Someone said that the French rumors are pertaining to the server products of Apple's not the desktops. If you reread the rumors, it makes a lot of sense. The supposed G5 probably requires a new process to make it run efficiently enough to be put in low profile rack-mount and why would the server products be OS9 bootable?
And since the expo days are focusing on XServe, and I'm sure the server technology will lag behind desktops when it comes to processors, the rumors seem quite believable in this context.
I don't think the current POWER4 would make it into any Apple machine, unless they decide to jump into the high end server space -- even then it doesn't have AltiVec so would be a little surprised at them using it at all. On the other hand, we don't really know how long they've been talking to IBM or how long IBM has been working on a SIMD equipped POWERx or PowerPC chip that Apple could use. They may have started yesterday or 2 years ago.
Here's a thought I haven't seen mentioned before: what if Apple's new G5 is just about ready (say late this year) but its on the 0.13 process and is just too big, too hot, and too expensive for a desktop machine. They are building U2/U3 servers, so perhaps they'd but this big beast of a chip into one of those in order to get it out into a market. When the better process version is available they can migrate the desktop to that. In the meantime the 0.13 version of the G4, hopefully with some improvements, will carry the PowerMac lineup.
[quote]Someone said that the French rumors are pertaining to the server products of Apple's not the desktops. If you reread the rumors, it makes a lot of sense.<hr></blockquote>
I thought exactly the same thing. The servers are mentioned throughout the story and after reading it I thought "Well, they're talking about a G5 server." I also find it hard to believe that if the G5 "really exists and runs reliably," it will be another 18 months before production ramps up.
I also find it hard to believe that if the G5 "really exists and runs reliably," it will be another 18 months before production ramps up.</strong><hr></blockquote>
Of course, 18 months is Motorola's usual ramp up time with current technology.
i keep seeing mention of apple building 3RU servers, i find this funney. at the unveiling of the xserve all that was mentioned was a 4RU RAID for Q4. might i be misteaken, or are you guys just bouncing around more rumors.
Something from that French site that caught my eye, as translated by babel fish:
[quote]- For the first time of the history of APPLE, a machine is provided out of standard with spare parts (alim....)
<hr></blockquote>
Anyone know what this could mean? A machine with spare parts? Conforming to some unspecified standard? Could this mean that the mobo will be a standard mobo, and Apple will use only a specialized CPU? Is this even possible?
<strong> what if Apple's new G5 is just about ready (say late this year) but its on the 0.13 process and is just too big, too hot, and too expensive for a desktop machine. They are building U2/U3 servers, so perhaps they'd but this big beast of a chip into one of those in order to get it out into a market. When the better process version is available they can migrate the desktop to that.</strong><hr></blockquote>
Seems reasonable enough. Especially when you say 'too hot'.
G5 must have been in development for at least 2 years and, I suppose, the first problem with it was the bus and memory throughput. Both Motorolla and Apple had to solve the bus problem first. Now, they have the bus, they may find that G5 drains too much power.
<strong>Um, wasn't the G3 only in Pro Desktops for 18-20 months?</strong><hr></blockquote>
601 debuted March, 1994
603 April, 1995
603e April, 1996
G3 November, 1997
604 August, 1995
604e August,1996
G4 October, 1999
*thanks to everymac
Looks like new PowerPC iterations WERE roughly once a year at one point. Somebody's slipping. Even extending the pattern with the g4 date slip, the g5 and/or g4+ should be out this year in August or October. The g3, however, is long overdue.
The g3 debuted as the "Pro" cpu. I seem to recall some argument at the time about whether it was really better than the 604e. Something to do with floating point math, I think? Anyone remember the reason? I do know that my 9600 seemed much faster than my Wallstreet (although that may be harddrive related).
As a side note, did I get this right, the g3 sprang from the 603, and g4 from 604? If so, did the 601 turn into the 603 or 604? <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" />
To answer your question, the 601 was the daddy of three chips, the 602, 603, and 604.
Basically three chips targeted for different segments.
The 602 was intended for servers, but never sold well, so Motorola killed it. The 603 was trageted towards low end desktops and the 604 covered the high end. Both of these chips were a success.
You missed something aswell...
G4 7450 January, 01
As for the G3/604e debate...
The G3 decended from the 603e and was smaller and less power hungry than the 604e. This made it suited for use in laptops aswell as desktops.
Also, the G3 intially ran at lower clock speeds than the 604e, which is the source of some misconceptions about its performace.
Clock for clock, the G3 was significanly faster than the 604e because it could execute as many as three instructions per clock cycle, while the 604e could perform no more than two. The G3 also sat on a faster bus and used faster L2 cache.
This is why a 300MHz G3 was 30% faster than the 350MHz 604e.
I am not a processor guru but there are mounting similarities in PowerPC design lines.
After the 601 was replaced with the release of the 604 [Pro] chip and the 603 [consumer] chip the design line seemed to have split.
The 603 and the G3 are both low cost, low heat, high clockable, low power. The 603 was clock for clock about 1/2-2/3 the speed of a 604. My understanding is that the G3 surpassed the 604 mainly in its support for a larger backside cache over the 604's smaller inline cache support.
The 604 and G4 in contrast are higher heat processor, costs more to make, uses more power, but gives more bang for the buck in terms of clock speed. The 604 also supported multiprocessing as does the G4. Both have had massive scaling issues with clock speed and have gone through several rewrites. The G4 has changed this thought process a little based on the newest revisions providing lower heat and power demands.
The G5 seems to promise to "reunite" the PowerPC line's feature sets, or had promised until it rumors came out that it is running hotter than was expected couple years ago.
All together this information leads me to think that the G3 line will be milked as long as IBM can pump out more efficient versions but maybe not in the products we expect. The newest versions of the G3 look to be excellent candidates for non-Mac devices while Motorola seems to really want to get as much out of the G4 line as possible. With the G3 replaced by the G4 in all but one product line [iBook] and Apple pressing the G4 as important for Mac OS X and newer applications [FCP 3] it is clear that the G3's days are numbered in the iBook.
I expect that the G5 will hit the Power Mac line and trickle down as the pattern goes. The G4 will still therefore be the "standard" processor but unless Apple does release another device with the G3 it seems unlikely that it will continue to support it after the G5 release.
<strong>To answer your question, the 601 was the daddy of three chips, the 602, 603, and 604.
Basically three chips targeted for different segments.
The 602 was intended for servers, but never sold well, so Motorola killed it. The 603 was trageted towards low end desktops and the 604 covered the high end. Both of these chips were a success.</strong><hr></blockquote>
Actually the Server chip was the PPC 620
The 603 chip was low power and eventually it morphed into the G4.
Think of the G4 as a 603 processor with the superior fpu unit of the 604+Altivec.
Remember the BiCimos Processor the X702<sp) that heat monger! Thank God the G3 turned out better than expected.
Here's a thought I haven't seen mentioned before: what if Apple's new G5 is just about ready (say late this year) but its on the 0.13 process and is just too big, too hot, and too expensive for a desktop machine. They are building U2/U3 servers, so perhaps they'd but this big beast of a chip into one of those in order to get it out into a market. When the better process version is available they can migrate the desktop to that. In the meantime the 0.13 version of the G4, hopefully with some improvements, will carry the PowerMac lineup.</strong><hr></blockquote>
The Programmer you seems to say that the G5 will be as huge as the power 4 (which is currently buitl in 0,18 micron process). If you consider that he will be too huge too hot, on 0,13 micron process with SOI it means that he will have near than 100 millions transistors : a real monster
The 604e could retire 4 instructions per clock cycle (like the current G4), which is more than the G3. The G3 turned out to be faster in most cases because processors had started to out run memory speeds in a significant way and the G3 had a backside L2 cache, whereas the 604e only had the L1 caches. The 604e's FPU had a throughput on double precision math of 1/cycle, whereas the G3 could only do 1/2 cycles. Roughly the same speed for single precision, however.
The G4 started out as a G3 with the better FPU, AltiVec and bus optimizations. Over the last few revisions the pipelines have gotten longer and the design has become more superscalar -- to the point where it is now more so than the 604e was.
You're wrong saying the 604e had no L2 cache. The PowerMac 9600 used the 604e and had 512KB of L2 cache. The G3's advantage here was that its L2 cache was directly linked to the processor and ran faster.
Comments
And since the expo days are focusing on XServe, and I'm sure the server technology will lag behind desktops when it comes to processors, the rumors seem quite believable in this context.
Here's a thought I haven't seen mentioned before: what if Apple's new G5 is just about ready (say late this year) but its on the 0.13 process and is just too big, too hot, and too expensive for a desktop machine. They are building U2/U3 servers, so perhaps they'd but this big beast of a chip into one of those in order to get it out into a market. When the better process version is available they can migrate the desktop to that. In the meantime the 0.13 version of the G4, hopefully with some improvements, will carry the PowerMac lineup.
I thought exactly the same thing. The servers are mentioned throughout the story and after reading it I thought "Well, they're talking about a G5 server." I also find it hard to believe that if the G5 "really exists and runs reliably," it will be another 18 months before production ramps up.
<strong>
I also find it hard to believe that if the G5 "really exists and runs reliably," it will be another 18 months before production ramps up.</strong><hr></blockquote>
Of course, 18 months is Motorola's usual ramp up time with current technology.
<strong>Apple's senior director of hardware (Greg Joswiak) has been quoted as saying "the g4 still has a long life ahead of it.
</strong><hr></blockquote>
When he said this, he may have been referring to the G4's long life in iMacs and eMacs (and soon maybe iBooks).
[quote]- For the first time of the history of APPLE, a machine is provided out of standard with spare parts (alim....)
<hr></blockquote>
Anyone know what this could mean? A machine with spare parts? Conforming to some unspecified standard? Could this mean that the mobo will be a standard mobo, and Apple will use only a specialized CPU? Is this even possible?
<strong>Won't support OS9???
How the hell are any print professional supposed to keep up with anything now?
the amount of software, drivers, etc etc required before we move to X hasnt even been scratched.
I told my boss to hold off on buying a new machine because of the newer ones coming out soon.
Man, this sucks.
OSX is fine for consumers - but no print /graphic design professional office is gonna have anything to do with it.
UGH</strong><hr></blockquote>
Yes they will. Eventually Adobe programs won't work in 9 and will have features designers want. In turn print shops will be forced to use it.
Saying nobody will switch to nine is like saying PC based corporations will never switch over to 2000 or XP because drivers don't exist.
You should try to start working in X ASAP. I've run Ill 10 in X since it came out. No problems yet. Maybe you need new equipment?
<strong> what if Apple's new G5 is just about ready (say late this year) but its on the 0.13 process and is just too big, too hot, and too expensive for a desktop machine. They are building U2/U3 servers, so perhaps they'd but this big beast of a chip into one of those in order to get it out into a market. When the better process version is available they can migrate the desktop to that.</strong><hr></blockquote>
Seems reasonable enough. Especially when you say 'too hot'.
G5 must have been in development for at least 2 years and, I suppose, the first problem with it was the bus and memory throughput. Both Motorolla and Apple had to solve the bus problem first. Now, they have the bus, they may find that G5 drains too much power.
Good point, Programmer!
Ibrahimi and Gill have taken most of the profits and NOT reinvested in the company nearly as well as they should have.
Xpress is a prime example of resting on one's laurels.
ID is now making very strong inroads thanks to X compatibility.
<strong>Um, wasn't the G3 only in Pro Desktops for 18-20 months?</strong><hr></blockquote>
601 debuted March, 1994
603 April, 1995
603e April, 1996
G3 November, 1997
604 August, 1995
604e August,1996
G4 October, 1999
*thanks to everymac
Looks like new PowerPC iterations WERE roughly once a year at one point. Somebody's slipping. Even extending the pattern with the g4 date slip, the g5 and/or g4+ should be out this year in August or October. The g3, however, is long overdue.
The g3 debuted as the "Pro" cpu. I seem to recall some argument at the time about whether it was really better than the 604e. Something to do with floating point math, I think? Anyone remember the reason? I do know that my 9600 seemed much faster than my Wallstreet (although that may be harddrive related).
As a side note, did I get this right, the g3 sprang from the 603, and g4 from 604? If so, did the 601 turn into the 603 or 604? <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" />
Basically three chips targeted for different segments.
The 602 was intended for servers, but never sold well, so Motorola killed it. The 603 was trageted towards low end desktops and the 604 covered the high end. Both of these chips were a success.
You missed something aswell...
G4 7450 January, 01
As for the G3/604e debate...
The G3 decended from the 603e and was smaller and less power hungry than the 604e. This made it suited for use in laptops aswell as desktops.
Also, the G3 intially ran at lower clock speeds than the 604e, which is the source of some misconceptions about its performace.
Clock for clock, the G3 was significanly faster than the 604e because it could execute as many as three instructions per clock cycle, while the 604e could perform no more than two. The G3 also sat on a faster bus and used faster L2 cache.
This is why a 300MHz G3 was 30% faster than the 350MHz 604e.
[ 06-24-2002: Message edited by: Keeksy ]</p>
After the 601 was replaced with the release of the 604 [Pro] chip and the 603 [consumer] chip the design line seemed to have split.
The 603 and the G3 are both low cost, low heat, high clockable, low power. The 603 was clock for clock about 1/2-2/3 the speed of a 604. My understanding is that the G3 surpassed the 604 mainly in its support for a larger backside cache over the 604's smaller inline cache support.
The 604 and G4 in contrast are higher heat processor, costs more to make, uses more power, but gives more bang for the buck in terms of clock speed. The 604 also supported multiprocessing as does the G4. Both have had massive scaling issues with clock speed and have gone through several rewrites. The G4 has changed this thought process a little based on the newest revisions providing lower heat and power demands.
The G5 seems to promise to "reunite" the PowerPC line's feature sets, or had promised until it rumors came out that it is running hotter than was expected couple years ago.
All together this information leads me to think that the G3 line will be milked as long as IBM can pump out more efficient versions but maybe not in the products we expect. The newest versions of the G3 look to be excellent candidates for non-Mac devices while Motorola seems to really want to get as much out of the G4 line as possible. With the G3 replaced by the G4 in all but one product line [iBook] and Apple pressing the G4 as important for Mac OS X and newer applications [FCP 3] it is clear that the G3's days are numbered in the iBook.
I expect that the G5 will hit the Power Mac line and trickle down as the pattern goes. The G4 will still therefore be the "standard" processor but unless Apple does release another device with the G3 it seems unlikely that it will continue to support it after the G5 release.
<strong>To answer your question, the 601 was the daddy of three chips, the 602, 603, and 604.
Basically three chips targeted for different segments.
The 602 was intended for servers, but never sold well, so Motorola killed it. The 603 was trageted towards low end desktops and the 604 covered the high end. Both of these chips were a success.</strong><hr></blockquote>
Actually the Server chip was the PPC 620
The 603 chip was low power and eventually it morphed into the G4.
Think of the G4 as a 603 processor with the superior fpu unit of the 604+Altivec.
Remember the BiCimos Processor the X702<sp) that heat monger! Thank God the G3 turned out better than expected.
<strong>I
Here's a thought I haven't seen mentioned before: what if Apple's new G5 is just about ready (say late this year) but its on the 0.13 process and is just too big, too hot, and too expensive for a desktop machine. They are building U2/U3 servers, so perhaps they'd but this big beast of a chip into one of those in order to get it out into a market. When the better process version is available they can migrate the desktop to that. In the meantime the 0.13 version of the G4, hopefully with some improvements, will carry the PowerMac lineup.</strong><hr></blockquote>
The Programmer you seems to say that the G5 will be as huge as the power 4 (which is currently buitl in 0,18 micron process). If you consider that he will be too huge too hot, on 0,13 micron process with SOI it means that he will have near than 100 millions transistors : a real monster
:eek:
[ 06-24-2002: Message edited by: powerdoc ]</p>
The G4 started out as a G3 with the better FPU, AltiVec and bus optimizations. Over the last few revisions the pipelines have gotten longer and the design has become more superscalar -- to the point where it is now more so than the 604e was.
IIRC.
[code]
IBM Power2 Motorola 88000 bus design
| | |
| | |
------- ---------- PPC 601 ----------
| PPC 601v
IBM | |
64 bit | |
PPC | ------------------------------------- Exponential
| | | | | | |
| | | | | | |
RS64 Power3 PPC 602 PPC 603 PPC 604 PPC 620 x704
| | (embedded) PPC 603e PPC 604e
| | (only ) |
| | |
RS64-II | PPC 750 0.29u to 0.22u
| PPC 755
| |
| -----------------------
| | | |
| | PPC 750cx PPC 7400 0.18 micron
| | PPC 750cxe PPC 7410
| | | |
| Nintendo | |
Power4 Gekko | PPC 7450
| PPC 7455 0.18 micron SOI
|
PPC 750fx 0.13 micron
</pre><hr></blockquote>
[ 06-24-2002: Message edited by: Keeksy ]</p>