Microsoft To Dump 32-Bit After Vista

SlinkySausage writes "Microsoft has used its annual hardware engineering conference to announce that Windows Vista and Server 2008 will be the last versions of Windows capable of booting on 32-bit CPUs such as Intel Pentium 4 and Core Duo. AMD, which introduced 64-bit CPUs early — much to the derision of Intel, which said there was no use for them at the time — must be delighted with Microsoft's decision. Owners of first-generation Intel Macs that used (32-bit only) Core Duo CPUs may not be so happy knowing that Vista will be the last Windows they will be able to run."

Comments

Don't fall into the trap

[ArcherB]

Windows Vista and Server 2008 will be the last versions of Windows capable of booting on 32-bit CPUs such as Intel Pentium 4 and Core Duo Before anyone falls into the trap I almost fell into, please note that the Core Duo is not the 64-bit capable Core2 Duo.

Re:Don't fall into the trap

[hackstraw]

Before anyone falls into the trap I almost fell into, please note that the Core Duo is not the 64-bit capable Core2 Duo.

Thanks for the heads up.

I find the "PC" world funny. I've used higher end equipment for quite some time, and I've found myself back into "commodity" land and its pretty confusing.

The summary says "AMD, which introduced 64-bit CPUs early". Huh?

DEC Alpha chips were introduced in 1992 and were 64-bit. SPARC went 64-bit in 1995. MIPS went 64-bit in 1991. PA-RISC in 1996.

AMD came out with 64bit/32bit hybrid chips in what? 2002-2003?

How is this early?

Also, Intel introduced the Itanium, a pure 64-bit chip in 2001. They had a strange i860 chip in the late 80s that was 64-bit.

I've been running 64-bit linux for about 10 years plus or minus 6 months.

To me, I find the x86_64 stuff to be a hack and late to the game. The only reason its remotely interesting is that its cheap, but calling this new or interesting is completely wrong.

Re:Don't fall into the trap

[Chas]

"DEC Alpha chips were introduced in 1992 and were 64-bit. SPARC went 64-bit in 1995. MIPS went 64-bit in 1991. PA-RISC in 1996."

And how many did you see, across the board, in consumer grade "Bob Everyman" systems?

"How is this early?"

Because Intel was basically telling people (rightly) that they really didn't need 64-bit on consumer workstations and laptops for another 10-15 years. AMD said "we've got it and we're releasing it". The adoption of the x64 stuff by Intel only reinforces this.

"Also, Intel introduced the Itanium, a pure 64-bit chip in 2001."

Which SUCKS for general computing use, and requires ridiculously complex "intelligent" compilers to eke decent performance out of your code.

"They had a strange i860 chip in the late 80s that was 64-bit."

Misconception. The data buses were 64-bits wide. And the FPU register depth was 64-bit (however, today's FPU have registers 80-bit deep by comparison).

"I've been running 64-bit linux for about 10 years plus or minus 6 months."

When did this become about dicksizing?

"To me, I find the x86_64 stuff to be a hack and late to the game."

As has been said about just about EVERY x86 development for most of the last 30 years. However, there's a reason why x86 is still the dominant platform extant. Underneath all the hacks and kludges and other cruft, the basic platform is stable, completely documented, and TIME TESTED. As such, the development community is orders of magnitude larger than any other. And, as a byproduct, the user community, who only cares about getting their hands on their personal favorite app, is several more orders of magnitude larger still compared to the user space of every other platform extant COMBINED.

Yes, inertia may NOT be the best reason for holding on to a platform. However, as long as the platform does what's required of it, what's the legitimate bitch?

Re:So this means

[ChetOS.net]

Windows XP came out in October 2001. Vista in November 2006. That is little over 5 years, not 20.

However, I understand your sentiment.

as the owner of a first gen PPC mac....

[gumbi west]

I've got 10.4.9 running on a computer I bought in 2000, that makes it 7 years, not 2 for running the latest OS X. It is true though that there were 1998 computer that can not run OS X, and OS X was released in 2001--so that was a 3 year window. But at the time, 10.1 wasn't really ready for prime time, and OS 9 was still the main OS.

Re:as the owner of a first gen PPC mac....

[Nimey]

You don't have a first-gen PPC Mac unless it has a PowerPC 601 processor. *Maybe* a 603. Those would be from about 1994.

Let me just be the first to say...

[Anonymous Coward]

"You'll be able to address every bit in the known universe with the memory map on *that* one." ...nobody will ever need to address more than every bit in the known universe.

Re:as the owner of a first gen intel mac....

[Bert64]

Well, mac users often keep their machines for much longer than windows users... I still use a 400mhz G4 and it runs OSX Tiger perfectly well.

Tho, it does bring up an important question....
Why did Apple start with core duo processors? They could have made a clean break to 64bit x86 hardware, instead of going 32bit and having to migrate later?

Re:as the owner of a first gen intel mac....

[GlobalEcho]

I believe semilog is more informative (as we discussed). One can see it in this sheet

Re:As though any processor

[Jozer99]

Quite a few people. I know someone who celebrated the XP launch by buying a new computer, 1.6 GHz P4 Geforce 4 Graphics 512MB of RAM 80GB HD DVD and CD-RW That runs Vista just fine, as you can see, and was actually not that expensive back in the latter half of 2001 (probably around the $1500 mark).

Re:as the owner of a first gen intel mac....

[AKAImBatman]

But you were inviting conclusions about Vista's RAM usage from it; thus, implicitly treating it as if it were a reasonable analysis.

"Since I've been itching to try out the new Charts feature on Google Spreadsheets anyway, I threw together a spreadsheet of the Windows memory requirements," hardly sounds like a reasonable analysis. Especially after the parent poster explicitly stated that such growth is not linear. It was information, and information only. i.e. A spreadsheet of Windows Memory Requirements. Exactly as stated.

Nonwithstanding that in any linear plot of explonentially increasing numbers, the most recent one is always going to far dominate the ones that came before.

I think that's obvious. However, the memory requirements did double with each generation. (The exception was 95 -> 98 which had a 4x increase.) The 8x increase with Vista is a bit of a first for Windows. Which is a true statement regardless of the time between the OSes. However, that doesn't mean it's an analysis. That's your invention, not mine.

Except you are wrong

[Sycraft-fu]

They don't require MS's signature on drivers, just a signature. You can self sign. Many drivers are signed that way. The Creative Labs audio drivers, for example. Those pieces of shit wouldn't pass WHQL validation, but they don't have to. Creative self signs them. Likewise the Truecrypt kernel module is self signed (though not because it sucks).

This doesn't stop free, open drivers, just requires the people releasing them to sign them. If the driver isn't MS signed, Vista pops up a little box saying "The driver is signed by Company X, do you want to trust them and install?" Click yes and it installs, and you aren't bothered again.

The idea isn't to stop third parties, it is to prevent people from compromising drivers. If you think you are installing Tryecrypt, but the digital signature is instead form "Joe's L33t Hax0r Shack", you should probably think about stopping the install and talking to the TC foundation since they have probably been hacked.

Re:As though any processor

[InvalidError]

This will not be an issue for desktops as most new desktops were 64-bit starting in 2004, with the last 32-bit desktop Semprons and Northwood Celerons bowing out around the end of 2005. The only mass-market 64bits desktop chip available in 2004 was the Athlon64... 64bits Prescott/Pentium D did not become widely available until early 2005. Pentium D 5x0 Prescotts that shipped in 2004 do not have EM64T but all other models (5x1, 5x2, 6xx, 8xx, 9xx) that started shipping in 2005 do.

Your first sentence becomes true about half-way through 2005 where even 32bits Celerons started being phased out and Northwood supplies completely dried up.

I remember building my current PC after a few weeks of pondering whether I preferred NW's lower power, the upcoming Prescott's 64bits or dual-dice 8xx. In the end, I opted for the 3.0GHz NW while it was still available at Prescott-like prices in September. Since I upgraded that computer from 1GB RAM to 3GB RAM last year, I plan to tough it out until I can get 2x2GB of reasonably priced DDR3 at some point through 2008 - assuming I will not be more interested in getting a new laptop by then.

Comment by Bill Laing clarified

[alienfluid]

So it seems like it's the Server SKU that is going to be 64-bit only, and not the client OS:

UPDATE | Microsoft's Alex Heaton has clarified the comment of Bill Laing, on which this story is based. "Bill Laing, a General Manager in the Microsoft Windows Server Division, has been quoted as saying that Windows Server 2008 will be the last 32-bit operating system. Bill is a server guy and indeed Windows Server 2008 is the last 32-bit server operating system - all future operating systems for server hardware from Microsoft beyond Windows Server 2008 will be 64-bit," Heaton said.

"A few folks took Bill's comments on Windows Server and applied them to Windows Client deriving that Windows Vista would be the last 32-bit operating system. That is an incorrect extension. While Windows Vista includes both 32-bit and 64-bit and there is a growing community of drivers for 64-bit Windows Vista we have not decided when Windows Client will follow Windows Server and become 64-bit only."