[quote] 2) The G4 is NOT topped out speed-wise. Adding SOI and a die shrink and copper interconnects will all add speed to the G4, probably well in excess of 1 GHz. <hr></blockquote>
Ok, so the Apollo will give us 933, 1.0ghz and 1.13ghz before it tops out. That's one speed bumped show. That's topped out to me.
[quote] Well, unlike the iMac, Apple don't even produce the G5 (Motorola do). And IIRC Moto gave technology presentations on both the original 7400 and the 7450 before Apple actually built systems based on them. <hr></blockquote>
So you're saying that if Steve says keep your mouth shut to Moto, they won't listen and they will release the G5 info conxcerning Apple? That's what I meant, that Apple can keep Moto quiet CIA style
As for the Apollo being for portables, it's only the same sentiment echoed by various people on various sites, so I can only form my opinion based on the info I have read.
Ergo, we have to state that there continues to be no good evidence of an imminent release of the G5, whereas the evidence behind the Apollo continues to mount.
"Man then proceeded to prove that black was white, and got run over at the next crosswalk."
[quote]Ergo, we have to state that there continues to be no good evidence of an imminent release of the G5, whereas the evidence behind the Apollo continues to mount. <hr></blockquote>
No one is disputing the fricking Apollo. The dispute is that the Apollo is intended for the iMac and the portables. No one has said the Apollo is for the towers and the G5 has been in production almost as long as the G4 (according to what i read).
Anyways, you believe what you want to believe and the rest of us will make up our own minds
[quote]i read more about g5 in two month than apollo in two years...<hr></blockquote>
50 rumors about the G5 and 1 statement from a MOTO rep saying that Apollo's were in production at .18mu (HIP6). Sometimes quality is better than quantity.
[quote]Ok, so the Apollo will give us 933, 1.0ghz and 1.13ghz before it tops out. That's one speed bumped show. That's topped out to me.<hr></blockquote>
If Apollo is just the 7450 with SOI process then we can extrapolate about 20% to 30% improvement with SOI process. The 7450 can clock at 933MHZ (people have o'clocked 867 to 1066, but have lost the L3 cache) so we can expect Apollo in 1.1GHZ to 1.3GHZ range if it is just a straight extrapolation over the 7450. Move to HIP7 and add more MHZ. Apollo must have more improvements (or else we'd have them in our hands by now) so there is no telling what the top end is.
Well as much as I beleive the G5 will be in the next imac, i can see the G4 scaling high above 1GHz. If the 7460 is simply a SOI process at 180nm then you can expect processors ranging 1GHz to 1.133GHz with no problem. I would expect 1.266GHz to be tops.With a die shrink (and add up to 512KB cache) to 130nm expect the G4 to scale to between 1.2GHz and 1.6GHz. Wouldn't these be great for PowerBooks and iMacs?
No one is arguing the existence of the Apollo here. It's what is the next chip for the towers. The Apollo fits with the iMac and portables because it has a shorted life before it tops out, even if it tops at 1.3ghz+ that's not enough for Apple to remain competetive in the power user arena.
That's where the G5 rumors seem to make sence and with the exposures from Epson, Amiga, Afsoft and Vuescan along with the 50 rumors puts together a legitiment argument.
Don't be dense. You code to the spec, which will get you 99% of where you need to be, you wait for the seeds to get out in the community, then you test against the seed and bash out the last few bugs.</strong><hr></blockquote>
This is completely bogus -- if you have any sizeable piece of code you certainly don't try to write 99% of before running it the first time! Only an idiot would try that, and he wouldn't be in the business very long.
As for why Absoft might want a G5-specific engineer...
1) Longer pipelines means a new scheduler, which can be very tricky to write.
2) A 64-bit processor requires a compiler that generates 64-bit code.
3) Any other special capabilities built into the processor that previous processors didn't have. New instructions are added occasionally, not to mention the AltiVec unit.
4) Absoft is primarily Fortran, IIRC. Any changes to the processor's math capabilities will want to be leveraged as much as possible since Fortran's primary market is in heavy math applications.
If Apollo is just the 7450 with SOI process then we can extrapolate about 20% to 30% improvement with SOI process. The 7450 can clock at 933MHZ (people have o'clocked 867 to 1066, but have lost the L3 cache) so we can expect Apollo in 1.1GHZ to 1.3GHZ range if it is just a straight extrapolation over the 7450. Move to HIP7 and add more MHZ. Apollo must have more improvements (or else we'd have them in our hands by now) so there is no telling what the top end is.</strong><hr></blockquote>
[quote] Position available immediately. <hr></blockquote>
Note, that it doesn't say, position available next year, when Apple finally yanks Moto's head out of their own ass so they can fab a few G5s for the MWSF 03 bake-off.
Now, about the Apollo...it's a G4, which is actually plays a very significant role in speculation about the G5. We know that the G4 has a design flaw that prevents it from being scaled easily. The G4 requires a disproportionate investment of time and resources to scale it's clockspeed. Apollo is yet another iteration of the G4 design, but like all of the G4 designs, Apollo is taking a very, very long time to finish. Remember when we had to wait 18 months for Moto get finish the 7450, just for a 233 MHz speedbump? Well, expect the same sort of delay and unwhelmingness from the Apollo. Specs change, but the architechture is still G4, and it's still a pain in the arse to crank up it's clockspeed.
Enter the G5, it scales like nobody's business, it's incredibly efficient, and it's been under development ever since the G4 debacle. The AIM alliance, or what was left of it, decided that since the G4 was such an awful pain in the ass to scale, that it would be easier and quicker to just design a whole new chip for high clockspeeds and performance.
Thus, it should not surprise anyone that the G5 would be finished before the Apollo G4, since the G4 design is inherently difficult to scale. However, designing a new CPU doesn't happen overnight, or in a few months, or even in a year, so Motorola had to do something a bit uncommon; they committed to developing both the G4 and the G5 in parallel. This also satisfies Apple's desire for both consumer and pro CPUs. The great thing about having both the G4 and G5 is that Apple's secret weapon, Altivec, can be made available to consumers. Developers can optimize for Altivec with the knowledge that millions of iMac owners are going to be looking for that little "altivec accelerated" sticker on the software box.
And what a great CPU lineup Apple is going to have! By this time next year; the monstrous G5 for Powermacs, and the G4 for virtually all other Macs. Maybe Apple can use surplus G3 chips in the iPod or something! Seriously, I see the G3 being used for some digital devices that fit in with Apple's digital hub strategy. But I'm getting ahead of myself. The Apollo has a place in Apple's lineup right alongside the G5. Since the Apollo will not be ready until this summer at the very earliest, we might expect that imacs will get a nice speedbump at MWNY. Alternatively, the Apollo may show up in the Titanium first, so that it can keep up with the G5 powermacs, and the iMacs will get a speedbump a bit later. After all, MWNY is a bit soon for another iMac revision, with G4 models not shipping in full until March.
So that's my long winded explanation of why the Apollo has little bearing on G5 availability and implementation.
Rock on, Apple! BTW, did you all hear? Apple rocks! Like a legendary rock star, Apple teeters on the brink of doom and failure, while playing in the realm of mad genius. Just when an ugly downfall seems inevitable, Apple triumphs over doom and arouses passion, hate, love, and lust, all while maintaining profitability in the midsts of a recession, and most importantly, without ditching more than a handful of employees. Apple is such an unamerican corporation, it's impossible not to respect a company that values the well being of its employees so highly as does Apple.
In other words, there is no difference between coding for G4 or G5.
Ergo, there is no difference between a G4 coder and a G5 coder.
Ergo, the alleged 'need' for G5 coders because of the imminent release of the G5 is nonsensical.
Ergo, we have to state that there continues to be no good evidence of an imminent release of the G5, whereas the evidence behind the Apollo continues to mount.
Ergo, despite all of the Tinkerbell Effect (if you wish hard enough....) in the world, the G5 will continue to be seriously in-the-future tech.
You know what?
I'd like to be wrong about the G5. I really would.
But I'm not.
Say hello to Apollo, kids! 'Cuz here it comes!
SdC</strong><hr></blockquote>
Sorry, but this is complete nonsense. There are always differences between processors. In a previous life I did some Mac programming, and the differences between the 601 vs. 603 vs. 604 were sometimes pretty substantial. For really high performance in graphics you had to write separate routines for each one - byte alignment problems could kill you. If you wrote it for just one, the other two would spend an enormous amount of time getting data moved, cutting performance as much as 1/2. And, as Programmer pointed out, this is not a generic "programmer" position - this is a compiler engineer, who must write everything down at machine level. I suspect there are a few interesting complexities in writing a 64-bit compiler vs. 32-bit. There's no way you could produce anything useful without a real chip to test it on. The thought of writing a compiler absolutely terrifies me, actually - I can't even imagine writing one "blind".
IMHO, there is clear evidence of G5 chips in prototype/testing situations. They have probably been out there for several months. My "gut feeling" odds at this point: 60% chance of a G5 before MWNY. An additional 30% chance before end of '02. 10% chance of after Jan. 1, 03.
<strong>A compiler writer really needs access to the chip the compiler is producing code for</strong><hr></blockquote>
that´s another indication of the G5 not coming anytime soon. If they are looking for an Engineer _now_ the chip is about a year away given the time it takes to find the right man and to code the compiler.
that´s another indication of the G5 not coming anytime soon. If they are looking for an Engineer _now_ the chip is about a year away given the time it takes to find the right man and to code the compiler.
bye.</strong><hr></blockquote>
What does Absoft have to do with the release of the G5? Just because they are looking for someone doesn't mean Apple has to wait for them to get an engineer. This just means it might take them a while to get their product up to G5 specs whatever they make.
So you're saying that if Steve says keep your mouth shut to Moto, they won't listen and they will release the G5 info conxcerning Apple? That's what I meant, that Apple can keep Moto quiet CIA style
</strong><hr></blockquote>
And the reason the same didn't happen with the 7400 and 7450 is, um, which?
[quote]<strong>
As for the Apollo being for portables, it's only the same sentiment echoed by various people on various sites, so I can only form my opinion based on the info I have read.</strong><hr></blockquote>
Yeah, but that's mostly based on the assumption that the G5 will be too hot to put into a notebook (which I don't doubt), so something cooler will have to be put there. Whether or not Apollo has been designed as an embedded processor is an entirely different matter (heck, on Wintel, they stick 1GHz+ desktop processors in their notebooks, and nobody is going to convince me those are really embedded chips ).
There will be a whole lot of 64-bit specific stuff to incorporate that is not part of the original spec either.
</strong><hr></blockquote>
What are you referring to here?
The PPC ISA does include 64 bit specific stuff already (and IIRC has been doing so since the beginning), but maybe you are referring to something else?
<strong>I really like this part of the job description:
(available immediately)
Note, that it doesn't say, position available next year, when Apple finally yanks Moto's head out of their own ass so they can fab a few G5s for the MWSF 03 bake-off.
</strong><hr></blockquote>
You did realize the "for Intel's IA64 (Itanium)" part, right?
Seen any shipping or near-shipping IA64 systems lately?
Comments
Ok, so the Apollo will give us 933, 1.0ghz and 1.13ghz before it tops out. That's one speed bumped show. That's topped out to me.
[quote] Well, unlike the iMac, Apple don't even produce the G5 (Motorola do). And IIRC Moto gave technology presentations on both the original 7400 and the 7450 before Apple actually built systems based on them. <hr></blockquote>
So you're saying that if Steve says keep your mouth shut to Moto, they won't listen and they will release the G5 info conxcerning Apple? That's what I meant, that Apple can keep Moto quiet CIA style
As for the Apollo being for portables, it's only the same sentiment echoed by various people on various sites, so I can only form my opinion based on the info I have read.
"Man then proceeded to prove that black was white, and got run over at the next crosswalk."
No one is disputing the fricking Apollo. The dispute is that the Apollo is intended for the iMac and the portables. No one has said the Apollo is for the towers and the G5 has been in production almost as long as the G4 (according to what i read).
Anyways, you believe what you want to believe and the rest of us will make up our own minds
just the guy from mot in mac central... but that was unclear ... he didn't deny the g5... and i was surely with a big nda above his head...
i read more about g5 in two month than apollo in two years...
50 rumors about the G5 and 1 statement from a MOTO rep saying that Apollo's were in production at .18mu (HIP6). Sometimes quality is better than quantity.
[quote]Ok, so the Apollo will give us 933, 1.0ghz and 1.13ghz before it tops out. That's one speed bumped show. That's topped out to me.<hr></blockquote>
If Apollo is just the 7450 with SOI process then we can extrapolate about 20% to 30% improvement with SOI process. The 7450 can clock at 933MHZ (people have o'clocked 867 to 1066, but have lost the L3 cache) so we can expect Apollo in 1.1GHZ to 1.3GHZ range if it is just a straight extrapolation over the 7450. Move to HIP7 and add more MHZ. Apollo must have more improvements (or else we'd have them in our hands by now) so there is no telling what the top end is.
That's where the G5 rumors seem to make sence and with the exposures from Epson, Amiga, Afsoft and Vuescan along with the 50 rumors puts together a legitiment argument.
<strong>
Don't be dense. You code to the spec, which will get you 99% of where you need to be, you wait for the seeds to get out in the community, then you test against the seed and bash out the last few bugs.</strong><hr></blockquote>
This is completely bogus -- if you have any sizeable piece of code you certainly don't try to write 99% of before running it the first time! Only an idiot would try that, and he wouldn't be in the business very long.
As for why Absoft might want a G5-specific engineer...
1) Longer pipelines means a new scheduler, which can be very tricky to write.
2) A 64-bit processor requires a compiler that generates 64-bit code.
3) Any other special capabilities built into the processor that previous processors didn't have. New instructions are added occasionally, not to mention the AltiVec unit.
4) Absoft is primarily Fortran, IIRC. Any changes to the processor's math capabilities will want to be leveraged as much as possible since Fortran's primary market is in heavy math applications.
<strong>
If Apollo is just the 7450 with SOI process then we can extrapolate about 20% to 30% improvement with SOI process. The 7450 can clock at 933MHZ (people have o'clocked 867 to 1066, but have lost the L3 cache) so we can expect Apollo in 1.1GHZ to 1.3GHZ range if it is just a straight extrapolation over the 7450. Move to HIP7 and add more MHZ. Apollo must have more improvements (or else we'd have them in our hands by now) so there is no telling what the top end is.</strong><hr></blockquote>
Should make for some fast TiBooks and iMacs.
[quote] Position available immediately. <hr></blockquote>
Note, that it doesn't say, position available next year, when Apple finally yanks Moto's head out of their own ass so they can fab a few G5s for the MWSF 03 bake-off.
Now, about the Apollo...it's a G4, which is actually plays a very significant role in speculation about the G5. We know that the G4 has a design flaw that prevents it from being scaled easily. The G4 requires a disproportionate investment of time and resources to scale it's clockspeed. Apollo is yet another iteration of the G4 design, but like all of the G4 designs, Apollo is taking a very, very long time to finish. Remember when we had to wait 18 months for Moto get finish the 7450, just for a 233 MHz speedbump? Well, expect the same sort of delay and unwhelmingness from the Apollo. Specs change, but the architechture is still G4, and it's still a pain in the arse to crank up it's clockspeed.
Enter the G5, it scales like nobody's business, it's incredibly efficient, and it's been under development ever since the G4 debacle. The AIM alliance, or what was left of it, decided that since the G4 was such an awful pain in the ass to scale, that it would be easier and quicker to just design a whole new chip for high clockspeeds and performance.
Thus, it should not surprise anyone that the G5 would be finished before the Apollo G4, since the G4 design is inherently difficult to scale. However, designing a new CPU doesn't happen overnight, or in a few months, or even in a year, so Motorola had to do something a bit uncommon; they committed to developing both the G4 and the G5 in parallel. This also satisfies Apple's desire for both consumer and pro CPUs. The great thing about having both the G4 and G5 is that Apple's secret weapon, Altivec, can be made available to consumers. Developers can optimize for Altivec with the knowledge that millions of iMac owners are going to be looking for that little "altivec accelerated" sticker on the software box.
And what a great CPU lineup Apple is going to have! By this time next year; the monstrous G5 for Powermacs, and the G4 for virtually all other Macs. Maybe Apple can use surplus G3 chips in the iPod or something!
So that's my long winded explanation of why the Apollo has little bearing on G5 availability and implementation.
Rock on, Apple! BTW, did you all hear? Apple rocks! Like a legendary rock star, Apple teeters on the brink of doom and failure, while playing in the realm of mad genius. Just when an ugly downfall seems inevitable, Apple triumphs over doom and arouses passion, hate, love, and lust, all while maintaining profitability in the midsts of a recession, and most importantly, without ditching more than a handful of employees. Apple is such an unamerican corporation, it's impossible not to respect a company that values the well being of its employees so highly as does Apple.
Apple rocks! They RAWK!
<strong>
This is from hmurchison in a different thread.
In other words, there is no difference between coding for G4 or G5.
Ergo, there is no difference between a G4 coder and a G5 coder.
Ergo, the alleged 'need' for G5 coders because of the imminent release of the G5 is nonsensical.
Ergo, we have to state that there continues to be no good evidence of an imminent release of the G5, whereas the evidence behind the Apollo continues to mount.
Ergo, despite all of the Tinkerbell Effect (if you wish hard enough....) in the world, the G5 will continue to be seriously in-the-future tech.
You know what?
I'd like to be wrong about the G5. I really would.
But I'm not.
Say hello to Apollo, kids! 'Cuz here it comes!
SdC</strong><hr></blockquote>
Sorry, but this is complete nonsense. There are always differences between processors. In a previous life I did some Mac programming, and the differences between the 601 vs. 603 vs. 604 were sometimes pretty substantial. For really high performance in graphics you had to write separate routines for each one - byte alignment problems could kill you. If you wrote it for just one, the other two would spend an enormous amount of time getting data moved, cutting performance as much as 1/2. And, as Programmer pointed out, this is not a generic "programmer" position - this is a compiler engineer, who must write everything down at machine level. I suspect there are a few interesting complexities in writing a 64-bit compiler vs. 32-bit. There's no way you could produce anything useful without a real chip to test it on. The thought of writing a compiler absolutely terrifies me, actually - I can't even imagine writing one "blind".
IMHO, there is clear evidence of G5 chips in prototype/testing situations. They have probably been out there for several months. My "gut feeling" odds at this point: 60% chance of a G5 before MWNY. An additional 30% chance before end of '02. 10% chance of after Jan. 1, 03.
<strong>A compiler writer really needs access to the chip the compiler is producing code for</strong><hr></blockquote>
that´s another indication of the G5 not coming anytime soon. If they are looking for an Engineer _now_ the chip is about a year away given the time it takes to find the right man and to code the compiler.
bye.
<strong>
that´s another indication of the G5 not coming anytime soon. If they are looking for an Engineer _now_ the chip is about a year away given the time it takes to find the right man and to code the compiler.
bye.</strong><hr></blockquote>
What does Absoft have to do with the release of the G5? Just because they are looking for someone doesn't mean Apple has to wait for them to get an engineer. This just means it might take them a while to get their product up to G5 specs whatever they make.
<strong>
Apple rocks! They RAWK!</strong><hr></blockquote>
Yeah...only if they can deliver
you all do know that ibm has a g4/g5/g6 line of computers/processors which is totally unrelated to anything from apple/motorola?
so, if our friend amir talks about his career and the g5, he certainly isn't talking about our g5.
[ 01-12-2002: Message edited by: Strangelove ]</p>
<strong>
So you're saying that if Steve says keep your mouth shut to Moto, they won't listen and they will release the G5 info conxcerning Apple? That's what I meant, that Apple can keep Moto quiet CIA style
</strong><hr></blockquote>
And the reason the same didn't happen with the 7400 and 7450 is, um, which?
[quote]<strong>
As for the Apollo being for portables, it's only the same sentiment echoed by various people on various sites, so I can only form my opinion based on the info I have read.</strong><hr></blockquote>
Yeah, but that's mostly based on the assumption that the G5 will be too hot to put into a notebook (which I don't doubt), so something cooler will have to be put there. Whether or not Apollo has been designed as an embedded processor is an entirely different matter (heck, on Wintel, they stick 1GHz+ desktop processors in their notebooks, and nobody is going to convince me those are really embedded chips
Bye,
RazzFazz
<strong>
There will be a whole lot of 64-bit specific stuff to incorporate that is not part of the original spec either.
</strong><hr></blockquote>
What are you referring to here?
The PPC ISA does include 64 bit specific stuff already (and IIRC has been doing so since the beginning), but maybe you are referring to something else?
Bye,
RazzFazz
<strong>I really like this part of the job description:
(available immediately)
Note, that it doesn't say, position available next year, when Apple finally yanks Moto's head out of their own ass so they can fab a few G5s for the MWSF 03 bake-off.
</strong><hr></blockquote>
You did realize the "for Intel's IA64 (Itanium)" part, right?
Seen any shipping or near-shipping IA64 systems lately?
Bye,
RazzFazz
<strong>What does Absoft have to do with the release of the G5?</strong><hr></blockquote>
see, absolutely nothing.
So why the f*ck is everybody taking it as an evidence of the G5 being released soon?
It means absolutely nothing in terms of the release date of the G5.
The same people collecting the "evidence" will bitch the loudest when the G5 is not making it´s way to the next PowerMac update.
bye.