Microsoft Corp. today announced the release to manufacturing (RTM) of Microsoft Windows XP 64-Bit Edition Version 2003. Windows XP 64-Bit Edition Version 2003 is optimized to enable customers to take advantage of the performance enhancements in the Intel Itanium 2 processor.
Looks like only Linux and OS X users get 64 bit desktops this Christmas season.
Did you forget to mention AMD’s x86-64 processors?
IIRC AMD 64 and Itanium 64 are distinct architectures, code for one can’t run on the other.
I’m just curious. Does anyone really care about this news? I don’t mean to slight Eugenia or OSnews in any way. I am just asking in a survey sort of way.
Yes. This is an Operating System news site. Last time I checked, Windoze was/is an OS. So, this is news worthy.
People who don’t want M$ related news should go to http://www.linuxtoday.com if they want just Linux news. Then they won’t be distressed by M$ information.
Having just done with two HW shows, I saw literally 2 dozen vendors of high end computing HW, some on Itaniums, some on Opterons & Athlon MPs, some even on Transmeta, PPC etc. So much IRON in so many forms but all massively dense packaged. Some vendors even supporting multiple competing options, whatever sells. I almost certainly walked past WinXP-64 without realizing it. BioTech show.
The interesting thing was seeing hundreds of densely packed cpus, esp AthlonMP & Itaniums in cabinets/racks where the heat & noise was taken care of.
My conclusion is that the 64 bit computing is looking less & less likely for us ordinary folks. Even if I could get an Itanium or Opteron board, the OS/power requirements are going to be severe, only the Linux RH/Suse option looks possible. I see little prospect of staying with MS at 64bits even if I wanted to.
If you go with Opteron, the Intel compilers won’t help at the 64 bit level so AMD will have to get its act together there instead of freeloading off Intel. Its going to take some time!
And does anyone care.
YES I do, I have potential computing problems to work on that will definately benefit from 64bit both address size and integer size, it’s just a matter of becoming affordable.
What are you talking about? MS has not _officially_ announced to support Hammer. Yes, that’s true. But have Apple or IBM announced that Macs will get PP970? No!
Why do you think that there will be 64 bit Macs, but not 64 bit Windows PCs?
On the Apple side there are only rumors. OTOH, Microsoft has this page: http://www.microsoft.com/ddk/debugging/installAMDbeta.asp
Is there any real-world benefit to a 64-bit processor and OS? Aside from making the code larger, what kinds of real-world benefits are there? Is this equivalent to the change from 16-bit to 32-bit? Will we see faster code, more clever behaviors and anything new that can’t be done in 32-bit?
Jace: The main real world benefit of a 64-bit processor is accessing large amounts of memory efficiently. By large amounts I’m talking greater than 2-4 gb. Who needs this kind of memory? In the sciences, lots’ of people.
I’ve personally worked in biotech on a machine with 178gb of memory. If I compiled a program as 32-bit which I mistakenly did once, I could only access 2gb of memory. A simple recompile as 64-bit and the problem was solved.
Is this usefull for desktop machines? Maybe in sciences. It won’t run your internet or games any faster though. It will probably be a long time before regular programs start requiring more than 2 gb of ram.
Well in development tools for games they are already beginning to use 64 bit computer (look at comments from the Unreal people) and also 64 bits will make for higher accuracy as they already use tricks of splitting 64 bit numbers into seperate 32 bit registers. 64 bits would also make room for more tricks and optimizations I believe.
Precisely!
AKH, you should write me to discuss your Bio work.
Not just Bio, I could list a dozen supercomp problems, mine are in ASIC/FPGA design, not as bad as Bio, but being HW guys we know atleast how to use the HW we create, well maybe not. As I saw in the BioTech show, most Bio Phds are clueless about the HW their SW runs on, so they get it worst, just buy more & more server/blades etc.
One of the really big issues for all of us is the relative slowness of HDs v CPU. As per Moores law, cpus continue to double performance every 18Mons or so.
All the HDs can muster is a 10x speed up say every 10yrs, after all these are moving parts. HDs do double in size every 12-20months, but not their speed. So HDs are part of the problem, not the solution. It becomes possible to store TBytes of data (Human Genenome etc) on disk arrays but ripping through it for string searches is incredibly inefficient and obviously getting much worse as cpus leave HDs in the dust.
The only solution is to whop it all into DRAM which might only be $100/Gig or so, so 178G might only cost the price of a server or car. At the Bio show, besides cpu clusters, the other big boxes were all TByte SANs with 2GByte DRAM caches. The real problems are in moving the data around from where its stored to where its processed.
Without >>32 address, the industries are thoroughly crippled by 2G space. The 16/32 crossover for general users a decade ago is entirely similar to the 32/64 crossover for the sciences.
I won’t be surprised if in 10yrs we are discussing 64/128 crossover for super sciences, after all a 128 bit cpu is only a 2x the cost of a 64 cpu design but don’t be craving for one. Actually your nVidia card by any reasonable definition is a super graphics computer with 64-256 data paths a plenty.
In the world of FPGA computing, the data paths are essentially as wide as the problem needs for it to be, ie 64 engines that might be 16 bits is equiv sort of to 1024 cpu design.
For most users, they should be concerned about unbloating their current systems if they want more speed.
JJ
Microsoft is so rad
Another reason for the switch to 64bit (well, actually just to x86-64 ie Opteron/Hammer) is that x86-64 has more General Purpose registers exposed to compilers. This is going to GREATLY enhance the speed of programs.
Go looking for the comments that the Sierra guys made a month or so ago about how they recompiled the Halflife server for x86-64 and got an almost 30% speed increase. They didn’t change the code at all – just recompiled it so it used the new registers (while not actually using any 64bit instructions).
x86 has had this problem for a long time of not having enought GP registers exposed – and x86-64 addresses this as well. So once all the applications get recompiled you will see large speed differences by going with an Opteron/Hammer chip in your comp.
I hope MS gets the x86-64 version of Windows out soon. I am going to be running 64bit Linux as well, but I have to use windows a lot and it would be sweet to have all 64bit OSs on my machine….
Derek
Check this out: http://www.oscast.com/stories/storyReader$221
That OScast guy doesn’t know what he’s talking about RE: Microsoft.
Along with the DDK info KAMiKAZOW posted, there’s also documentation on MSDN about x86-64, and here’s an interview where Brian Valentine talks about Microsoft’s ententions of supporting AMD64.
————–
PT: When does 64-bit technology become viable, both on the server and on the desktop?
BV: We have native Intel Itanium support in Windows Server 2003, and that will help 64-bit computing go mainstream. Moving forward, the 64-bit architectures that we’ve committed to support include Itanium and AMD’s AMD-64. These architectures are native parts of the build process now, and every day we build them as well as the 32-bit products.
With Longhorn [the next Windows release], our goal is that everything that’s native to Windows will be 64-bit native across those platforms.
(more info at the link)http://www.winnetmag.com/WindowsServer2003/Index.cfm?ArticleID=3827…
“Yes. This is an Operating System news site. Last time I checked, Windoze was/is an OS. So, this is news worthy.”
I did not mean that it was not newsworthy, I am asking if anyone personally cares about Itanium Windows. If the news was for Hammer Windows, I would care, albeit ever so slightly.
To further explain, does anyone reading this have a serious interest in Itanium Windows considering the severe lack of application compatibility? I note that almost all posters seem to have a greater interest in either;
1) AMD 64 bit Windows
2) Linux on Itanium
with the exception of the enthusiastic “yay” “Microsoft is so rad.”.
>To further explain, does anyone reading this have a >serious interest in Itanium Windows considering the severe >lack of application compatibility?
Well, given the relative “new-ness” of the platforms, it’s going to be a while before commercial stuff is ported. I know there are companies working on ports.
Mostly, one would expect to see a lot of custom/in-house software ported. These are the early adopters that have real computing needs for 64-bit stuff. Maybe financials/bio-tech/science, etc.
> I note that almost all posters seem to have a greater >interest in either;
> 1) AMD 64 bit Windows
> 2) Linux on Itanium
What apps have been ported to Linux Itanium? It would be cool to see some of the graphics/rendering stuff ported and to see how they benefit.
“What apps have been ported to Linux Itanium?”
Well, I can tell you about the two apps we use the most in my workplace: Abaqus and Fluent (Finite Element Analysis and Fluid Dynamics Modeling respectively) – they have both been ported to Linux on Itanium.
Not too mention that you can just recompile most open-source apps and they just work…
Derek
In case anyone reads this far back, I appreciate the responses to my “how is this useful” question. It’s been a log time since I messed with development and even longer since I looked at issues of CPU registers, memory addressing and such. Thanks for the info!