DigiTimes Reports PowerBook G5 and iBook G5 Contracts

[HomesliceJ]

64-BIT ch!

So, the "rah rah" cheerleading for a 64-bit laptop is really misguided - most software will remain 32-bits even on the 64-bit system. This is Apple's recommendation, not just Aiden's opinion.[/url])

That's an interesting fact! Kinda stupid for Apple to go "rah" hitting a 64-bit computer saying "the First 64-bit computer" when it has no benefit whatsoever. But they are foreseeing the future. The reasoning i think is that Apple is setting the platform for 64-bit right now, taking it a step at a time. Giving the resource now, so that later people may really take advantage of a 64-bit chip. That seems to make sense.

Remember when the 386's came out? The First 32-bit processor! That was 1985. Remember? Ok i was only 4, but i'm a nerd. Anyway, 386's later on had companion 387 math co-processors. And it was still DOS! Then the newly found 16-bit Windows up to 3.11 Workgroups. 486 came out later and Windows started to gain fame to become the big thang. Windows slowly made it's way to 32-bit using a "win32" extention allowing 32-bit programs. Yea remember Windows NT? The only 32-bit Windows at the time? We didn't really see a complete movement of 32-bit until Windows 95 in 1995 (10 years later) which was the bloom of 32-bit applications.

The push of the 32-bit from Apple seems like a healthy move so that Apple can be sure 64-bit can be fully utilized in the future. Seeing that on x86 systems, it took 10 years before 32-bit made it's way. Also, there has been so many revisions of the chip (math co-processors, combined core, Pentium, MMX, etc.); which gave a nice foundation for powerful 32-bit programs.

So yea 64-bit processing... I guess we won't see software for it for a while.

 

[zot72]

Apple will stick with a 32-bit OS until L1 cache gets big enough that it no longer matters that the pointers double in size for a 64-bit computer. 64 bit applications are always inferior to otherwise-identical 32-bit apps unless the app needs to address more than two gigabytes of RAM or many values larger than 32 bits. Few applications meet these criteria.

You're ignoring Apple's core market of users that know when dates can no longer be expressed in seconds-since-the-epoch and have large numbers of events after that date that they're constantly performing complex operations on...in iCal.

 

[Kagetenshi]

That's an interesting fact! Kinda stupid for Apple to go "rah" hitting a 64-bit computer saying "the First 64-bit computer" when it has no benefit whatsoever. The reasoning i think is that Apple is setting the platform for 64-bit right now, taking it a step at a time.

It does have benefit. Plenty of benefit. Look at Final Cut Pro. Having it be able to address four or more gigs of RAM is a godsend.

For the average user? Not so much.

~J

 

[soapsuds]

64-bit and virtual memory

One thing I haven't seen mentioned here is that 64-bit addressing can still be useful on a machine with "only" 4GB of physical RAM. For one thing, it allows applications to use memory-mapped file I/O to access large files. It also allows for more virtual memory--for instance, having an 8 GB swapfile would be reasonable on a 4GB machine. Lastly, when you are using a 32-bit address space to access 4GB of RAM, you can have serious problems due to fragmentation of the address space. For instance, in a hypothetical bad case, lets say you're running the new spiffy app "iCal HD" and it makes four 1GB allocations, frees the 2nd and 4th blocks, and then tries to make a single 2GB allocation--it will fail, because the address space has now been fragmented, even though there is now 2GB of unused memory! Most programmers have gotten spoiled by working on modern machines with paged memory architectures and haven't been worrying about fragmentation for a while...

Obviously, none of this matters for most users at this point, and that true 64-bit GUI apps will not be possible on OS X for quite some time--right now to make a 64-bit app you have to split your program into a 32-bit GUI process and a 64-bit "engine" process and have them communicate indirectly, which sounds onerous enough that only people who *really* need 64 bit addressing will make 64-bit applications--it's not a simple recompile. I don't know when "Cocoa64" or whatever they will call it will debut, but it doesn't appear to be in line for the Tiger release. I also don't know when G5 PowerBooks are coming out, but clearly nobody else on this forum does either...

 

[HomesliceJ]

OSX Tiger and 64-bit

And one more thing....

If Tiger is truly 64-bit, wasup with support for the Mac Mini's and eMacs? Is it possible to have an Apple retail store with everything running Tiger except Mac Mini's and eMacs? That is totally not Apple's style.

My guess, if Tiger really is 64 bit, there has to be an "alternative" Tiger installation that runs all the neat new Tiger Apps, which means, like i've said before, probably a "watered-down" version of CoreImage.

But yea the 64-bit in G5's isn't the key to it's chip, so i agree with all of you saying it's the architecture handling higher system bus and larger RAM.

 

[Kagetenshi]

Tiger will have 64-bit components. It may install everything and just use the proper components, or it may identify and install the proper components. However, the entire OS being 64-bit is undesirable for the same reason any other app not needing a lot of +32 bit numbers or large memory addressing being 64-bit is undesirable.

~J

 

[Tthx3]

Any significance?

Hi everybody, I'm new at the forum and still posting with a Windows machine. However, I'm looking closely the prospected PowerBook update discussion as I might get one as soon as the line gets at least a speed bump.

Considering the recent Think Secret article about the confirmed PB updates and remembering it's Tuesday a week after the MW, I wonder if the little green balls with the word 'New' pasted on top of the PB line pictures at the Australian and some but not all Asian Apple Store web pages have any significance. I don't think I've seen those before. Opinions?

 

[Pringolian]

Bummer!

No significance I'm afraid. They are highlighting the last revisions of the PowerBook from 1.33Ghz to 1.5Ghz G4, god knows how many months ago that was. Australia/Asia have just been a tad slow on removing them.

 

[3Memos]

PB updates are just inching along. Most likely 1.5-1.67 Ghz upgrade. Perhaps Airport Extreme2 to take advantage of 802.11 Pre-N. Price cuts for sure across the board.

Some things I hope to see in the updates. Better video card. Higher resolution displays than currently offered. Backlit KB standard. Superdrives standard.

 

[Pringolian]

PowerBook G4 Revision

PowerBook G4 1.67Ghz
100Gb Hard Drive (5400rpm) on high end models
8x Superdrive with official support for +R format
Bluetooth 2.0
No new displays, unfortunately (until G5?)

These are the predictions for the G4 update, if there is one.

 

[Eastend]

Need a PowerBook

PowerBook G4 1.67Ghz
100Gb Hard Drive (5400rpm) on high end models
8x Superdrive with official support for +R format
Bluetooth 2.0
No new displays, unfortunately (until G5?)

These are the predictions for the G4 update, if there is one.

Unfortunately I think you may be right. If they do not improve the video card and get the PowerBook screen more up to date, plus increase the performance by at least 14 per cent or so, then I'll still be waiting on the next revision. Let's hope Apple has a surprise for us all.

Brian

 

[Pringolian]

Please, Please, Please, Please Please!!!!!!!!!!!!!!

Unfortunately I think you may be right. If they do not improve the video card and get the PowerBook screen more up to date, plus increase the performance by at least 14 per cent or so, then I'll still be waiting on the next revision. Let's hope Apple has a surprise for us all.

Brian

The PowerBooks haven't been updated for 10 months... That is apparently the longest revision Apple has ever had on the PowerBooks.
I don't think the rumoured G4 upgrade warrants such a delay (they should have been updated along with the iBook revision before Xmas).

This taken into account; I still have my fingers crossed for a surprise from Apple!!!

 

[Eastend]

Surprise?

The PowerBooks haven't been updated for 10 months... That is apparently the longest revision Apple has ever had on the PowerBooks.
I don't think the rumoured G4 upgrade warrants such a delay (they should have been updated along with the iBook revision before Xmas).

This taken into account; I still have my fingers crossed for a surprise from Apple!!!

I do not think it will happen, but you never know what Apple may spring on you, I do hope dreams come true. Whatever, still waiting.

Brian

 

[Eastend]

announcement Possible?

Of course it is only about 5:00 AM in California now, there could possibly be some kind of announcement today about a PowerBook, but keep in mind this is just rumors floating around the Internet. Who Knows?

Brian

 

[3Memos]

Of course it is only about 5:00 AM in California now, there could possibly be some kind of announcement today about a PowerBook, but keep in mind this is just rumors floating around the Internet. Who Knows?

Brian

Kewl.. Just 3 hours away from a possible announcement. But so soon after MacWorld? Perhaps next tuesday.

 

[i_am_a_cow]

where on earth is my powerbook

 

[SiliconAddict]

where on earth is my powerbook

In the R&D labs of Apple.

 

[Timboid]

Intel's Dual Core space heater...

Well, i don't know if anybody has spotted this, but it's nice to see that after all the problems that IBM?Apple have had with hot CPU's, that Intel seem to have run into problems that are just as bad, if not worse than IBMs!

See http://www.tomshardware.com/hardnews/20050117_225242.html

Is this going to be an issue for the future IBM dual cores as and when they surface, I wonder? Anybody know any more?

 

[AidenShaw]

don't be quite so negative about a 64-bit O/S

Tiger will have 64-bit components. It may install everything and just use the proper components, or it may identify and install the proper components. However, the entire OS being 64-bit is undesirable for the same reason any other app not needing a lot of +32 bit numbers or large memory addressing being 64-bit is undesirable.

I feel that there's a bit of mis-information here....

First of all, the slowdown due to 64-bit pointers is real, but it's usually from a minor effect to un-measureable. It depends on the application, the percentage of the application's active RAM use that is pointers, and how the pointers are used.

It usually is only significant for programs that have a lot of data in many, many small structures that are linked together in lists or trees (especially if they have multiple links). Even then, the slowdown is seldom significant - a fraction of a percent to a few percent (it can be statistically significant, but not practically significant). Of course, there are some programs that spend a lot of time traversing pointers that can see tens of percent of a hit.

It does not follow that having the O/S fully 64-bit is a liability. Most of the use of pointers within the O/S falls into the insignificant category. (Look at the CPU usage of a typical CPU intensive application, and see how much is in the kernel vs in the user application. Adding a few tenths of a percent to the kernel CPU usage isn't going to be noticeable.)

If the 64-bit O/S supports 32-bit applications, then it will need to have a set of 32-bit libraries for those apps. That means that time spent in libraries for a 32-bit app is not increased by running on the 64-bit O/S - even by fractions of a percent.

The biggest advantage of a fully 64-bit O/S is that it can support GUI applications that are 64-bit capable. Windows XP 64-bit and Server 64-bit don't force you to break an application into GUI front-end and 64-bit back end in order to use extra memory.

Also, note that PPC970 and OS X support 64-bit arithmetic in a completely 32-bit system. It has always supported 64-bit native double precision floating point, way back to the 68K days. A compiler switch lets you use native 64-bit integers (the only new math in the PPC970) while using 32-bit addresses. You don't need a 64-bit O/S or a 64-bit application to use native 64-bit integers - you can do that on any PPC970 with the 32-bit O/S.

Finally, and just to make sure that nothing is ever simple, the Xeon 64-bit and AMD 64-bit systems are often quite a bit faster running in 64-bit mode. This isn't because the pointers are bigger, but because the 64-bit x86 instruction set has twice as many registers (and nearly 3 times as many "free' ones) and twice as many SSE vector registers. Any program compiled for 64-bit mode gets to use the extra registers for better program optimization. This will almost always more than compensate for the 64-bit addressing slowdown. (x64 doesn't have a "64-bit math with 32-bit addressing" mode like PPC970.)

 

[JonMaker]

For all you 64 bit naysayers

I'll admit that 64-bit won't make anything faster. It only allows for more RAM.

Remember when Bill said that 640k should "be enough for anyone"?
In another few years, what you're saying about only needing 4GB will seem equally foolish.

Bring on the 64-bitness!

 

[AidenShaw]

in essence, I think that you're trying to fit 64-bit apps into 32-bits

One thing I haven't seen mentioned here is that 64-bit addressing can still be useful on a machine with "only" 4GB of physical RAM. ... It also allows for more virtual memory--for instance, having an 8 GB swapfile would be reasonable on a 4GB machine.

Sorry, but I don't think that an 8 GiB swapfile would be reasonable under any circumstances where you actually need it.

Your CPU can process 8 GiB of RAM in about 4 seconds -- it would need at least 4 minutes to get that memory from the swapfile, and probably a lot longer on a laptop swap drive....

For one thing, it allows applications to use memory-mapped file I/O to access large files. ... Lastly, when you are using a 32-bit address space to access 4GB of RAM, you can have serious problems due to fragmentation of the address space.

These are valid points, but you're also really talking about applications that should be 64-bit anyway.

Trying to completely utilize your 4 GiB of virtual memory is a boundary case - it's trying to get much of the advantages of 64-bit programming inside a 32-bit container.

Your enormous memory-mapped file is useful if you have the RAM to hold it - if you don't you'll often be better off treating it as a real file. The file system can usually do better buffering and read-ahead/write-behind of file data than the memory system.
______________

One other example of where a 64-bit application can be useful on a 2 GiB machine is a program with huge but sparse data structures.

You could have a TiB matrix with only a few MiB of data - with a sparse structure only those memory addresses with real data would have physical RAM behind them. All the rest would be "empty virtual memory" - addresses without a page of RAM assigned to them.

 

[AidenShaw]

In another few years, what you're saying about only needing 4GB will seem equally foolish.

Bring on the 64-bitness!

Actually, what I'm saying is that without the 8 GiB of RAM in the system - bragging about 64-bit is just willy-waving. There's almost no point in running a 64-bit operating system on a PowerPC with 512 MiB of RAM. (It does make sense to run 64-bits on an Intel/AMD x64 laptop with 512 MiB of RAM, however, since x64 can be a lot faster in 64-bit mode.)

In a few years, when we have 8 GiB SO-DIMMs (or 32 GiB holographic sugar cubes ) 64-bits will be common in laptops.

In particular, what I'm saying is that a dual-core G4 PowerBook would be a lot more bang-for-the-buck than a G5 PowerBook at this point in time. Most of the complaints here are about the speed of the current 'books, not about their memory addressability.

Note that the dual-core G4 is not stuck with the 133/167 MHz FSB, and also note that a follow-on 64-bit version is planned. It might be a year or two longer for the 64-bit version - but maybe by then we'll have reasonably large memory chips for laptops.

The need for 64-bits in a PPC laptop won't be important until there's both a mass of 64-bit applications and the ability to put enough RAM in a laptop to run them.

I would predict that until OS X is truly 64-bit (that is, when Carbon64/Cocoa64 exist and a GUI app can be 64-bit) the first condition won't be met.

For now, 10.4's 64-bit can only be run in batch mode in a command-line app. That'll be good for VATech, but it really blows for Photoshop and Final Cut - unless you want "batch mode" Final Cut

 

[3Memos]

Aidenshaw, do you have charts on RAM capacity increases through the years? It would be interesting to see what minimum config of RAM is projected to be in 1 year, 2 years, and 3 years down the line. This would be a good predictor of requiring 64-bitness or not. Just as we have hit the Ghz limit, are we to see similar limitations in memory chip densities?

In the immortal words of Bill Gates, "640KB ought to be enough for anybody".

 

[AidenShaw]

Aidenshaw, do you have charts on RAM capacity increases through the years? It would be interesting to see what minimum config of RAM is projected to be in 1 year, 2 years, and 3 years down the line. This would be a good predictor of requiring 64-bitness or not. Just as we have hit the Ghz limit, are we to see similar limitations in memory chip densities?

Your idea, go find them!

I would look, however, at maximum memory capacities supported by real systems. Apple is screaming about 64-bit, but most of their systems top out at 1 GiB to 2 GiB. Only the Xserve and the top two models of the PM G5 support more than 4 GiB - and even then only 8 GiB.

To me, that says much more about what Apple thinks about the need for 64-bit than any chip chart. (And the lame implementation of 64-bit support in 10.4 says even more about how unimportant Apple thinks that 64-bit is.)

(x86 supports up to 64 GiB on 32-bit systems, and has for years. IA64 systems with 1 TiB of RAM are available. EM64T with up to 16 GiB, and AMD64 up to 64 GiB.)

In the immortal words of Bill Gates, "640KB ought to be enough for anybody".

Before you quote, check your facts:
http://www.wired.com/news/politics/0,1283,1484,00.html

"He never said it."

 

[BeoVir]

Aidenshaw, do you have charts on RAM capacity increases through the years? It would be interesting to see what minimum config of RAM is projected to be in 1 year, 2 years, and 3 years down the line. This would be a good predictor of requiring 64-bitness or not. Just as we have hit the Ghz limit, are we to see similar limitations in memory chip densities?

In the immortal words of Bill Gates, "640KB ought to be enough for anybody".

1yr 256mb across the board
2yrs 256mb across the board
3yrs 256mb across the board

And in the fourth year after intense pressure from consumers being unable to boot computers with the stock RAM configuration Apple will increase RAM to 384mb!