So when an LCD with a fixed pixel ratio to screen size suddenly becomes dynamic.
No, no, no. This works with any screen. Remember, QuickDraw supported resolution independence in 1983!
All it means is that the hardware pixel is no longer a unit of measurement. A font glyph is 12 points tall, and the denser your screen is, the more accurately that glyph will be rendered.
So, basically, all I'm saying is that with a resolution independent OS X, you could put your laptop side by side with a laptop that had a 600dpi screen with the same physical dimensions, and with both running at full native resolution all the text and images and widgets would be the same size, but the 600dpi screen would be as crisp as a page from a color laser printer. The way it is now, with things measured in hardware pixels, everything on screen would be so tiny that you couldn't read it.
It's exactly analogous to printing a document on a 144dpi ImageWriter and then printing the same document on a 600dpi laser printer, because printers have been resolution independent for decades now. Everything will be roughly the same size in both documents, but the laser printer will produce a much crisper, more accurate rendering.
As an added bonus, you can scale in software, and blow everything up to double the size (or shrink it to half the size) and still have the screen render everything at its full native resolution. Because once you've made rendering independent of the hardware, you have a great deal of freedom.
Some interesting news from Think Secret on 970GX processors and little-known news of a low-power 970. PowerBook G5s are coming. No denying it.
I am reading this a little differently: despite the brave efforts IBM and Apple put into production of the 970GX and design of the Powerbook G5, it is still probable that this Powerbook will not come to daylight anytime soon, forcing in this case Apple to use the 7448 chip from Freescale, which, according to sources, is almost finished. And we should be thankful if this is the case, since otherwise there would be no new processor for a Powerbook update.
According to an old article at The Register, the 7448 will sport not only 1 Mb of L2-cache but also a 200 MHz MPX-frontside bus (up from 167 MHz). The TS-article seems to corraborate the part about the double L2-cache, but it mentions nothing of the speedier memory controller (MPX-bus) and I'm not sure if that has been confirmed elsewhere (public Freescale documents for instance).
Still, this seems like a really nice mobile chip and I'd love to see it in the next revision of the Powerbook-line.
NAll it means is that the hardware pixel is no longer a unit of measurement. A font glyph is 12 points tall, and the denser your screen is, the more accurately that glyph will be rendered.
Ah, now I am on the same page. Is there any information or news regarding this independence with regards to Mac OS X?
Remember, QuickDraw supported resolution independence in 1983!
Well, not really... it was all integer based and had no base scaling matrix. Printer drivers had to scale up or applications had to scale to fit the page resolution, and there was no screen resolution independence to speak of. QuickDrawGX was Apple's first attempt at that, but it didn't catch on and they'd made the mistake of not using floating point (started a little too early -- before the advent of ubiquitous FPUs).
As for the processor in the next PowerBook, TS doesn't really say either way. If the 970GX (and the machine around it) is ready we'll see that next, otherwise it'll be the 90nm G4. Given that this is Freescale's first 90nm processor and everyone who has gone 90nm so far has had troubles... I'd give a PBG5 in the next revision about a 50/50 chance, but whether it shows up in January or April is a crap shoot.
Well, not really... it was all integer based and had no base scaling matrix. Printer drivers had to scale up or applications had to scale to fit the page resolution, and there was no screen resolution independence to speak of. QuickDrawGX was Apple's first attempt at that, but it didn't catch on and they'd made the mistake of not using floating point (started a little too early -- before the advent of ubiquitous FPUs).
No released version of QuickDraw supported res independence, which I hinted at when I said 1983 rather than 1984. The twin optimizations of assuming a 72ppi screen and CopyBits() killed any attempt at res independence, but on the other hand, they were sorely needed.
The point is that the original QD API was originally designed not to make any assumptions about resolution. They were trying to look ahead to widespread FP support, so that the platform could easily be migrated. But that didn't happen.
Quote:
As for the processor in the next PowerBook, TS doesn't really say either way. If the 970GX (and the machine around it) is ready we'll see that next, otherwise it'll be the 90nm G4. Given that this is Freescale's first 90nm processor and everyone who has gone 90nm so far has had troubles... I'd give a PBG5 in the next revision about a 50/50 chance, but whether it shows up in January or April is a crap shoot.
It can always show up in January and ship in April. Wouldn't be the first time.
There are indeed a lot of ifs here, but it's worth remembering that Freescale has been able to play with 90nm tech for almost as long as everyone else; they've just had a more cautious timeline for attempting to fab a CPU. So I think it's possible that the 7448 rollout will be relatively smooth, just because they've taken their time.
IBM, meanwhile, are back to trying to get a new process tech going and immediately fabbing a CPU on the new tech...
Barring some unannounced 7447 revision (B?), this makes a PowerBook G5 at MWSF in January more and more like a shoo-in.
We can only hope. I'll probably end up getting a powerbook when the new models come out regardless of what processor it is, unless it's a tiny stop-gap speed bump with a high probability of a major step being announced *and* shipped to my doorstep by the middle of the summer.
No released version of QuickDraw supported res independence, which I hinted at when I said 1983 rather than 1984. The twin optimizations of assuming a 72ppi screen and CopyBits() killed any attempt at res independence, but on the other hand, they were sorely needed.
You been chatting with Atkinson again?
Quote:
The point is that the original QD API was originally designed not to make any assumptions about resolution. They were trying to look ahead to widespread FP support, so that the platform could easily be migrated. But that didn't happen.
Not making assumptions is quite a bit different than designing for independence. And there was no sign of an attempt at planning for FP migration -- even in QDGX they didn't do that.
Quote:
There are indeed a lot of ifs here, but it's worth remembering that Freescale has been able to play with 90nm tech for almost as long as everyone else; they've just had a more cautious timeline for attempting to fab a CPU. So I think it's possible that the 7448 rollout will be relatively smooth, just because they've taken their time.
Then again, IBM and Intel didn't run into troubles until they actually attempted production. Everything up to that point went smoothly. And given the trouble Freescale had getting to even a full 130nm process, I'm taking a wait-and-see attitude.
Quote:
IBM, meanwhile, are back to trying to get a new process tech going and immediately fabbing a CPU on the new tech...
Are you talking about 65nm or the new materials on 90nm? They've been experimenting with new materials on 90nm for over a year now and its still pretty much a 970 (or two) with minor tweaks, so I wouldn't dismiss that as "new process tech... immediately fabbing a new CPU". If I were a betting man my money would be on IBM at this point... Freescale has a longer history of screwups, and they've been leaking braintrust for years now.
Looks like a good probability that new G5 notebooks are in order once again. With all the news breaking, it looks more and more to be the case (but I really hope they don't change the most fetching notebook enclosure design in the industry to much).
What's all this talk about the G5 beeing hot. Wasn't the problem with the G5 heat density not heat. The desktop G4's were really hot but Apple managed to put them in their laptops. We also don't have a clue what adjustments they made to make them into a laptop chip (well we acually don't know if they made it into a laptop chip). Also who tought that the iPod photo would have longer battery life than the iPod when it has a coulor screen an stuff.
I sure there will be a G5 PowerBook, the ony question is when.
I've done more reading than is healthy, and that cropped up multiple times.
Quote:
Not making assumptions is quite a bit different than designing for independence. And there was no sign of an attempt at planning for FP migration -- even in QDGX they didn't do that.
Well, this is getting into a level of detail far beyond what I read, which is that the QuickDraw API was originally designed not to make any assumptions about resolution, and that design had to be broken to make a needed optimization—the blitter. Maybe they'd planned to go straight from 72dpi to 144dpi on screen, and avoid the FP issue altogether? Or maybe they were just doing it because it was prettier? Who knows? Given that QD was compromised out of the starting gate, we'll never know.
Quote:
Then again, IBM and Intel didn't run into troubles until they actually attempted production. Everything up to that point went smoothly.
On the other hand, very little had happened "up to that point." They rushed a new CPU design onto a new process, which is exactly what Motorola did with the first G4. And we all know what happened there. I'm willing to give Freescale and AMD credit for letting the dust settle first, because that does seem to make a difference.
Quote:
And given the trouble Freescale had getting to even a full 130nm process, I'm taking a wait-and-see attitude.
So, apparently, is Freescale.
Quote:
Are you talking about 65nm or the new materials on 90nm? They've been experimenting with new materials on 90nm for over a year now and its still pretty much a 970 (or two) with minor tweaks, so I wouldn't dismiss that as "new process tech... immediately fabbing a new CPU". If I were a betting man my money would be on IBM at this point... Freescale has a longer history of screwups, and they've been leaking braintrust for years now.
I'm talking about new process tech—not "experimenting with new materials," but actually trying to put all the fixings onto a production 90nm process with the goal of producing the 0.8v part that has eluded them sofar. This is the "laptop" 970 that emerged in early IBM documentation, and then vanished.
Stepping back and looking at the industry, I don't really see the point in betting anymore. Of course, there will be successes, and there will be progress, and products will ship, but it seems like a lot more time is being spent on things that used to be taken for granted, and any industry that requires paradigm-shifting breakthroughs to continue forward is going to be unpredictable, at least until it settles on a new set of paradigms suited to the new reality sub-110nm. And frankly, while IBM is doing much better than Motorola was doing four years ago, I still don't like everything I'm seeing over in upstate NYC. And I'm old enough to be deeply, almost reflexively skeptical of IBM.
I met IBM processor people in October or so. It was internal exhibition and I met there IBM processoer people. Exhibition was only for our company, where I work. IBM people couldn't say much, but they said, that Apple might use new IBM low power processor next year.
Comments
Originally posted by Rhumgod
So when an LCD with a fixed pixel ratio to screen size suddenly becomes dynamic.
No, no, no. This works with any screen. Remember, QuickDraw supported resolution independence in 1983!
All it means is that the hardware pixel is no longer a unit of measurement. A font glyph is 12 points tall, and the denser your screen is, the more accurately that glyph will be rendered.
So, basically, all I'm saying is that with a resolution independent OS X, you could put your laptop side by side with a laptop that had a 600dpi screen with the same physical dimensions, and with both running at full native resolution all the text and images and widgets would be the same size, but the 600dpi screen would be as crisp as a page from a color laser printer. The way it is now, with things measured in hardware pixels, everything on screen would be so tiny that you couldn't read it.
It's exactly analogous to printing a document on a 144dpi ImageWriter and then printing the same document on a 600dpi laser printer, because printers have been resolution independent for decades now. Everything will be roughly the same size in both documents, but the laser printer will produce a much crisper, more accurate rendering.
As an added bonus, you can scale in software, and blow everything up to double the size (or shrink it to half the size) and still have the screen render everything at its full native resolution. Because once you've made rendering independent of the hardware, you have a great deal of freedom.
Originally posted by Rhumgod
Some interesting news from Think Secret on 970GX processors and little-known news of a low-power 970. PowerBook G5s are coming. No denying it.
I am reading this a little differently: despite the brave efforts IBM and Apple put into production of the 970GX and design of the Powerbook G5, it is still probable that this Powerbook will not come to daylight anytime soon, forcing in this case Apple to use the 7448 chip from Freescale, which, according to sources, is almost finished. And we should be thankful if this is the case, since otherwise there would be no new processor for a Powerbook update.
Still, this seems like a really nice mobile chip and I'd love to see it in the next revision of the Powerbook-line.
Originally posted by Amorph
NAll it means is that the hardware pixel is no longer a unit of measurement. A font glyph is 12 points tall, and the denser your screen is, the more accurately that glyph will be rendered.
Ah, now I am on the same page. Is there any information or news regarding this independence with regards to Mac OS X?
Originally posted by Amorph
Remember, QuickDraw supported resolution independence in 1983!
Well, not really... it was all integer based and had no base scaling matrix. Printer drivers had to scale up or applications had to scale to fit the page resolution, and there was no screen resolution independence to speak of. QuickDrawGX was Apple's first attempt at that, but it didn't catch on and they'd made the mistake of not using floating point (started a little too early -- before the advent of ubiquitous FPUs).
As for the processor in the next PowerBook, TS doesn't really say either way. If the 970GX (and the machine around it) is ready we'll see that next, otherwise it'll be the 90nm G4. Given that this is Freescale's first 90nm processor and everyone who has gone 90nm so far has had troubles... I'd give a PBG5 in the next revision about a 50/50 chance, but whether it shows up in January or April is a crap shoot.
Originally posted by Programmer
Well, not really... it was all integer based and had no base scaling matrix. Printer drivers had to scale up or applications had to scale to fit the page resolution, and there was no screen resolution independence to speak of. QuickDrawGX was Apple's first attempt at that, but it didn't catch on and they'd made the mistake of not using floating point (started a little too early -- before the advent of ubiquitous FPUs).
No released version of QuickDraw supported res independence, which I hinted at when I said 1983 rather than 1984. The twin optimizations of assuming a 72ppi screen and CopyBits() killed any attempt at res independence, but on the other hand, they were sorely needed.
The point is that the original QD API was originally designed not to make any assumptions about resolution. They were trying to look ahead to widespread FP support, so that the platform could easily be migrated. But that didn't happen.
As for the processor in the next PowerBook, TS doesn't really say either way. If the 970GX (and the machine around it) is ready we'll see that next, otherwise it'll be the 90nm G4. Given that this is Freescale's first 90nm processor and everyone who has gone 90nm so far has had troubles... I'd give a PBG5 in the next revision about a 50/50 chance, but whether it shows up in January or April is a crap shoot.
It can always show up in January and ship in April. Wouldn't be the first time.
There are indeed a lot of ifs here, but it's worth remembering that Freescale has been able to play with 90nm tech for almost as long as everyone else; they've just had a more cautious timeline for attempting to fab a CPU. So I think it's possible that the 7448 rollout will be relatively smooth, just because they've taken their time.
IBM, meanwhile, are back to trying to get a new process tech going and immediately fabbing a CPU on the new tech...
(H2-05 for the 7448, H1-06 for the 8641 and the 8641D)
Originally posted by Eric_Z
I don't remember seeing production dates in it.
(H2-05 for the 7448, H1-06 for the 8641 and the 8641D)
In the introduction of the new chips one(?) month ago, Freescale had said that they are going to sample the 7448 in H1-05 and the 8641D in 2H-05.
Originally posted by DHagan4755
Barring some unannounced 7447 revision (B?), this makes a PowerBook G5 at MWSF in January more and more like a shoo-in.
We can only hope. I'll probably end up getting a powerbook when the new models come out regardless of what processor it is, unless it's a tiny stop-gap speed bump with a high probability of a major step being announced *and* shipped to my doorstep by the middle of the summer.
Originally posted by Amorph
No released version of QuickDraw supported res independence, which I hinted at when I said 1983 rather than 1984. The twin optimizations of assuming a 72ppi screen and CopyBits() killed any attempt at res independence, but on the other hand, they were sorely needed.
You been chatting with Atkinson again?
The point is that the original QD API was originally designed not to make any assumptions about resolution. They were trying to look ahead to widespread FP support, so that the platform could easily be migrated. But that didn't happen.
Not making assumptions is quite a bit different than designing for independence. And there was no sign of an attempt at planning for FP migration -- even in QDGX they didn't do that.
There are indeed a lot of ifs here, but it's worth remembering that Freescale has been able to play with 90nm tech for almost as long as everyone else; they've just had a more cautious timeline for attempting to fab a CPU. So I think it's possible that the 7448 rollout will be relatively smooth, just because they've taken their time.
Then again, IBM and Intel didn't run into troubles until they actually attempted production. Everything up to that point went smoothly. And given the trouble Freescale had getting to even a full 130nm process, I'm taking a wait-and-see attitude.
IBM, meanwhile, are back to trying to get a new process tech going and immediately fabbing a CPU on the new tech...
Are you talking about 65nm or the new materials on 90nm? They've been experimenting with new materials on 90nm for over a year now and its still pretty much a 970 (or two) with minor tweaks, so I wouldn't dismiss that as "new process tech... immediately fabbing a new CPU". If I were a betting man my money would be on IBM at this point... Freescale has a longer history of screwups, and they've been leaking braintrust for years now.
Originally posted by 9secondko
Wait for 10.3.7-8.
Why have it in Panther now, when at least one more update is coming before MWSF?
No one is going to downgrade the OS that comes with the new PB.
Hmm, I thought I remembered me saying that.
http://www.appleinsider.com/article.php?id=747
Looks like a good probability that new G5 notebooks are in order once again. With all the news breaking, it looks more and more to be the case (but I really hope they don't change the most fetching notebook enclosure design in the industry to much).
10.3.7 is comin' down the pike.
the powerbooks are already hot on your lap. with a g5 it would be unberable.
IBM would have to come out with a 65nm chip. given it's current problems that dosn't look like it could do that soon
I sure there will be a G5 PowerBook, the ony question is when.
Viktor
Originally posted by Programmer
You been chatting with Atkinson again?
Yeah, he's a fixture here in Iowa.
I've done more reading than is healthy, and that cropped up multiple times.
Not making assumptions is quite a bit different than designing for independence. And there was no sign of an attempt at planning for FP migration -- even in QDGX they didn't do that.
Well, this is getting into a level of detail far beyond what I read, which is that the QuickDraw API was originally designed not to make any assumptions about resolution, and that design had to be broken to make a needed optimization—the blitter. Maybe they'd planned to go straight from 72dpi to 144dpi on screen, and avoid the FP issue altogether? Or maybe they were just doing it because it was prettier? Who knows? Given that QD was compromised out of the starting gate, we'll never know.
Then again, IBM and Intel didn't run into troubles until they actually attempted production. Everything up to that point went smoothly.
On the other hand, very little had happened "up to that point." They rushed a new CPU design onto a new process, which is exactly what Motorola did with the first G4. And we all know what happened there. I'm willing to give Freescale and AMD credit for letting the dust settle first, because that does seem to make a difference.
And given the trouble Freescale had getting to even a full 130nm process, I'm taking a wait-and-see attitude.
So, apparently, is Freescale.
Are you talking about 65nm or the new materials on 90nm? They've been experimenting with new materials on 90nm for over a year now and its still pretty much a 970 (or two) with minor tweaks, so I wouldn't dismiss that as "new process tech... immediately fabbing a new CPU". If I were a betting man my money would be on IBM at this point... Freescale has a longer history of screwups, and they've been leaking braintrust for years now.
I'm talking about new process tech—not "experimenting with new materials," but actually trying to put all the fixings onto a production 90nm process with the goal of producing the 0.8v part that has eluded them sofar. This is the "laptop" 970 that emerged in early IBM documentation, and then vanished.
Stepping back and looking at the industry, I don't really see the point in betting anymore. Of course, there will be successes, and there will be progress, and products will ship, but it seems like a lot more time is being spent on things that used to be taken for granted, and any industry that requires paradigm-shifting breakthroughs to continue forward is going to be unpredictable, at least until it settles on a new set of paradigms suited to the new reality sub-110nm. And frankly, while IBM is doing much better than Motorola was doing four years ago, I still don't like everything I'm seeing over in upstate NYC. And I'm old enough to be deeply, almost reflexively skeptical of IBM.
When Apple execs have commented on the problems posed by heat to the PowerBook range, I think they may have been actually referring to the 12"
Apple has to upgrade the whole range if there's a new processor.
They can't have two G5 models and 1 G4.
With news of the quiet development of a low power G5, I think it's certain that MWSF will see the debut of the PBG5 range.
<Where's that predictions thread?...
If the market wants it, and sales of G4s are hurting, it makes no sense to wait until all metal-clad books from Apple are G5'd.
They could simply designate a (one-model) PBG4 line and a PBG5 line.