Although Microsoft has refused to confirm the many reports that say so, it appears the company is working on a version of its Windows Server platform specifically tailored for the high performance computing market… That Microsoft would branch off from Windows 2003 Server to create an HPC Edition makes perfect sense for a number of reasons.
Windows HPC Edition in the Works
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Will we see a Chess playing super computer with a M$ Windows logo on its case?
I don’t think so. It won’t work.
I think MS gonna base on UNIX system, becouse if it will be based on windows it won’t work.
Why not keep using Solaris & Linux?? Both are free!
Microsoft is creating this system for a low-volume market that has better alternatives. I think although Microsoft will gain marketshare they will loose revenue.
What could be better than the BSOD than several dozen BSOD’s at the same time?
Seriuosly though, I think they’re way out classed in this arena and serving in general. They are good at some things, but they will have to make very very unix like enhancements to the OS, specifically the kernel if it can hope to succeed.
I see only ONE main reason :$$$$$$$$$$
they will likely do what apple did – and put an interface over a BSD bases system. the winows NT kernel and inteernals (eg filesystems, scheduling) are not suited to HPC.
A few weeks ago, I downloaded a large PDF about “migrating high-performance computing to windows,” according to which, the model they are trying to pursue tries to implement the distributed programs as web-services & clients rather than open standards like MPI. In fact, the said PDF drones on about different stages in the migration of programs (its aimed at decision-makers, not researchers), in which Services for UNIX has been suggested as one of the migration paths.
I am very skeptical about this since the number of MPI programmers who will be excited at trying out anything of this sort is not likely to be large, while most programmers who implement things like web-services using MS platforms are typically not scientifically oriented to begin with … which begs the question, “who’s going to use this, if at all?” Nonetheless, it will be interesting to see which direction this goes.
Before making jokes that MS can’t possibly make this High Performance Computing version of Windows, I would wait till it’s out and seriously tested.
It’s like judging a film you’ve never seen
Windows makes people ignorant about computation, and if you don’t believe me have a look of joe user. People working in HPC are in general clever and well educated and they come up which thier own FREE solutions and aren’t expecting to pay M$ for a GUI to the open source utilities which M$ will make closed source.
>That Microsoft would branch off from Windows 2003 Server to
>create an HPC Edition makes perfect sense for a number of
>reasons.
and those are?
and those are?
To generate internet traffic of course, and making advertising on websites work.
Because all geeks against MS will rant and rant, and one more rant!
Your maybe right but since MS histroy in computing its more like judging the comfort of a Volkswagen you never drove. It might be a good car but its no where near a Volvo or Mercedes.
I think (as the article points out) MS is trying to make the advatage of their Unix services and .NET services on the big irons. Before you know it Windows and Linux are batteling on big irons and Sun & SGI are dead. Pitty but this might happen.
Linux however is a hard one to beat is POSIX, its Free/Open, stable, virus/trojan free, Unix, and well spread and loved by many current HPC sysops.
>Although Microsoft has refused to confirm the many reports
>that say so, it appears the company is working on a version
>of its Windows Server platform specifically tailored for the
>high performance computing market…
Where is that stated? it would be a logic step for MS to spread their .NET services etc. but is nowhere stated or written down. Its pure speculation like IBM is building an Linux distrobution toghter with Novell(suse), Redhat, Orale, CA and SGI.
Hmm, first, lots of ignorant crap on this thread.
But I’ll be ignorant myself and ask a question:
is this going to be clustering e.g. grids, progress migration etc, or is this going to be for more closely coupled big iron?
All I have to say, is I can’t wait. M$ Rules. They will dominate this market just like they do all of the others. I have used Linux for 4 years now. But it is not as fast as my WinXP Machine.
So either MS is doing this only for marketing purposes or who knows why.
A cluster implies lots and lots of CPUs. So what educational or governmental institution wants to pay $$$ per node when they can pay NOTHING per node for the OS. And the free node per node versions make up more than half the top 500 HPC clusters in the world.
MS must be wanting to give this all away for free then.
If they do release an HPC version of Server 2003 and if I got an inquiry about it I would reccomend it.
I think maybe you’re all seeing this from the wrong angle.
This HPC version of Windows could be the first step to bringing cluster computing to the masses. Think about it, Microsoft has a huge presence on the desktop, if they could build a set of tools that allow developers to easily leverage the untapped processing power of the office computers imagine what the developers could achieve.
The mind boggles with all the possibilities… (at least it would if it wasn’t a Friday afternoon!)
Do not confuse HPClustering with HPComputing. MS seems to be working on HPClustering and not HPComputing.
So rather than implementing and selling (HPClustering) something that is already available for free (MPI+freeOS), MS should concentrate more on HPComputing by increasing/optimizing/tweaking the HT, SMP, and DC (http://www.xbitlabs.com/news/cpu/display/20040610151158.html, note disclaimer) functionality in current MS “workstation” products.
I see only ONE main reason :$$$$$$$$$$
Agreed, god WinXP Home is still $100 at stores at a nearby town. When you can get SuSE Pro for like $20 less MS can kiss my ___
First M$ will give away the HPC software tools
just like they did with IE to fight against Netscape.
Then they will embrace and extend the PVM and/or
MPI software. Then add .Net on top.
MacOsX doesnt use the BSD kernel it uses the mach Kernel, so if MS were to use BSD as a base they’d also use there NT Kernel
You should know by now that you should not question authority. (Democracy is a myth)
authority: slashdot model – majority of mods.
OSnews model – majority of 1/owner.
If you don’t like the model… you know the rest.
MacOsX doesnt use the BSD kernel it uses the mach Kernel
Wrong again!
http://developer.apple.com/documentation/Darwin/Conceptual/KernelPr…
Both BSD and Mach are in the Mac OS X/Darwin kernel. Quit talking about things that you know nothing about, and stop spreading misinformation.
http://www.kernelthread.com/mac/osx/arch.html
“If they do release an HPC version of Server 2003 and if I got an inquiry about it I would reccomend it.”
Me too…. It has the smell of a fat payday written all over it. Might require massive amounts of aspirin to make it run, but what the hell…..
Does that mean we won’t see you post here again?
” don’t think so. It won’t work.
I think MS gonna base on UNIX system, becouse if it will be based on windows it won’t work.”
This is a completely untrue statement.
This seems to be the only way for Windows to deal with the flexibility of Linux. Rather than having kernel ports and distributions with different goals, the Windows world gets 9x, NT, CE, tablet, XP media center, XP embedded, XP 64 bit, X-BOX, X-BOX2, and now HPC.
While it would be cool if Microsoft did the BSD thing like Apple, I don’t think it would happen. I wonder what Theo would say if Microsoft based the hypothetical WinBSD on OpenBSD, like SFU. Actually, that sounds about right for Microsoft. They had former DEC employees make a VMS clone for the NT kernel, right? VMS is one of the most stable OSs ever, but now look at NT. I’m sure they’d make OpenBSD insecure somehow. Anyway, the problem is Microsoft can’t use UNIX because UNIX is the enemy. Also, since I’m sure everyone will expect Windows compatibility, they’d have to port win32 to UNIX, but they may already have done that. I often suspect Microsoft has minor side-projects, Plan Bs. If X-BOX2 is NT5 on PPC, that’ll be one side-project confirmed.
Windows NT was buggered up for one reason, and one reason alone, unlike OpenVMS, Windows NT was written in C/C++ where as with OpenVMS MACRO/BLISS and a small amount of assembly was used thus made it very secure. I’m sure that if any operating system was written in MACRO/BLISS, it would as secure.
With that being said, 1/2 the stability and security can be put down to the hardware and the fact that the hardware and software teams worked together, each adding features as each other required it.
You are confusing the Mach microkernel and the BSD-based userland.
Read the links he provided, BSD is part of the kernel as well as in userland.
“Read the links he provided, BSD is part of the kernel as well as in userland”
No, the BSD userland could be *considered* part of the kernel – but only in the same manner as a filesystem driver.
No, the BSD userland could be *considered* part of the kernel – but only in the same manner as a filesystem driver.
Fine. Stay in fantasyland. I really don’t care.
“Fine. Stay in fantasyland. I really don’t care.”
“Fantasy land”? Hey, if you don’t understand what a microkernel is, don’t bleme me. Darwin does NOT have a BSD kernel. No amount of screaming from you will change that.
Hey, if you don’t understand what a microkernel is, don’t bleme me. Darwin does NOT have a BSD kernel. No amount of screaming from you will change that.
Idiot. I’ve posted links to Apple’s web site numerous times that state clearly that XNU is a monolithic kernel based on Mach and BSD. If you’re too fucking stupid to read them and understand them before continuing to waste our time, then yes, you are living in a fantasy land.
Stop arguing this point. Kingston is a tactless prick, but in this case, what he’s saying is true.
In case you’ve changed you mind, and decide to actually learn something, it’s in the very first paragraph.
http://www.apple.com/lae/macosx/technologies/darwin.html
“Idiot. I’ve posted links to Apple’s web site numerous times that state clearly that XNU is a monolithic kernel based on Mach and BSD. If you’re too fucking stupid to read them and understand them before continuing to waste our time, then yes, you are living in a fantasy land.”
It’s quite OBVIOUS you have no clue what a “microkernel”
is nor are you capable of understanding my posts. When Apple marketing refers to the “kernel” they are including the Mach microkernel, filesystem drivers, graphics, and the BSD-based userland. A microkernel based system isn’t a “monolithic” kernel no matter what Apple marketing calls it. Darwin no more has a real “monolithic” kernel than Windows NT does.
Now it’s clear that you are either a complete retard, a troll, or both. Have a nice life sped.
If’you’d pay attention to the world outside your feeble little mind, you’d know that I damned well do know what a microkernel is, and that I favor them over monolithic ones.
If you’d read the developer documentation on Apple’s site, or looked at the Darwin source code, you’d see that although XNU (the Darwin kernel) is based on Mach 3 (which was a microkernel), that it also has BSD kernel code built right into it, meaning that the resulting kernel (XNU) is no longer a microkernel, despite the fact that it has a message passing infrastructure.
Alot of the BSD kernel bits are compiled into the kernel image, meaning it’s all running in one single address space, which (if your feeble memory serves you), is not how microkernels are supposed to be done. They are supposed to implement the barest functionality in the kernel, and use a defined API (generally using message passing) to communicate with other components in userspace (each component in their own protected memory space.).
It’s not Apple marketing, saying that it’s a monolithic kernel, it’s a fact. It’s a choice the developers made, in order to avoid the performance penalty involved in the extra context switching that would take place were the entire BSD subsytsem to have been in userspace. They gave up some stability for performace (if you thought Mac OS X 10.0 was slow, just imagine how much slower it’d be were it to really have been implemented as a microkernel based OS).
But by all means, don’t believe me. Don’t read the extensive developer documentation they have publically available. And don’t let any more garbage leak out of your brain and onto your keyboard untill you get a gawdamned clue, read the docs, and know that I am right.
Not that the fact that I’m right matters. But hey, it could be worse. I could be a cluless git like you.
There are two points of contention here:
(1) Is “XNU” a monolithic kernel?
A monolithic kernel is a single image (executable piece of software) that contains drivers, resource allocation, security and all the other central functions. An example of a *pure* monolithic kernel is that of 4.4BSD. Linux is a monolithic kernel with the capability to load modules into user space for additional functionality.
A microkernel is a design that implements a minimum of functionality in the kernel itself and everything else running above kernel space.
In XNU, the Mach microkernel loads modules into kernel space, not user space. It is NOT a monolithic kernel. It’s actually more a hybrid.
Apple agrees with me:
“This modular structure results in a more robust and extensible system than a monolithic kernel would allow, without the performance penalty of a pure microkernel.”
http://developer.apple.com/documentation/Darwin/Conceptual/KernelPr…
Ergo, it is NOT a monolithic kernel. Those modules are running in kernel space, but they are NOT compiled into a single image any more than a linux driver module is compiled into the kernel.
(2) Is XNU a BSD kernel.
XNU is not a monolithic kernel, but a hybrid microkernel design and as such has little to nothing in common with the 4.4BSD kernel design. It does have code from BSD, but so does Windows NT 4.0 (TCP/IP stack) – of course, XNU borrows more from BSD than NT4 did. Those subsystems are only defined as being “in the kernel” as much as any other Ring 0 code would be.
As to your tone, kid, you really ought to look into anger management treatment before you burst something. Does your Mom know you’re on the computer?
Now, go play with your toys and leave the adults alone – until said point when you can communicate as an adult.
A microkernel is a design that implements a minimum of functionality in the kernel itself and everything else running above kernel space.
Uhm, yeah. I think I said something to that effect. And BSD is essential why?
Ergo, it is NOT a monolithic kernel. Those modules are running in kernel space, but they are NOT compiled into a single image any more than a linux driver module is compiled into the kernel
In a microkernel, they wouldn’t be running in kernelspace, they’d be entirely in userspace. Modularity has precious little to do with wheather or not a kernel is “micro” or monolithic.
From the link provided:
However, in Mac OS X, Mach is linked with other kernel components into a single kernel address space. This is primarily for performance; it is much faster to make a direct call between linked components than it is to send messages or do remote procedure calls (RPC) between separate tasks.
Sounds monolithic to me.
XNU is not a monolithic kernel, but a hybrid microkernel design and as such has little to nothing in common with the 4.4BSD kernel design. It does have code from BSD
No arguments. It’s probably the only correct thing you’ve said this entire time.
Now, go play with your toys and leave the adults alone – until said point when you can communicate as an adult.
Adults try to be correct, and attempt to avoid spreading misinformation. I tried correcting you politely, and you continue to fight back with nothing but incorrect information. When I say that XNU is not a microkernel, and provide evidence, I am speaking truth. When you continue to call my facts into question, you are providing the evidence I require to justify calling you what I suspect you are, an idiot.
Anger management isn’t an issue of mine. For whatever twisted reason, arguing with idiots on OSnews in my spare time is.
At any rate, keep right on living in that (ahem) adult fantasy land of yours.
This was not a polite introduction:
>”Fine. Stay in fantasyland. I really don’t care. ”
” I tried correcting you politely, and you continue to fight back with nothing but incorrect information”
When were you polite again? Perhaps you aren’t straight on who your arguing with from one moment to the next?
“In a microkernel, they wouldn’t be running in kernelspace, they’d be entirely in userspace. Modularity has precious little to do with wheather or not a kernel is “micro” or monolithic. ”
In a *monolithic* kernel you would have a SINGLE image loaded into its own address space, not modules loaded into *kernel* space.
As *APPLE* said – it isn’t a monolithic kernel. BSD is a monolithic kernel.
“Sounds monolithic to me.”
Not surprising, but it isn’t – and Apple agrees with me.
A monolithic kernel loads a SINGLE image into ring 0, not loadable modules.
“When I say that XNU is not a microkernel, and provide evidence, I am speaking truth. ” When you say it’s a monolithic kernel, you aren’t.
“When you continue to call my facts into question, you are providing the evidence I require to justify calling you what I suspect you are, an idiot.”
See, this kind of thing makes your “facts” so easily dismissed. That and starting off with the whole “fantasy land” thing.
XNU is NOT a monolithic kernel. Live with it.
“Anger management isn’t an issue of mine.”
Oh, yes it is. I suppose I asked for it, arguing with children and all. Get back to us after you grow up. Your manifestos are a bit hard to take seriously, especially when I can produce Apple developer documentation that disagrees with your links from the marketing pages.
Oh, BTW – this was a true statement:
“No, the BSD userland could be *considered* part of the kernel – but only in the same manner as a filesystem driver.”
Unless you don’t agree with Apple’s developer documentation on the XNU kernel.
Have fun screaming at people on OSNEWS. I really don’t care what you do or think.
Perhaps you aren’t straight on who your arguing with from one moment to the next?
Clearly, I’m arguing with “JCS,” be it a single person with multiple proxies, or a small number of incompetant trolls using the same name and arguments.
Either way, it’s incorrect to claim a kernel to not be monolithic just because it’s modular. If the modulues get loaded into kernel space at runtime, you still have a giant chunk of code running in ring 0, therefore monolithic.
Just shut up, the both of you. Kingston’s right in that it’s monolithic. JCS is right that Kingston’s more than a tad touchy.
You’d both be better served by getting a new hobby.
I don’t intend to reply to that joker further, but the thing is – he *isn’t* right.
“This modular structure results in a more robust and extensible system than a monolithic kernel would allow, without the performance penalty of a pure microkernel.”
http://developer.apple.com/documentation/Darwin/Conceptual/KernelPr…
“XNU, the Mac OS X kernel, is also a modified microkernel, due to the inclusion of BSD kernel code in the Mach based kernel.”
…
“Some people confuse the term “Hybrid kernel” with monolithic kernels that can load modules after boot. This is not correct. “Hybrid” implies that the kernel in question shares architectural concepts or mechanisms with both monolithic and microkernel designs – specifically message passing and migration of “non-essential” code into userspace while retaining some “non-essential” code in the kernel proper for performance reasons.”
http://en.wikipedia.org/wiki/Monolithic_kernel#Hybrid_kernels_.28mo…
http://www.bellevuelinux.org/kernel.html
“Hybrid kernels are similar to microkernels, except that they include additional code in kernel space so that such code can run more swiftly than it would were it in user space. These kernels represent a compromise that was implemented by some developers before it was demonstrated that pure microkernels can provide high performance. Hybrid kernels should not be confused with monolithic kernels that can load modules after booting (such as Linux).”
…
“Most modern operating systems use hybrid kernels, including Microsoft Windows NT, 2000 and XP. Mac OS X also uses a modified microkernel, as it includes BSD kernel code in its Mach-based kernel. DragonFly BSD, a recent fork (i.e., variant) of FreeBSD, is the first non-Mach based BSD operating system to employ a hybrid kernel architecture.”
Mr. Kingston is confused. XNU is a hybrid kernel, not a monolithic kernel. Loading modules into kernel space does not a monolithic kernel make. XNU is also not the BSD kernel, which *was* a monolithic kernel, but merely contains BSD code.
This isn’t my hobby, I just do not appreciate being attacked by people like Kingston.
You’ve got a point there, but so does Kingston. You’re right that it shares architectural concepts or mechanisms with both monolithic and microkernel designs but he is right that XNU is monolithic, because all of the running code is in one address space (one, mono, you get the idea 🙂
I guess that this particular issue is going to remain a point of contention here, so I’ll say no more about it.
I still think that it’s a silly thing to spend so much time arguing over (I say while participating in said discussion :-D)
“mechanisms with both monolithic and microkernel designs but he is right that XNU is monolithic, because all of the running code is in one address space (one, mono, you get the idea 🙂 ”
That isn’t a “monolithic kernel”. A “monolithic kernel”
is a single executable file running in kernel space.
To quote again:
“Some people confuse the term “Hybrid kernel” with monolithic kernels that can load modules after boot. This is not correct.”
That’s hardly a trustworthy source. I’d never trust something written by an anonymous public to be accurate or at all otherwise correct.
Why are you using my name, Kingston?
You seem to forget the other links I’ve provided. Oh, wait!
They don’t agree with you, therefore they must all be wrong!
No matter. Your opinion means nothing.
Bye.
which you? the real you you? i’m having such a hard time keeping you all straight! no one’s opinion matters boy.
“which you? the real you you? i’m having such a hard time keeping you all straight! no one’s opinion matters boy.”
Yes, the real me. There is only one me. I think you’ve got me confused with Youlle and the rest. You’ve posted as “Anonymous”, “Kingston”, and “JCS”. YOU are posting with multiple names, not me.
Try arguing your case again when you’ve grown up. I have no more time for 12 year olds.
You seem to forget the other links I’ve provided. Oh, wait! They don’t agree with you, therefore they must all be wrong!
For crying out loud! You could use the exact same argument about the links that I provided first! How is this relevant? Of course I disagree, just like you did with my links.
Think.
I would like to further add that my sources (various Apple sites) are far more credible than the wikipedia link etc. that you gave.
“I would like to further add that my sources (various Apple sites) are far more credible than the wikipedia link etc. that you gave.”
That “etc” includes developer.apple.com. Of course, it doesn’t matter. Any site or link that backs you up, you consider fact and any site that disagrees with you is wrong.
All you could do was hurl insults from the start. Like I said: 12 years old.
All you could do was hurl insults from the start
Not true. I started by providing accurate information with links to back it up. You started off saying that “no amount of screaming would change the fact that you’re wrong,” and then proceded to continue spouting misinformation and tried to back it up with less than solid evidence. Then, when I realized that you were merely trolling, I started to call you what you really are.
Sod off. You bore me sped.
“Not true. I started by providing accurate information with links to back it up”
No, you responded with inaccurate information. I have, on a number of occasions here, responded with links that back ME up. You ignore all but the Wikipedia link and mention that only to dismiss it.
“You started off saying that “no amount of screaming would change the fact that you’re wrong,”
Your memory is faulty. YOU started the insult war, Kingston.
” less than solid evidence”
Hardly. Try the definition of the term.
“Then, when I realized that you were merely trolling, I started to call you what you really are. ”
I’m not trolling. That’s another word you don’t know the meaning of.
“Sod off. You bore me sped.”
Funny. I’ve told you the same thing, yet you keep coming back. Please, do go away this time. Permanently, if possible. At least until you get past being 12.
Funny
That you are.