance arguement for the G4 is over. It's time to give another cpu a shot at the champ's belt.
And of course, if each G4 had the same performance of a P4 at 2.8 gig...then 'really silky' would be an understatement.
Nothing gets Apple or Moto' off the hook here. The only reason Apple is all dual is because that's the best they've got.
It just so happens that it aint that good either.
"Even if the clock speed was double. If anything, the above-quote post just illustrates the shallowness of the whole cry for more MHz."
Crap. There's nothing shallow about the Pentium 4's might on Lightwave. Go look at some Tommy hardware baby benchmarks on the Lightwave for the P4 or XP Athlon. Benchmarks elsewhere sez any Athlon XP of 1.6 upwards owns the G4. Second, the top Pentium 4s destroy the fastest Athlon XP overclocked to crazy XP on Lightwave.
"If you want Lightwave to run fast,"
You run them on a Pentium 2.8. For now. That may change when the G5 ships.
"you want robust 64 bit FP support, which is not tied to MHz."
I never said it was tied to mhz. Any Power 4 will tell you that!
Lightwave will run fast on a Mac when Apple gets a 'robust' 64-bit G5 out the door. Until then, its a huff and puff '3rd place' for the 'power'Mac.
A chip like that may not be ready for PowerBook use but think of what a 130nm process G4 would offer. We would be looking at 1.2-1.5GHz (extrapolating from a 1.25GHz 180nm G4). For a PowerBook, this is not bad.</strong>
If the x86 world can use the Pentium 4 as a notebook processor, Apple can use this IBM 64 bit PPC as a notebook processor
Well, sooner or later, Motorola has to start fabbing processors on its 0.13u process, otherwise, they will start losing their bread and butter to MIPS and ARM. We can maybe expect it in fall of 2002, but who knows. They have been behind in fab (which is as important as or more important than processor design) since the 0.25 micron days...
Either way (IBM or Moto) will be fine, especially if Moto adds a second FP unit onto the 7455 design. Apple needs 0.13 micron processors. All this bus bandwidth talk is more boogeyman than real, and gets tiring.
<strong>I've tried to forget those dark years Amorph. The G4 was a and still is a struggling flop.</strong><hr></blockquote>
Ugh. Lemon, PLEASE. PLEASE stop with the unbelievable negativity. After reading your posts for months now it's becoming too much to bear, mostly because so much of it is so completely inaccurate or based on fiction. For God's sake give it a rest would you?
The G4 is a struggling flop? Yeah, right. Try a "struggling" 7 Gigaflops on my DP 1 GHz tower. Slow my ass. Wake up.
To sum up the rest of your post: Whine whine whine bitch bitch moan moan whine moan bitch complain complain cry whine cry cry puke.
Exactly!! Lemon BB. You need to get a reality check, or at least try to document your complaints. Show us how bad the G4 is. Links please. I suggest you start a new thread on it in current hardware though. Nobody here wants to hear this crap anymore.
Or do like I do, and simply skip over anything that is by 'the Bon'. It actually makes even long theads like this one more reasonable in length.
But, lets not drag this thread further off topic that it is getting with the economy/product release debates. This thread had stayed on target through beginning of page 12.
Just over a month until MPF. Who knows how long until we (if ever) get to see this new chip in a PM. What have we learned since the original announcement about MPF? Well, we have the Altivec support on 64bit PPC, and we have several people on this board saying that they have contacts/info/birds etc telling them this is for Apple. I myself have heard two different similar 'birds' saying the same.
But, what do we really know? Not much more. But 'tis only a month away.
I'm pretty sure this thread will get locked before then.
I'm still curious what the cogniscenti and the little birds think of my two-pronged IBM SIMD idea. A full-bore high end PPC and a SIMD-light PPC G3 derivative would leave a nice, comfortable middle spot for subsequent G4s. And I do think that Mot will not leave the picture entirely. There are some nice things on their roadmap too, and they'll look nicer once they're no longer asked to hold up Apple's top of the line.
Interestingly, it looks like both IBM and Mot are taking the fast lane to .09m processes.
[quote] I'm still curious what the cogniscenti and the little birds think of my two-pronged IBM SIMD idea. A full-bore high end PPC and a SIMD-light PPC G3 derivative would leave a nice, comfortable middle spot for subsequent G4s. And I do think that Mot will not leave the picture entirely. There are some nice things on their roadmap too, and they'll look nicer once they're no longer asked to hold up Apple's top of the line.
Interestingly, it looks like both IBM and Mot are taking the fast lane to .09m processes.
<hr></blockquote>
I guess that I would wonder how Apple would use all of these chips. Say Motorola and IBM switch roles, and the G4 runs into the consumer line, and the Power4 derivative runs into the powermac line. Where is the room for a SIMD-lite G3 that you propose.
I don't buy the common line that the Power4 lite will first appear in an "uberPro" box. The xServe is a server; the PowerMac is supposed to fill the role of the Pro Desktop machine. They are different markets, and I don't see athird market in the high end. I just don't.
There is no middle spot in my mind. Nor does Apple generate enough demand to maintain 2 high end chip providers.
The G4 is a good chip and will serve the consumer lines well. Look at these numbers from a photoshop benchmark developed in the Battlefront at Ars. Many of these guys are very Mac unfriendly, so this is a pretty unbiased presentation.
[code]
2x 2000 P4 Xeon 286
2x 1533 Athlon MP 285
2x 1466 Athlon XP 279
2x 1000 G4 DDR 10.2 267
2x 1000 G4 OS 9 260
2x 1000 G4 OS X 10.1.5 254
2340 P4 (overclock) 239
2400 P4 234
1548 Athlon XP 214
1667 Athlon XP 211
1x 1533 Athlon MP 197
2000 P4 Xeon 194
2x 533 G4 175
1800 P4 173
1508 Celeron (overclock) 167
1400 PIII Tualatin 160
2x 500 G4 OSX 152
2x 450 G4 OS9 151
1333 Athlon TBird 147
2x 450 G4 OSX 10.1.5 143
800 G4 Pbook OSX 1 MB L3 135
733 G4 (miro7) 134
667 G4 OS 9 no L3. 127
466 G4 OS 9 123
667 G4 OSX Pbook no L3 121
466 G4 OS X 133 MHz bus 112
450 G4 OSX 100 MHz bus 101
1000 Athlon TBird (PS6.01) 100
700 G3 iBook 74
</pre><hr></blockquote>
The G4 is right up there. I just don't see such a big performance deficit right now. I guess many people feel that a full SIMD-enabled Power4 derivative would be ideal for dealing with hammer, opteron etc. in the 64 bit OS space. Again, I'm not sure what the market would be for a G3 with pseudo-SIMD from IBM. iPod?
I imagine it will be another 8 months before we see this IBM G5 chip in a PowerMac. It could be only 6 or so but don't count on it. Sometime in may. When's the expo around then? This is my bet.
<strong>"Remember when Apple was offering a single-proc 733 and a dual 533, and the smart money was on the dual 533?"
I've tried to forget those dark years Amorph. The G4 was a and still is a struggling flop.</strong><hr></blockquote>
The G4 was NOT a 'flop'. Specifically in the year surronding intoduction. It was touted as pentium crushing - and it was _for_me_. You can whine about current performance all you like, but there was some indisputable truth to the 'export controlled' supercomputer business. Single-precision flops out the wazzu. It has taken until the last season or so for the x86 side to catch up (again, _for_me_).
[quote]<strong>What a crock of crap. A single G5, all things being vapourware, will destroy any dual G4 machine Apple puts out. Hands down. Maybe if we go quad G4...we'll get some 'snap', eh? </strong><hr></blockquote>
?
Even a prospective single CPU 2 GHz G5 will be given a run for its money by a Dual 1.2 GHz G4, what are you _ON_?
Some of the speculation about what a G5 would or could be are particularly contradictory to what you are saying - a dual _core_ is pretty much precisely the same as a dual CPU with less expensive overhead (in packaging & whatnot).
Sure, if you choose to deliberately ignore the other CPU or care about Apps that ignore the other CPU, that will indubitably be true. It sure isn't anywhere near true for me.
On another note, I'm hoping that AltiVec in the IBM chip is a distinct unit _and_ that there's dual double precision FPUs. More GFLOPS=better, I don't care if it maxxes out at 1Mhz so long as it is a numerical beast.
"I'm still curious what the cogniscenti and the little birds think of my two-pronged IBM SIMD idea. A full-bore high end PPC and a SIMD-light PPC G3 derivative would leave a nice, comfortable middle spot for subsequent G4s. And I do think that Mot will not leave the picture entirely. There are some nice things on their roadmap too, and they'll look nicer once they're no longer asked to hold up Apple's top of the line.
Interestingly, it looks like both IBM and Mot are taking the fast lane to .09m processes."
A lot of speculation has centered around the fact that this processor will be single core. What are the prospects of it actually being a dual core? The current POWER4 is a beast to say the least but how would you go about making it's die smaller?
Some ideas:
Cut out two thirds of the L2 cache
Eliminate the advanced GX bus in favor of smaller RIO buses
Eliminate the MCM bus because a workstation will not need this
Simplify the memory controller for exclusive DDR and/or DDR-II use
Keep it simple stupid: Dual core is enough for now
<strong>Well, sooner or later, Motorola has to start fabbing processors on its 0.13u process, otherwise, they will start losing their bread and butter to MIPS and ARM. </strong><hr></blockquote>
Motorola has said that they will be fabbing at 90nm by the end of the year (4 months ahwad of Intel). Remains to know what they will be fabbing.
As for performance enhancement, speed normally doubles for each sqrt(2) process shrinkage, so if they were to jump from 180nm to 90nm (2 steps - I know it sounds crazy), you could expect 4-5HGz G4s.
That's a good question, wormboy (and thanks for the benchmarks!). So, what could Apple do with three chips?
The whole point of using the Book E spec to implement SIMD - with register sharing, that is - is to keep the processor small, cool and cheap. The G3 is already very good at this. A SIMD G3 would do what the G3 currently does, only it would be marginally more capable at vector processing as well. Possible uses: low-end/educational machines, iBooks, any other portable form that Apple settles on where small size and battery life are more important than raw power. A Sahara fabbed on a .09m process could conceivably power a handheld.
The G4 can handle PowerBooks, iMacs, the XServe and other space-optimized servers.
The IBM processor could sit in a heretofore unseen level of PowerMac targeted at the Maya/Shake/UNIX workstation high-end, and in higher-end (by Apple's standards, not IBM's!) servers. That won't require too many processors per year, true, but then IBM will be using the processors in its own RS/6000 lines and elsewhere, and I'm sure they won't have any trouble finding other customers. Or it could end up in the PowerMac at current price points, in which case sales would probably shoot through the roof.
..."All this bus bandwidth talk is more boogeyman than real, and gets tiring."</strong><hr></blockquote>
That is interesting. I understand the MPX bus is very efficient, but in the "Altivec Programing Anyone" thread @ Arstechnica some very explicit posts showed that the bus was choked if the data could not be held in L2/3 cache?? correct??
[quote] A Sahara fabbed on a .09m process could conceivably power a handheld. <hr></blockquote>
That would be one scary fast handheld! But seriously, I could see something like this if Apple was to go back down the Newton road. But that's a BIG if. Isn't it?? [Apologies to Fran for my pessimism]
I would like to see where the DP 1.25 G4 PM would feature in this little league table.
I think the slowness is a perception problem, and it all stems from the GUI. Even after Quartz, my G3 450 on OS9 has a faster GUI than any current Mac using 10.2, and that is the problem. As a user experiance it just feels slow. I really don't understand why this is, if just about every other OS can have a lighting fast GUI, with the power of the G4 we should have one too.
It's strange but besides IE which I don't use any more almost everything is faster on 10.2 than 9 on my iMac G4 800. Ummm I don't really understand what people are going on about with this slow stuff. 10.1 was not slow by any means and 10.2 is quit fast. Just don't get these people complaining about the GUI being slow. oh well...
Comments
There is only one MDF per year so, based on the sampling frequency, you can't really drawn any conclusions from when the new processor will be ready.</strong>
Hehehe... IBM revealed the technical details of the Power4 at MPF in 1999. Power4 machines didn't ship until Fall of 2001! A mere 2 years later.
On the other hand, if this upcoming 64 bit desktop PPC from IBM is just a single core Power 4 with SIMD unit and 512KB L2 cache fabbed on CMOS9S, it could come out in a couple of months ready for a Q1 03 launch.
Cutting the Power4 down to 1 core, 512 KB L2 and fabbing on CMOS9S would reduce the die size to <100 sq mm, which would definitely be desktop die size affordable. If it's dual core and 512 KB L2, probably in the 130 sq mm range.
I've tried to forget those dark years Amorph. The G4 was a and still is a struggling flop.
"Anyone who's seen just how silky OS X is on a DP machine"
Sorry, I'm lost here, Amorph, shouldn't 'X' be able to run 'silky' smooth on a single G4? Y'know, just 'XP' runs 'silky' and 'snappy' on any low end Athlon XP? A £50 chip tops.
"...would still be complaining if the towers had single CPUs."
What a crock of crap. A single G5, all things being vapourware, will destroy any dual G4 machine Apple puts out. Hands down. Maybe if we go quad G4...we'll get some 'snap', eh?
Mhz, smeghz. Performance is where its at. The Pentium 4 has it and the G4 doesn't. Photoshop, Lightwave, After Affects... It takes Apple's best machine costing almost twice the cost of a Pentium 2.5 to come trailing in all those apps. I don't like it. But that's the way it is. (And don't get me started on 'browser' performance...)
If 'X' was running on anything approaching the performance of a 2.8 Pentium 4...'X' would feel very silky indeed. It would have a dollop of snap too...running on a big '500' bus as opposed to a measly 167.
The mhz myth maybe had 'balls' when it was Pentium 3 and G3 at 500 and 450 mhz. The mhz and performance arguement for the G4 is over. It's time to give another cpu a shot at the champ's belt.
And of course, if each G4 had the same performance of a P4 at 2.8 gig...then 'really silky' would be an understatement.
Nothing gets Apple or Moto' off the hook here. The only reason Apple is all dual is because that's the best they've got.
It just so happens that it aint that good either.
"Even if the clock speed was double. If anything, the above-quote post just illustrates the shallowness of the whole cry for more MHz."
Crap. There's nothing shallow about the Pentium 4's might on Lightwave. Go look at some Tommy hardware baby benchmarks on the Lightwave for the P4 or XP Athlon. Benchmarks elsewhere sez any Athlon XP of 1.6 upwards owns the G4. Second, the top Pentium 4s destroy the fastest Athlon XP overclocked to crazy XP on Lightwave.
"If you want Lightwave to run fast,"
You run them on a Pentium 2.8. For now. That may change when the G5 ships.
"you want robust 64 bit FP support, which is not tied to MHz."
I never said it was tied to mhz. Any Power 4 will tell you that!
Lightwave will run fast on a Mac when Apple gets a 'robust' 64-bit G5 out the door. Until then, its a huff and puff '3rd place' for the 'power'Mac.
Lemon Bon Bon <img src="graemlins/hmmm.gif" border="0" alt="[Hmmm]" />
[ 09-11-2002: Message edited by: Lemon Bon Bon ]
[ 09-11-2002: Message edited by: Lemon Bon Bon ]</p>
A chip like that may not be ready for PowerBook use but think of what a 130nm process G4 would offer. We would be looking at 1.2-1.5GHz (extrapolating from a 1.25GHz 180nm G4). For a PowerBook, this is not bad.</strong>
If the x86 world can use the Pentium 4 as a notebook processor, Apple can use this IBM 64 bit PPC as a notebook processor
Well, sooner or later, Motorola has to start fabbing processors on its 0.13u process, otherwise, they will start losing their bread and butter to MIPS and ARM. We can maybe expect it in fall of 2002, but who knows. They have been behind in fab (which is as important as or more important than processor design) since the 0.25 micron days...
Either way (IBM or Moto) will be fine, especially if Moto adds a second FP unit onto the 7455 design. Apple needs 0.13 micron processors. All this bus bandwidth talk is more boogeyman than real, and gets tiring.
<strong>I've tried to forget those dark years Amorph. The G4 was a and still is a struggling flop.</strong><hr></blockquote>
Ugh. Lemon, PLEASE. PLEASE stop with the unbelievable negativity. After reading your posts for months now it's becoming too much to bear, mostly because so much of it is so completely inaccurate or based on fiction. For God's sake give it a rest would you?
The G4 is a struggling flop? Yeah, right. Try a "struggling" 7 Gigaflops on my DP 1 GHz tower. Slow my ass. Wake up.
To sum up the rest of your post: Whine whine whine bitch bitch moan moan whine moan bitch complain complain cry whine cry cry puke.
ENOUGH. Take it somewhere else.
But, lets not drag this thread further off topic that it is getting with the economy/product release debates. This thread had stayed on target through beginning of page 12.
Just over a month until MPF. Who knows how long until we (if ever) get to see this new chip in a PM. What have we learned since the original announcement about MPF? Well, we have the Altivec support on 64bit PPC, and we have several people on this board saying that they have contacts/info/birds etc telling them this is for Apple. I myself have heard two different similar 'birds' saying the same.
But, what do we really know? Not much more. But 'tis only a month away.
I'm pretty sure this thread will get locked before then.
Interestingly, it looks like both IBM and Mot are taking the fast lane to .09m processes.
your not the guy in a different thread that was complaining about the g4 speed gap and yet owned an ibook?
maybe it was someone else...
Interestingly, it looks like both IBM and Mot are taking the fast lane to .09m processes.
<hr></blockquote>
I guess that I would wonder how Apple would use all of these chips. Say Motorola and IBM switch roles, and the G4 runs into the consumer line, and the Power4 derivative runs into the powermac line. Where is the room for a SIMD-lite G3 that you propose.
I don't buy the common line that the Power4 lite will first appear in an "uberPro" box. The xServe is a server; the PowerMac is supposed to fill the role of the Pro Desktop machine. They are different markets, and I don't see athird market in the high end. I just don't.
There is no middle spot in my mind. Nor does Apple generate enough demand to maintain 2 high end chip providers.
The G4 is a good chip and will serve the consumer lines well. Look at these numbers from a photoshop benchmark developed in the Battlefront at Ars. Many of these guys are very Mac unfriendly, so this is a pretty unbiased presentation.
[code]
2x 2000 P4 Xeon 286
2x 1533 Athlon MP 285
2x 1466 Athlon XP 279
2x 1000 G4 DDR 10.2 267
2x 1000 G4 OS 9 260
2x 1000 G4 OS X 10.1.5 254
2340 P4 (overclock) 239
2400 P4 234
1548 Athlon XP 214
1667 Athlon XP 211
1x 1533 Athlon MP 197
2000 P4 Xeon 194
2x 533 G4 175
1800 P4 173
1508 Celeron (overclock) 167
1400 PIII Tualatin 160
2x 500 G4 OSX 152
2x 450 G4 OS9 151
1333 Athlon TBird 147
2x 450 G4 OSX 10.1.5 143
800 G4 Pbook OSX 1 MB L3 135
733 G4 (miro7) 134
667 G4 OS 9 no L3. 127
466 G4 OS 9 123
667 G4 OSX Pbook no L3 121
466 G4 OS X 133 MHz bus 112
450 G4 OSX 100 MHz bus 101
1000 Athlon TBird (PS6.01) 100
700 G3 iBook 74
</pre><hr></blockquote>
The G4 is right up there. I just don't see such a big performance deficit right now. I guess many people feel that a full SIMD-enabled Power4 derivative would be ideal for dealing with hammer, opteron etc. in the 64 bit OS space. Again, I'm not sure what the market would be for a G3 with pseudo-SIMD from IBM. iPod?
[aside: are you paying attention Lemon BB??]
<strong>"Remember when Apple was offering a single-proc 733 and a dual 533, and the smart money was on the dual 533?"
I've tried to forget those dark years Amorph. The G4 was a and still is a struggling flop.</strong><hr></blockquote>
The G4 was NOT a 'flop'. Specifically in the year surronding intoduction. It was touted as pentium crushing - and it was _for_me_. You can whine about current performance all you like, but there was some indisputable truth to the 'export controlled' supercomputer business. Single-precision flops out the wazzu. It has taken until the last season or so for the x86 side to catch up (again, _for_me_).
[quote]<strong>What a crock of crap. A single G5, all things being vapourware, will destroy any dual G4 machine Apple puts out. Hands down. Maybe if we go quad G4...we'll get some 'snap', eh?
?
Even a prospective single CPU 2 GHz G5 will be given a run for its money by a Dual 1.2 GHz G4, what are you _ON_?
Some of the speculation about what a G5 would or could be are particularly contradictory to what you are saying - a dual _core_ is pretty much precisely the same as a dual CPU with less expensive overhead (in packaging & whatnot).
Sure, if you choose to deliberately ignore the other CPU or care about Apps that ignore the other CPU, that will indubitably be true. It sure isn't anywhere near true for me.
On another note, I'm hoping that AltiVec in the IBM chip is a distinct unit _and_ that there's dual double precision FPUs. More GFLOPS=better, I don't care if it maxxes out at 1Mhz so long as it is a numerical beast.
[ 09-11-2002: Message edited by: Nevyn ]</p>
Interestingly, it looks like both IBM and Mot are taking the fast lane to .09m processes."
Lemon Bon Bon
Some ideas:
Cut out two thirds of the L2 cache
Eliminate the advanced GX bus in favor of smaller RIO buses
Eliminate the MCM bus because a workstation will not need this
Simplify the memory controller for exclusive DDR and/or DDR-II use
Keep it simple stupid: Dual core is enough for now
Make on 130nm or 90nm process
These alone should make it managable.
<strong>Well, sooner or later, Motorola has to start fabbing processors on its 0.13u process, otherwise, they will start losing their bread and butter to MIPS and ARM. </strong><hr></blockquote>
Motorola has said that they will be fabbing at 90nm by the end of the year (4 months ahwad of Intel). Remains to know what they will be fabbing.
As for performance enhancement, speed normally doubles for each sqrt(2) process shrinkage, so if they were to jump from 180nm to 90nm (2 steps - I know it sounds crazy), you could expect 4-5HGz G4s.
The whole point of using the Book E spec to implement SIMD - with register sharing, that is - is to keep the processor small, cool and cheap. The G3 is already very good at this. A SIMD G3 would do what the G3 currently does, only it would be marginally more capable at vector processing as well. Possible uses: low-end/educational machines, iBooks, any other portable form that Apple settles on where small size and battery life are more important than raw power. A Sahara fabbed on a .09m process could conceivably power a handheld.
The G4 can handle PowerBooks, iMacs, the XServe and other space-optimized servers.
The IBM processor could sit in a heretofore unseen level of PowerMac targeted at the Maya/Shake/UNIX workstation high-end, and in higher-end (by Apple's standards, not IBM's!) servers. That won't require too many processors per year, true, but then IBM will be using the processors in its own RS/6000 lines and elsewhere, and I'm sure they won't have any trouble finding other customers. Or it could end up in the PowerMac at current price points, in which case sales would probably shoot through the roof.
<strong>
..."All this bus bandwidth talk is more boogeyman than real, and gets tiring."</strong><hr></blockquote>
That is interesting. I understand the MPX bus is very efficient, but in the "Altivec Programing Anyone" thread @ Arstechnica some very explicit posts showed that the bus was choked if the data could not be held in L2/3 cache?? correct??
[ 09-11-2002: Message edited by: rickag ]</p>
[quote] A Sahara fabbed on a .09m process could conceivably power a handheld. <hr></blockquote>
That would be one scary fast handheld! But seriously, I could see something like this if Apple was to go back down the Newton road. But that's a BIG if. Isn't it?? [Apologies to Fran for my pessimism]
2x 2000 P4 Xeon 286
2x 1533 Athlon MP 285
2x 1466 Athlon XP 279
2x 1000 G4 DDR 10.2 267
2x 1000 G4 OS 9 260
2x 1000 G4 OS X 10.1.5 254
2340 P4 (overclock) 239
2400 P4 234
1548 Athlon XP 214
1667 Athlon XP 211
1x 1533 Athlon MP 197
2000 P4 Xeon 194
2x 533 G4 175
1800 P4 173
1508 Celeron (overclock) 167
1400 PIII Tualatin 160
2x 500 G4 OSX 152
2x 450 G4 OS9 151
1333 Athlon TBird 147
2x 450 G4 OSX 10.1.5 143
800 G4 Pbook OSX 1 MB L3 135
733 G4 (miro7) 134
667 G4 OS 9 no L3. 127
466 G4 OS 9 123
667 G4 OSX Pbook no L3 121
466 G4 OS X 133 MHz bus 112
450 G4 OSX 100 MHz bus 101
1000 Athlon TBird (PS6.01) 100
700 G3 iBook 74
</pre><hr></blockquote>
[/qb]<hr></blockquote>
I would like to see where the DP 1.25 G4 PM would feature in this little league table.
I think the slowness is a perception problem, and it all stems from the GUI. Even after Quartz, my G3 450 on OS9 has a faster GUI than any current Mac using 10.2, and that is the problem. As a user experiance it just feels slow. I really don't understand why this is, if just about every other OS can have a lighting fast GUI, with the power of the G4 we should have one too.
[ 09-11-2002: Message edited by: Addison ]</p>