OK, guys, I ran across a few articles on ZDNet that you might be interested in. In fact, it's almost as if these authors are lurking on AI, taking notes on what were discussing. In particular, the following two articles caught my attention.
The first deals with all of the various *flavors* of Windows (24 current versions!) that are on the market. It discusses how Gates and Co. argued under oath that forking and fragmentation would be extremely *bad* for customers, businesses, developers and especially Microsoft. The part they didn't mention was that it was apparently *OK* if it was Microsoft who was doing the forking and fragmenting. Odd, eh? This ties directly into what a bunch of us have been discussing here at AI regarding Windows support for various platforms. It really does seem to be turning into a nightmare for them. Ironically, it also ties into the second article which talks about how AMD is essentially betting the house on the Opteron. In the end, I get the feeling that x86 will be around a long, long time -- exactly as I was surmising in earlier threads. Since it's already been brought up here, I'd like to delve deeper into the discussion and see what people like Programmer, Amorph and Matsu think. Personally I just see the whole Windows world in an ever accelerating state of decay, confusion and fragmentation and basically stuck with 32-bit x86 for a very long time. I'd like to hear input from others as well and see what this might mean for Apple and the 970 with regard to 64-bit computing. Here are the articles:
. . . Personally I just see the whole Windows world in an ever accelerating state of decay, confusion and fragmentation and basically stuck with 32-bit x86 for a very long time. I'd like to hear input from others as well and see what this might mean for Apple and the 970 with regard to 64-bit computing.
The Windows world will certainly milk the 32-bit x-86 for as long as it's profitable, but I believe they are looking for a way out. Who knows what they will try. If they take a clue from IBM, Intel may produce a mini version of the Itanium processor, like the 970 is a mini version of the Power4. That's one possibility. It will take them much longer than Apple and IBM to make such a transition, since the Itanium will not run 32 bit x-86 code native. A much bigger software issue for people to change over.
This gives Apple a chance to build market share for a while I believe.
Edit: Thanks to wfzelle for noticing I had the word "native" in the wrong place. Yes, people will not want to run all their existing software in emulation if they switch from x-86 to Itanium.
Last night I saw an "expert" at CNN news state that the big advantage of 64 bit CPUs is that they can handle twise the data chunks of 32 bits CPUs and thus are twise as fast This in context of the new Hammer CPU from AMD.
What is important is that Apple rely on the strenghts of IBMs PPC CPUs and the strenght of Apples implementation and not rely on bad things happening on the Wintel side.
Regarding IBM selling Intel CPUs in blade servers. The most important thing for IBM is that the customer buy the blade server from IBM and not from an other vendor! A secondary thing is that IBM when prefer that the blade box has as much IBM stuff on the inside as well be it the IBM 970 CPU or IBM hard disks or IBMs OS or other software.
So even if IBM is happy to sell cutomers Intel CPUs and Maxtor har disks they are even happier with IBM CPUs and hard disks, and they will work hard to be "even happier"
I believe that this would be in regard to the "970 on track" part of the title and to perhaps help dispell any remaining doubts about the imminent arrival...
Interview with Tim Sweeney (Unreal Tournament) on www.firingsquad.com about porting the code to the 64 bit AMD solution...
I was wondering, if you could use a 64 bit register to work simultaniously on 2 32 bit words.
Today we don´t have single thread situations anymore. On the software side of things a lot works in "parallel" and without interrelations.
Supose you detect in the load unit what kind of instruction is going to be worked on. Supose you find 2 32 bit instructions with equal opcodes. I guess this will happen frequently. Then you can load the 2 32 bit words in the 64 bit register. Of course you need to double the carry bits, etc..
Instruction reordering is reality today. So why not extended it in the proposed way?
Another variation would be to present the 64 bit register to 32 bit APIs as 2 32 bit registers. This has also been done before.
That way (enough ALUs need to be present) the cycle speed can doubled.
I'm positive about the information and have, and even if you don't believe me now you'll get to see sometime after June.
Fair enough.
Quote:
You'll see IBM 970 64bit w/ "Altivec Velocity Engine Extreme"
What would this entail?
Quote:
You will see Mac OS 10.3 available on x86 Intel Processors.
I share Amorph's concern about the travails of such an avenue whether by design or not. Apple might not want to repeat the clones failure, but they might not have complete control of the situation.
Quote:
64bit OS, including rewritten finder to take advantage! Jokes have been made about "finder extreme".
...OSXFS - New filesystem, based off of BeOS Filesytem. Very cool features built right into the finder! 10.3 Includes the ability to lay windows out with rich HTML based on the webcore. You can use "simple/regular" finder mode to bypass these. The ability to assign files categories like ID3 tags. Ever finder window can be searched realtime similar to the itunes search thanks to OSXFS. You can find all files that were assigned "category" business in 1 second flat.
I bet the engineers have as much fun with the marketing names as we do. It will be interesting to see ifthis is actually a new FS, a modified HFS++++! or something to do with the VFS model Amorph has mentioned before. I cringe at the idea of using html in Finder windows for two reasons: do we really want the IE model of file organizer as web browser?, plus if it's for rendering neat stuff, why use html?, why not QT sprites and crap like that?
Quote:
New Navigation services! Three different ways to drill down files, you can even click a button to "spot" the folder. This means your cursor changes and you can move through the finder to locate the folder then hit enter key to finish saving. You can also assign the file categories at this time or the application can by default. files created by apple sound track editor can automatically be assigned "audio", so could files from an updated pro tools.
Lemme guess: icon, list and column? The "spotting" feature sounds interesting, but I also assume you would be able to do a real-time search in the save or open dialog? Categories sounds a lot like labels, but are they only one criteria per file like labels?
Quote:
Bundles (some of you call them piles) work sort of like this, you could select 20 files and say "these are all the files for my project" monitor them. when you go to move 1 file it will automatically copy the rest with it. You can also condense them to look like a folder. Think files on different hard drives, different networks all as one project "folder" all updating and all copying when you move them. Very cool.
Bundles, packages, both taken. Someone referred to them as "stacks," maybe that would be the official name? Piles makes me reach for Preparation H. Anyhoo, this could be a nice idea, but it does sound like the concept of "smart folders" too. How could you differentiate them from regular folders, or what exactly would their relationship be to them? Could these "bundles" contain sub-folders or sub-bundles?
[quote]New dock, so many new features, split mode, text mode, "start menu" like mode for people who prefer windows. dock is very slick in this version, better than the start menu.[quote]
...and the question on everyone's mind: would it have spring-loaded pop-ups?!
Quote:
new mail.app uses webcore for parsing html/mime email now, much faster and better, offers exchange server support, and many other new features, including a new "view" that looks exactly like outlook 2003 preview for windows. (you can still use old ones) mail is much faster.
The use of webcore is a no-brainer, but this third mention of this idea makes me windows why Apple feels the need to make options to imitate Windows more?
Quote:
much improved sharing controls for sharing with windows and unix enviroments, lots of complex options.
Hopefully these things will be more GUI/Finder-oriented.
Quote:
new iphoto (very very very very very very very very fast)
This would be nice, but how would they do it without completely redoing the way iPhoto displays the pictures and uses the library?
I was wondering, if you could use a 64 bit register to work simultaniously on 2 32 bit words.
Today we don´t have single thread situations anymore. On the software side of things a lot works in "parallel" and without interrelations.
Supose you detect in the load unit what kind of instruction is going to be worked on. Supose you find 2 32 bit instructions with equal opcodes. I guess this will happen frequently. Then you can load the 2 32 bit words in the 64 bit register. Of course you need to double the carry bits, etc..
Instruction reordering is reality today. So why not extended it in the proposed way?
Another variation would be to present the 64 bit register to 32 bit APIs as 2 32 bit registers. This has also been done before.
That way (enough ALUs need to be present) the cycle speed can doubled.
Perhaps I´m missing something obvious.
See ya.
Cocoa Tree
Missing several things, actually.
1) The execution resources (ALU, FPU, VPU etc.) are not the limiting factors in modern processor design, nor do they occupy a lot of real estate on the die. The hard work is done in the scheduling and tracking of instructions. Adding another ALU would have been done a long time ago otherwise, they are relatively tiny.
2) Identical opcodes often wouldn't fit well into a single register, consider rotate/shift instructions. The added complexity of the ALU would probably be greater than that of adding another ALU.
3) PPC is not short of registers, 32 is OK, most compilers cnnot yet use the existing ones efficiently. Adding another 32 architectural registers would alter the whole ISA for very little benefit, and would also considerably complicate the design of the register file.
I'll add a few more questions to BuonRotto's list for Inky, who's either an inspired liar, or the first guy to break through the Panther information firewall...
- how are dialog windows better?
- what are the "new themes"?
- how exactly does a split, or text-based Dock work? What does the Apple Start menu look like, how does it work?
- what's the system of window management, incl. minimization?
I'll add a few more questions to BuonRotto's list for Inky, who's either an inspired liar, or the first guy to break through the Panther information firewall...
Hmm, more details and answers to questions asked would be far more convincing.
Ah, well, time to collect some nice-sized rocks...
Quote:
Originally posted by BuonRotto
Hey, who cares, it's fun to consider the ideas. Certainly some intriguing stuff to consider regardless of how "confirmed" it is.
Yeah, definitely, I'm not a **confirmation** diehard. Ideas are what matter. But though I love to speculate as much as next AI attendee, it's also really nice to have a nugget -- just a scrap! -- of semi-reliable sounding information to speculate from. Or better yet, a conceptual direction.
For example, will Apple make the next iteration of the Dock as Windows-friendly as Inky suggest...? I had considered and speculated on that, but rejected it as just too against the contrarian, do-it-our-own-way Apple DNA. That would be something chew on... Speculation can go in a dozen different valid directions, and having even just a hint of Apple's way makes a big difference. That's what I'm often looking for among the many would-be leaks and foggy glimpses.
Stated by Steve Jobs at its annual share holders meeting when asked about clock speed and it's low performing chips..
"there will be a time" when Apple would speak about its relationship with Motorola, and that the particular shareholder who asked the question would be invited, if he desired.
Stated by Steve Jobs at its annual share holders meeting when asked about clock speed and it's low performing chips..
"there will be a time" when Apple would speak about its relationship with Motorola, and that the particular shareholder who asked the question would be invited, if he desired.
Stated by Steve Jobs at its annual share holders meeting when asked about clock speed and it's low performing chips..
"there will be a time" when Apple would speak about its relationship with Motorola, and that the particular shareholder who asked the question would be invited, if he desired.
Comments
The first deals with all of the various *flavors* of Windows (24 current versions!) that are on the market. It discusses how Gates and Co. argued under oath that forking and fragmentation would be extremely *bad* for customers, businesses, developers and especially Microsoft. The part they didn't mention was that it was apparently *OK* if it was Microsoft who was doing the forking and fragmenting. Odd, eh? This ties directly into what a bunch of us have been discussing here at AI regarding Windows support for various platforms. It really does seem to be turning into a nightmare for them. Ironically, it also ties into the second article which talks about how AMD is essentially betting the house on the Opteron. In the end, I get the feeling that x86 will be around a long, long time -- exactly as I was surmising in earlier threads. Since it's already been brought up here, I'd like to delve deeper into the discussion and see what people like Programmer, Amorph and Matsu think. Personally I just see the whole Windows world in an ever accelerating state of decay, confusion and fragmentation and basically stuck with 32-bit x86 for a very long time. I'd like to hear input from others as well and see what this might mean for Apple and the 970 with regard to 64-bit computing. Here are the articles:
News: AMD's future rides with Opteron
http://zdnet.com.com/2100-1103-997529.html
News: Windows faces new competition: Itself
http://zdnet.com.com/2100-1104-997509.html
--
Ed
Originally posted by Ed M.
. . . Personally I just see the whole Windows world in an ever accelerating state of decay, confusion and fragmentation and basically stuck with 32-bit x86 for a very long time. I'd like to hear input from others as well and see what this might mean for Apple and the 970 with regard to 64-bit computing.
The Windows world will certainly milk the 32-bit x-86 for as long as it's profitable, but I believe they are looking for a way out. Who knows what they will try. If they take a clue from IBM, Intel may produce a mini version of the Itanium processor, like the 970 is a mini version of the Power4. That's one possibility. It will take them much longer than Apple and IBM to make such a transition, since the Itanium will not run 32 bit x-86 code native. A much bigger software issue for people to change over.
This gives Apple a chance to build market share for a while I believe.
Edit: Thanks to wfzelle for noticing I had the word "native" in the wrong place. Yes, people will not want to run all their existing software in emulation if they switch from x-86 to Itanium.
Originally posted by snoopy
It will take them much longer than Apple and IBM to make such a transition, since the Itanium will not run native 32 bit x-86 code.
That's not true. It's just slow as hell. Personally I see x86-64 winning out, it allows for a much more gentle transition to 64 bit.
What is important is that Apple rely on the strenghts of IBMs PPC CPUs and the strenght of Apples implementation and not rely on bad things happening on the Wintel side.
Regarding IBM selling Intel CPUs in blade servers. The most important thing for IBM is that the customer buy the blade server from IBM and not from an other vendor! A secondary thing is that IBM when prefer that the blade box has as much IBM stuff on the inside as well be it the IBM 970 CPU or IBM hard disks or IBMs OS or other software.
So even if IBM is happy to sell cutomers Intel CPUs and Maxtor har disks they are even happier with IBM CPUs and hard disks, and they will work hard to be "even happier"
something is up www.apple.se
"We are busy updating the store for you and will be back within the hour."
My guess is bigger iPODs or price adjustments or other mundane things, but hope springs eternal
http://zdnet.com.com/2100-1103-997788.html
Any thoughts?
--
Ed
Interview with Tim Sweeney (Unreal Tournament) on www.firingsquad.com about porting the code to the 64 bit AMD solution...
http://firingsquad.gamers.com/featur...ew/default.asp
I was wondering, if you could use a 64 bit register to work simultaniously on 2 32 bit words.
Today we don´t have single thread situations anymore. On the software side of things a lot works in "parallel" and without interrelations.
Supose you detect in the load unit what kind of instruction is going to be worked on. Supose you find 2 32 bit instructions with equal opcodes. I guess this will happen frequently. Then you can load the 2 32 bit words in the 64 bit register. Of course you need to double the carry bits, etc..
Instruction reordering is reality today. So why not extended it in the proposed way?
Another variation would be to present the 64 bit register to 32 bit APIs as 2 32 bit registers. This has also been done before.
That way (enough ALUs need to be present) the cycle speed can doubled.
Perhaps I´m missing something obvious.
See ya.
Cocoa Tree
Originally posted by inkhead
I'm positive about the information and have, and even if you don't believe me now you'll get to see sometime after June.
Fair enough.
You'll see IBM 970 64bit w/ "Altivec Velocity Engine Extreme"
What would this entail?
You will see Mac OS 10.3 available on x86 Intel Processors.
I share Amorph's concern about the travails of such an avenue whether by design or not. Apple might not want to repeat the clones failure, but they might not have complete control of the situation.
64bit OS, including rewritten finder to take advantage! Jokes have been made about "finder extreme".
...OSXFS - New filesystem, based off of BeOS Filesytem. Very cool features built right into the finder! 10.3 Includes the ability to lay windows out with rich HTML based on the webcore. You can use "simple/regular" finder mode to bypass these. The ability to assign files categories like ID3 tags. Ever finder window can be searched realtime similar to the itunes search thanks to OSXFS. You can find all files that were assigned "category" business in 1 second flat.
I bet the engineers have as much fun with the marketing names as we do.
New Navigation services! Three different ways to drill down files, you can even click a button to "spot" the folder. This means your cursor changes and you can move through the finder to locate the folder then hit enter key to finish saving. You can also assign the file categories at this time or the application can by default. files created by apple sound track editor can automatically be assigned "audio", so could files from an updated pro tools.
Lemme guess: icon, list and column?
Bundles (some of you call them piles) work sort of like this, you could select 20 files and say "these are all the files for my project" monitor them. when you go to move 1 file it will automatically copy the rest with it. You can also condense them to look like a folder. Think files on different hard drives, different networks all as one project "folder" all updating and all copying when you move them. Very cool.
Bundles, packages, both taken.
[quote]New dock, so many new features, split mode, text mode, "start menu" like mode for people who prefer windows. dock is very slick in this version, better than the start menu.[quote]
...and the question on everyone's mind: would it have spring-loaded pop-ups?!
new mail.app uses webcore for parsing html/mime email now, much faster and better, offers exchange server support, and many other new features, including a new "view" that looks exactly like outlook 2003 preview for windows. (you can still use old ones) mail is much faster.
The use of webcore is a no-brainer, but this third mention of this idea makes me windows why Apple feels the need to make options to imitate Windows more?
much improved sharing controls for sharing with windows and unix enviroments, lots of complex options.
Hopefully these things will be more GUI/Finder-oriented.
new iphoto (very very very very very very very very fast)
This would be nice, but how would they do it without completely redoing the way iPhoto displays the pictures and uses the library?
Originally posted by cocoa tree
Hi all,
I was wondering, if you could use a 64 bit register to work simultaniously on 2 32 bit words.
Today we don´t have single thread situations anymore. On the software side of things a lot works in "parallel" and without interrelations.
Supose you detect in the load unit what kind of instruction is going to be worked on. Supose you find 2 32 bit instructions with equal opcodes. I guess this will happen frequently. Then you can load the 2 32 bit words in the 64 bit register. Of course you need to double the carry bits, etc..
Instruction reordering is reality today. So why not extended it in the proposed way?
Another variation would be to present the 64 bit register to 32 bit APIs as 2 32 bit registers. This has also been done before.
That way (enough ALUs need to be present) the cycle speed can doubled.
Perhaps I´m missing something obvious.
See ya.
Cocoa Tree
Missing several things, actually.
1) The execution resources (ALU, FPU, VPU etc.) are not the limiting factors in modern processor design, nor do they occupy a lot of real estate on the die. The hard work is done in the scheduling and tracking of instructions. Adding another ALU would have been done a long time ago otherwise, they are relatively tiny.
2) Identical opcodes often wouldn't fit well into a single register, consider rotate/shift instructions. The added complexity of the ALU would probably be greater than that of adding another ALU.
3) PPC is not short of registers, 32 is OK, most compilers cnnot yet use the existing ones efficiently. Adding another 32 architectural registers would alter the whole ISA for very little benefit, and would also considerably complicate the design of the register file.
michael
- how are dialog windows better?
- what are the "new themes"?
- how exactly does a split, or text-based Dock work? What does the Apple Start menu look like, how does it work?
- what's the system of window management, incl. minimization?
and
- care to provide a screenshot?
Originally posted by Hobbes
I'll add a few more questions to BuonRotto's list for Inky, who's either an inspired liar, or the first guy to break through the Panther information firewall...
Given this, I'd guess the former...
http://www.theregister.co.uk/content/39/30372.html
Originally posted by inkhead
I know this won't prove anything but it might help some of you see that I do know what I'm talking about :
http://www.theregister.co.uk/content/39/30372.html
Hmm, more details and answers to questions asked would be far more convincing.
Ah, well, time to collect some nice-sized rocks...
Originally posted by BuonRotto
Hey, who cares, it's fun to consider the ideas. Certainly some intriguing stuff to consider regardless of how "confirmed" it is.
Yeah, definitely, I'm not a **confirmation** diehard. Ideas are what matter. But though I love to speculate as much as next AI attendee, it's also really nice to have a nugget -- just a scrap! -- of semi-reliable sounding information to speculate from. Or better yet, a conceptual direction.
For example, will Apple make the next iteration of the Dock as Windows-friendly as Inky suggest...? I had considered and speculated on that, but rejected it as just too against the contrarian, do-it-our-own-way Apple DNA. That would be something chew on... Speculation can go in a dozen different valid directions, and having even just a hint of Apple's way makes a big difference. That's what I'm often looking for among the many would-be leaks and foggy glimpses.
"there will be a time" when Apple would speak about its relationship with Motorola, and that the particular shareholder who asked the question would be invited, if he desired.
Originally posted by Producer
Stated by Steve Jobs at its annual share holders meeting when asked about clock speed and it's low performing chips..
"there will be a time" when Apple would speak about its relationship with Motorola, and that the particular shareholder who asked the question would be invited, if he desired.
That sounds very ominous.
Odd use of language though isn't it?
--
Ed
Originally posted by Ed M.
Can someone post a link to that SJ quote about Motorola? When was that stated?
http://www.thinksecret.com/
It is the Notes From Apple Shareholders' Meeting, near the end.
Originally posted by Producer
Stated by Steve Jobs at its annual share holders meeting when asked about clock speed and it's low performing chips..
"there will be a time" when Apple would speak about its relationship with Motorola, and that the particular shareholder who asked the question would be invited, if he desired.
sounds like a trial is coming up....