Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
dguisinger said:
They may not have good 64 bit support; but 64-bit support is a true waste for all but 1% of the computing market at most.
On what do you base this slight?

And what is "bad 64-bit support", other than what Apple is shipping in Tiger? (and that's bad software support, not a bad ISA)

I haven't noticed anything "not good" about Windows x64 on Intel and AMD 64-bit chips - the entire system is true 64-bit, with a subsystem that transparently runs 32-bit applications.

bigwig said:
A CPU with built-in Rosetta isn't my idea of heaven. I'd bet that translation layer is a significant performance hit. The x86 is extremely register-starved, which doesn't help.
Actually, that translation layer is what makes the Pentium (all variants since the Pentium Pro) so freaking fast. Note that the Pentium doesn't have an "instruction cache" or I-cache per se - the instruction cache really stores the translated (or decoded) instruction stream - an "I-cache hit" picks up the already decoded micro-ops.

The micro-engine has many more registers than are visible at the x86 ISA - which helps a lot. Also, x64 (which unfortunately Apple isn't using, since Mac OSx86 abandons 64-bit support) has twice as many 128-bit SSE registers, and nearly three times as many usable integer registers.

Which, going back to the earlier quote, is why x64 is so important in the Intel world. Not because programs need more than 2 GiB of memory space, but because 64-bit code is usually quite a bit faster. (As opposed to PowerPC, where 64-bit code tends to be somewhat slower than 32-bit code.)

The terms CISC and RISC are archaic
 
Maestro64 said:
What we all have to be concerned about here is what will happen to the quality and reliability of the systems. No one can argue that Apple has the highest quality of all systems manufacturers out there with a few except of lemons that Apple has made I the past.

This is by no means an accident on Apple's part. Part of the reason we all pay so much more for an Apple is becuase of the amount of testing, qualification and attention to details Apple puts in to every product. Intel does not have the same reputation. I concerned that these lower end products being mentioned here will fall well short of a standard apple products.

Plus with the rumors of the pending new Intel product to be announce with the help of Intel's design center I am afraid we will all get the same crap the PC world has been getting. I do not believe Apple had the bandwith or the time to fully test and qualify an Intel design, plus it so hard to transfer lessons learned from previous designs to a new design group. One of the reasons Apple designs are so good is their design teams have been together so long and they learn from the past.

I think it will be prudent before we all buy to find out who actually did the design, Apple or Intel, personally I stay way from anything that has an intel PCB in it, remember intel is a Chip company not a systems company go with what you do best and stayway from everything else.


Lest we forget, OS X has been living a dual life the past how many years? Apple has played with intel processors and OS X all this time. I do not see them lowering their standards just because of a switch to intel processors...
 
sbrhwkp3 said:
Lest we forget, OS X has been living a dual life the past how many years? Apple has played with intel processors and OS X all this time. I do not see them lowering their standards just because of a switch to intel processors...

Precisely - Jobs would not have rushed into a transition to Intel, and the fact that OS X has been running on Intel platforms secretly since its inception proves this. Intel is nothing new, in many respects for Apple, and I have every confidence that with such a landmark transition like this, Jobs is going to ensure they do it right - and right from the start. 😎
 
Maestro64 said:
I believe you are correct, I could not remember which of the 6xx series was not actually used by Apple, I was pulling from memory.

I also know at one point there was suppose to be a 050 chip but it was canned when the deal was struck for the PPC. Also, there was a road map for a 605, 606, 607, but those must have been canned when the G series came along. I Personally think the G series is actually the 6xx series they just renamed them for ease of marketing the new chips.

No. Totally off base.

Motorola's successor to the 040 was the 68060 which shipped late and initially only at 50Mhz, eventually reaching 66Mhz. Apple never used it as the PowerPC 601 emulating the 040 was almost as fast as the 060 and quickly surpassed it. IIRC the 050 had already been allocated to a chipname by Phillips. 060's shipped in some really expensive Amiga CPU cards but I don't think they ever made it into a production computer. I did have a laser printer with one in at one point.

Back then it was Motorola's policy to ship even number chips as the first architectural change and then odd numbers as what were essentially fixes or cheaper versions. so 68000 was first, 68010 fixed bugs. 68020 was 32bit, 68030 essentially the same but built in MMU. and so on.

The 601 was the first generation PowerPC chip. Generation 2 was the 603/604. G3 was Apple's marketing terminology and not used by Motorola or IBM. Just as Apple gave us 'Velocity Engine' which was cringe worthy.
 
Peace said:
My Dev Kit runs better than the G4..
So I'm guessing Mac users are in for a really good surprise😉

[disclaimer] This post in no way means anything specific.It is meant purely as an opinion and does not break the NDA [/disclaimer]

ok, but G5s run better than the G4 so that means nothing. Your NDA is probably safe. 🙂
 
...one more thing

AidenShaw said:
Actually, that translation layer is what makes the Pentium (all variants since the Pentium Pro) so freaking fast. Note that the Pentium doesn't have an "instruction cache" or I-cache per se - the instruction cache really stores the translated (or decoded) instruction stream - an "I-cache hit" picks up the already decoded micro-ops.
There's another advantage to the u-ops architecture - it eliminates any performance cost involved in supporting legacy instruction sets (16-bit, 32-bit, ...).

The legacy instructions exist only in the decoder, and not in the execution engine. The decoder translates the 16-bit instructions into u-ops, which then are run through the 64-bit execution engine just like any other instructions.

The people clammering for a "legacy-free Apple-only superfast chip" miss the fact that the legacy stuff is a few transistors in a corner of the decoder, and eliminating it would have almost no effect on performance.

The dual-core Xeon powers up as a 16-bit x86 chip, and looks for the 16-bit boot loader. (It's odd to see an 8-core Opteron or Xeon running 16-bit DOS from a floppy....) The boot loader and subsequent initialization code then enables 32-bit or 64-bit mode as requested by the OS.

Ontogeny recapitulates phylogeny...
 
If Apple + Intel = Synergy.

I don't mind having a sticker "Intel Inside" or whatever...

I'll just remove it..🙄
 
Darth Bill's Conspiracy...

Seasought said:
Perhaps this is part of the agenda of the "Bill and Melinda Gates Foundation"; establish a healthcare monopoly in overseas developing countries, integrate Microsoft as part of the healthcare/medical industry and not just computer software... Maybe I'm going overboard? Too much conspiracy theory? I don't know, sounds like an idea to me.
Not so far fetched as all that. I seem to recall that a few years back the Gates Foundation (or maybe it was MS, itself) had some kind of pilot program with a school district in... Kansas, Nebraska or the Dakotas... one of those states... Anywayz, as part of the effort to increase learning with computers & media the school(s) received vouchers (or credits) that could be used to buy the computer hardware & software needed... Needless to say, the program specified Windows systems...

I think there was even a whole town that the Gates Foundation and/or Microsoft was going to "wire up" for high speed internet & interactive media... I don't know. It's a few years back in my memory...

Stonecoldcleric said:
Hey, I hate the Gateopoly as much as anyone, but using generosity as a pretense for gaining control and power? That's quite a conspiracy. I think too many would call him on it if he bagan to shove MS down their throats.🙂
Naw, they'd wait until the check cleared. Then they'd tar & feather him...

Though, if he succeeded he could set himself up as "Papa Bill", Windoze-Tater for life... Then he could thumb his nose at the US & EU courts from the beach while drinking his virgin daquiri
 
AidenShaw said:
and nearly three times as many usable integer registers.

To be clear by "usable" you mean named registers, ones that are under direct controll of a compiler or programmer. Those extra registers are still used (at least potentialy) even if not under direct control of a programmer or compiler. Likely not as efficiently but they don't go unused.

AidenShaw said:
Not because programs need more than 2 GiB of memory space

Just to be clear 32 bit address yields an address space of 4 GiB (2^30 = 1GiB, 2^2 = 4) and on average on Mac OS X that gives you a usable address space of around 3-3.5 GiB once you knock out shared memory mappings, etc. Also on average the largest contigous allocation you can make is in the 2-2.5 GiB range because of memory space segmentation from shared mappings, etc.
 
AidenShaw said:
But if the Intel one is cheaper, uses less power, and has zero R&D cost ????

"Centrino" is a much stronger brand that AE ...

Since this story is about "image" - it looks like Apple will be dropping the "Think Different" mindset and will be embracing "Think Same".
Weren't the AE cards made by Lucent, anyways? And I have no clue who makes the chips for the AE-BS... Point is, they were/are pretty much industry-standard chipsets - they'd have to be for certification as 802.11a/b/g...

It was the Airport Software, developed by Apple, that made it different. And that will not change, even if they start using the Centrino bundle... Also, it's important to note that Centrino is not just wireless technology (802.11a/b/g). It is integral to the Napa platform, which includes the cpu (Yonah dual-core to begin), the (optional) Calistoga Intel integrated graphics (think iBook - but likely single-core), the DDR2 memory controller, power management, WiMax/3G options, and PCI-Exp...

Dell, HP, Lenova, et al will have (pretty much) the same Napa platform as Apple (except for whatever TPM Apple "welds on"). However, they are shackled to Windows, so their "counterpart" (used loosely in this case) to Apple's Airport hardware/software duet will "look more like a drunken mule strapped to a rhino's butt that's rollerblading down an avalanche in a too-tight g-string while bellowing Pavrotti tunes in an uncanny imitation of Roseanne Barr" compared to Apple's AE...
 
ha ha

Norse Son said:
However, they are shackled to Windows, so their "counterpart" (used loosely in this case) to Apple's Airport hardware/software duet will "look more like a drunken mule strapped to a rhino's butt that's rollerblading down an avalanche in a too-tight g-string while bellowing Pavrotti tunes in an uncanny imitation of Roseanne Barr" compared to Apple's AE...
Cute image, but the truth is that wireless and networking support in XP is pretty seamless and transparent.

I use several WiFi access points, and XP remembers my keys and automatically connects to the nearest AP (home, office, local coffee shop...). At the airport, I can see all the pay-for options and connect, or plug in my cellular modem PC-card and work about anywhere.

So, make your joke, but Windows does in fact work OK for most people - as evidenced by its market share.
 
Please, do shed some light...

dguisinger said:
That used to be important. But as the powerpc has been evolving, it now has nearly as many instructions as an x86 processor. Plus, an x86 processor no longer executes x86 code, it converts it into microcode and executes that code instead; combining / reducing the real number of instructions executed.

The fact is, with the last G5s apple was falling way behind on performance; they couldn't hit the MHz required to compete against Intel machines which keep improving. At the same time, these new Intel processors do even more instructions per cycle, allowing the clock speed to be reduced resulting in less power and head consumption. The new upcoming Intel processors beat the living day out of the G5s in overall performance and in performance per watt.

They may not have good 64 bit support; but 64-bit support is a true waste for all but 1% of the computing market at most. Floating point and vector math is where it matters for science and games; and 32-bit integer is what matters for office/internet applications. 64-bits is used for programs with HUGE amounts of ram.

bigwig said:
RISC vs CISC is about instruction complexity, not instruction count. Plus many x86 instructions aren't merely complex, they're stupid. SIMD instructions that destroy their inputs, for example, so you're forced to execute extraneous load instructions.

A CPU with built-in Rosetta isn't my idea of heaven. I'd bet that translation layer is a significant performance hit. The x86 is extremely register-starved, which doesn't help.
Okay, you two sound like "gear heads" (no insult, either), so maybe you can both enlighten us...

Tiger seems to be the first time we had the real option of dumping the Classic Environment. I'm guessing that Leopard may not even have Classic included - which I think is past due to die. The big question is AltiVec - I know Intel has their own fancy Vector Processing engine/instructions; so will Apple and developers rewrite apps to recognise Intel's instructions? And how does Opel GL support on x86 stack up - I hope Apple never tries to license & "cobble together" support for Direct-X...

And down the line, probably in MacOS X v10.7 ("alley cat"?), will we start to see universal-binaries being phased out? Will we be able to rely on just loading the Intel-versions of the OS & apps by then? If so, I think the performance would increase slightly - again, I don't know how all that Rosetta stuff works when there's less & less for it to translate from PPC to x86...

Intel will likely have 45nm quad-cores, possibly even "octa-cores"(?), by that point (late 2008-09). And Vista "might" see its first SP by then, but I think the MacOS may have "pulled back the curtain on the Wizard" by then, exposing just how shaky Windows STILL is compared to MacOS X.
 
shawnce said:
To be clear by "usable" you mean named registers, ones that are under direct controll of a compiler or programmer. Those extra registers are still used (at least potentialy) even if not under direct control of a programmer or compiler. Likely not as efficiently but they don't go unused.

Just to be clear 32 bit address yields an address space of 4 GiB (2^30 = 1GiB, 2^2 = 4) and on average on Mac OS X that gives you a usable address space of around 3-3.5 GiB once you knock out shared memory mappings, etc. Also on average the largest contigous allocation you can make is in the 2-2.5 GiB range because of memory space segmentation from shared mappings, etc.
My head hurts... owww... get them out of my mind! They're screaming at me... They want me to do dirty things with window blinds...
 
AidenShaw said:
Cute image, but the truth is that wireless and networking support in XP is pretty seamless and transparent.
...
So, make your joke, but Windows does in fact work OK for most people - as evidenced by its market share.
Yes, yes... Point noted! I guess I just don't like the "look & feel" of it. I owned one Toshiba laptop in my life, with WinME (or was it 98?), and I hated every minute of it until I could get back to the Mac. Then, as soon as MacOS 10 v10.1 came out I "shunned" Classic - I think the only 2-3 times I went back to Classic was a couple years back when a "computer illiterate" friend got an old 7200 that he wanted to make work (and then I had to poke around a few minutes just to remember how things "worked"). Luckily that friend bought an iMac G5 (orig.) and started using Panther.

My point is I like MacOS X - I even liked the Dock from the beginning. It represents the future. MacOS 9 belongs in a museum. And Windows, with all the legacy code buried below its foundations, still looks second-rate to me. XP's desktop still looks cluttered, and the Start Menu is like trying to walk through one of those topiaries (hedge row mazes?), but at night, and one that was designed by Dr. Seuss & Stephen King...

And is XP still just a shell on top of DOS? How long can it possibly take for the Windows' world to shed DOS? I mean, when was the last release of Lotus 1-2-3? Longhorn was supposed to be a ground-up rewrite, yes(?), but it seems to be taking them an eternity to release even a "viable" beta - unless they "delay" more features.
 
shawnce said:
To be clear by "usable" you mean named registers, ones that are under direct controll of a compiler or programmer.
Yes, you are correct that I mean "named" or "architectural" registers.

The x86 ISA has 8 integer registers, but 3 of those are usually dedicated to certain uses, leaving 5 freely usable. The x64 ISA has 16 integer registers, which leave 13 to be freely usable. "13" is more than 2.5 times 5, so I said "nearly 3 times".

shawnce said:
Just to be clear 32 bit address yields an address space of 4 GiB (2^30 = 1GiB, 2^2 = 4) and on average on Mac OS X that gives you a usable address space of around 3-3.5 GiB once you knock out shared memory mappings, etc. Also on average the largest contigous allocation you can make is in the 2-2.5 GiB range because of memory space segmentation from shared mappings, etc.
Yes, this is true for OS X. For Windows, a process normally gets 2 GiB of usable space, and the other 2 GiB is used for the system and I/O mappings. (Windows Server has an option to make the split 3GiB/1GiB.)
________________

My real point is that for PPC, there's no real reason to go to 64-bit addressing (that is, true 64-bit) unless you're hitting that memory barrier in the 2 GiB to 3.something GiB range.

(Note that 64-bit is not needed for supporting more than 4 GiB of physical memory - 32-bit OSX can do it, and 32-bit Windows can also. 64 GiB total memory is the max for 32-bit Windows on 32-bit chips.)

For Intel/AMD, however, the increase in the number of registers means that programs with no need whatsoever for large memory will benefit from a true 64-bit O/S. Many things will run faster if recompiled for a 64-bit O/S.
 
Some of these are very old

zv470 said:
hmm, strange love decrangle:

Excel on Apple,
MS DOS on Intel,
Windows on MS DOS on Intel,
Office on Windows on MS DOS on Intel,
Apple on PowerPC,
Office on Windows on MS DOS on AMD,
iTunes on Motorola ROKR,
Microsoft uses PowerPC chips in XBox,
Palm Treo running Windows Mobile,
Apple uses Intel processors,

Whats next?

😀


Excel on Apple, that's over 20 years old now.
MS DOS on Intel, That's probably 25 years old now.

Bill the TaxMan
 
Norse Son said:
Tiger seems to be the first time we had the real option of dumping the Classic Environment.
My "real option" was to dump Classic on 10.1. Never needed it.

I'm guessing that Leopard may not even have Classic included - which I think is past due to die.
AFAIK Classic won't run on Intel-based Macs, which will apparently will use 10.4.x. So that gives Apple a head start for dropping it completely in 10.5 even if it were still possible to run on PPC-based Macs. I've long been warning my Classic-using friends to ditch it ASAP since it's eventually a dead end. But there will still be people using it long after it's officially EOL'd, just like systems still running pre-X Mac OS.

Anyone know anyone still using floppy drives? 😱
 
Power on pins 5 & 6

Randall said:
Yeah that's a good point. FireWireless is a very interresting concept, seeing as how (at least in my experience) that the majority of FW400 hardware is 6-pin (external Hard Drives etc.) and as such, they utilize the built in power in pins 5 and 6. That said, we all know that there is no such thing as wireless power, so these devices would have to have an alternate power source in order to utilize FireWire wireless or whatever it's called. Only the 4-pin FW400 stuff will be able to use this new technology.


The power supplied on pins 5 & 6 only supplies enough power in just a few cases. Most of the time the power is supplied by a power supply just for that item. Most 2.5" hard drives are designed to be able to use bus (pin 5 & 6) power. FW card holders work the same way. Few 3.5" drives have a low enough power requirement to allow bus power. FW printers & scanners would have their own PS.

The only pats that are ever wireless is the signal. Power is always supplied locally.

Bill the TaxMan
 
sjk said:
My "real option" was to dump Classic on 10.1. Never needed it.

AFAIK Classic won't run on Intel-based Macs, which will apparently will use 10.4.x. So that gives Apple a head start for dropping it completely in 10.5 even if it were still possible to run on PPC-based Macs. I've long been warning my Classic-using friends to ditch it ASAP since it's eventually a dead end. But there will still be people using it long after it's officially EOL'd, just like systems still running pre-X Mac OS.

Anyone know anyone still using floppy drives? 😱
I stopped using it, but for some reason it seemed that every time I upgraded (10.1-10.2-10.3) it never gave me the option of not installing Classic. And I sometimes wonder if the only reason it let me do it with Tiger was because I took it to an Apple store... See, my 3rd DVD-ROM drive died on my Pismo around August of 2004 (the first 2 died within 3-4 months of buying the lemon); so I could not boot from my install disk to run disk utility.

I had the Apple Store erase my hard drive and use a PwrMac G5 (via FrWr) to install Tiger... I desparately need Apple to release a PowerBook dual-core Yonah at MWSF...

Oh, by the way, I think I've seen a few floppies at work - on the Windows PCs. They make good coffee coasters, or frisbees (watch'em smash on a concrete wall).
 
Norse Son said:
And is XP still just a shell on top of DOS? How long can it possibly take for the Windows' world to shed DOS?

XP is not a shell on top of DOS. Windows NT, Windows 2000, Windows XP and Windows Server 2003 don't have the DOS legacy at all.

Windows ME has about as much in common with Windows XP as Mac OS 9.2 has with Mac OS X 10.4. Both WinME and Mac OS 9.2 would crash if you looked at them funny. With good hardware and stable drivers both Windows XP and Mac OS X 10.4 are rock solid.
 
ktlx said:
XP is not a shell on top of DOS. Windows NT, Windows 2000, Windows XP and Windows Server 2003 don't have the DOS legacy at all.

Windows ME has about as much in common with Windows XP as Mac OS 9.2 has with Mac OS X 10.4. Both WinME and Mac OS 9.2 would crash if you looked at them funny. With good hardware and stable drivers both Windows XP and Mac OS X 10.4 are rock solid.
How is XP related to DOS? The "terminal" is DOS, so there has to be some relation. 😕
 
AidenShaw said:
Cute image, but the truth is that wireless and networking support in XP is pretty seamless and transparent.

I use several WiFi access points, and XP remembers my keys and automatically connects to the nearest AP (home, office, local coffee shop...). At the airport, I can see all the pay-for options and connect, or plug in my cellular modem PC-card and work about anywhere.


XP is a POS when it comes to Wireless. I'm speaking from someone who is at this moment typing on a IBM X31 Thinkpad with XP W/ SP2. XP's internal app for using WIFI is a kludge. That is why I'm using Intel's WIFI tools. Disabled XP's built in crap and let Intel manage the radio.
Then lets talk about WIFI power management in XP. Sometimes you are lucky to get the thing to go to sleep. I sit there listening to my ThinkPad beep. . .beep. . .beep. . .beep. . .beep. . .beep. . .beep. . .beep. . .for 2 minutes waiting for the standby light to go on...open the lid. Turn off the radio then try it again. Its gotten a heck of a lot better with SP2 and the latest and greatest drivers but its still nowhere as good as Apple's.
 
ktlx said:
XP is not a shell on top of DOS. Windows NT, Windows 2000, Windows XP and Windows Server 2003 don't have the DOS legacy at all.

Windows ME has about as much in common with Windows XP as Mac OS 9.2 has with Mac OS X 10.4. Both WinME and Mac OS 9.2 would crash if you looked at them funny. With good hardware and stable drivers both Windows XP and Mac OS X 10.4 are rock solid.


The voice of reason and sanity.
 
EricNau said:
How is XP related to DOS? The "terminal" is DOS, so there has to be some relation. 😕

The terminal is NOT DOS. The "terminal" is cmd.exe. Go to any 2K XP system. In the start menu it is called a Command Prompt. The command prompt is simple a command interpreter that sits on top of the OS. Its a shell is what it is.
Oh and before someone tells me about well what about hitting F8 and selecting command prompt. The OS simply loads cmd.exe instead of explorer.exe

PS- MS is revamping the command line in Vista so it becomes much more powerful. Most apps will be accessable from the command line now. Something the OS X and pretty much the entire *nix world has had well....forever.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.