Fog Creek Software
Discussion Board




Longhorn Requirements:Has Microsoft Shot itself?

<<
Analysts expect minimum requirements to start with a 4 GHz CPU and 1 GB of RAM, built-in wireless, and a DirectX 9-capable 3D accelerator with 128 MB of RAM or more. But the recommended system could be far more powerful>>

Noww folks tell me. If Bill Gates had spent 1$Billion on advertising Linux, it could not have gotten worse. If there are 10 million (An example) computers running Visual Basic, swear anywhere from 1 million-5 million will be outraged with what Microsoft is demanding. It means their systems are obsolete, unless they cough up  hundreds of thousands of dollars JUST TO GET THE OLD APPLICATIONS RUNNIING on the new OS. Add to that the fact that many of them would have to be rewritten, it sure will make IT managers take another look at Linux.

There are millions  of critical applications running VB6 . Stuff like train reservations etc. No one will be amused at M$ dictating terms to them.

Longhorn is Microsoft's gift to Linux camp. M$ seems to forget that business drives technology and technology DOES NOT/CANNOT dictate to business.

I am not predicting the end of Microsoft. But a new age in which Linux is taken very very seriously. There are millions (hundreds of milllions) places where this kind of hardware requirements are absurd.  Linux will be attractive to many of them, thanks to Microsoft.

We have an ERP system in my company. And boy, will my boss like it if i ask 300 machines to be upgraded and dozens of systems to be Retested just because Microsoft wants it?

Full article:
http://www.winnetmag.com/windowspaulthurrott/Article/ArticleID/42522/windowspaulthurrott_42522.html

Karthik
Monday, May 03, 2004

Another fact
According to Google Zeitgist, only 40- 45% have upgraded to Windows XP despite all the brainwashing. Guess how many will upgrade to Longhorn?

Karthik
Monday, May 03, 2004

No


Monday, May 03, 2004

An awesome display of intelligence. I am overwhelmed.

Karthik
Monday, May 03, 2004

that kind of hardware will be quite standard in 2006-2007.

Upgrade to longhorn will be a process, the first set would be the people who buy new hardware, and then there would probably the ones running 95/98/2000, and then the folks running XP.

Also the cost of hardware would be much cheaper than what it is now, so I think the hardware requirements are modest!

Prakash S
Monday, May 03, 2004

That hardware spec ain't all that.

I bought a new development PC a while back that far exceeded those specs and cost a few thousand dollars. By the time Longhorn actually ships it'll be a whole lot cheaper.

As for the other comments, only time will tell, but I suspect the death/demise of Microsoft/Windows/etc has been, and will continue to be, greatly exaggerated.

Steve Jones (UK)
Monday, May 03, 2004

With every release of Windows since 1.0 I have asked myself the same question. Each time I have answered "Yes", and each time I have been wrong.

Now, with Longhorn, I'm making an assumption that the user upgrade cycle will continue, and that the requirements will be realistic.

*grin* ... of course, that could be a continuation of my losing streak :)

HeWhoMustBeConfused
Monday, May 03, 2004

Unfortunately for the nay sayers. Microsoft has just about never hurt itself bad when it bets heavily on the idea of future hardware being readily available. (In fact, there's a fleet of nuclear submarines paid for by Bill G swimming outside of South Korea and Taiwan right now to fend off the communist missles from destroying the chip foundries. Lol.. okay bad joke)

Remember expanded memory? Extended memory? Seem so pointless now doesn't it?

Remember how people worried that thunking of 16 bit instructions into 32bit was too expensive? By the time people even started installing OS/2 95 (and the clones you needed to run them) was around to make it a non-issue.

How about resources being used up by 95? How about NT being too heavy weight in 3.51 for the Pentium 1s? Would you even hesitate to put Windows 5.0+ now on any of today's computers? No right?

So IT IS a scary bet, but in time we won't even think twice to have inefficient VMs protect processes.

Li-fan Chen
Monday, May 03, 2004

Even with something like stackless python, IronPort sends out 600,000 emails in one hour. Yeah it's running FreeBSD, very cool and efficient, but stackless Python has both inefficiencies and efficiencies that cancels out. The reason it is fast is mostly because of architecture (thread-less multiplexing, fibers or IO completion ports for the NT coders) and the fact that hardware is cheap. Behind IronPort is a $3,000 Dell rack, and these things are getting faster by the quarter for the same price.

Li-fan Chen
Monday, May 03, 2004

The important question is.. will it be worth it? What is using up those resources. Why the graphics card requirement.. who cares about eyecandy?
Also, dont overestimate the willingness to upgrade. I know of a few companies that bought small form factor 1 ghz boxes instead of going for the latest and greatest for their last upgrade.

Eric Debois
Monday, May 03, 2004

Eric, then in those cases you'll have to trust they know what they are doing and will leverage the platform but keep it as thin as possible. A library will only take up memory if you use it, so if you want to move to Longhorn so early in the game and still use 1Gig machines with 256Megs of ram, you should be careful to write small programs, otherwise thousands of users will revolt and your migration will back-fire on you.

Li-fan Chen
Monday, May 03, 2004

Microsoft has ALWAYS bet on hardware improvements.

Remember how slow Windows 1.0, 2.0, 3.x were, on average PCs of their time?

slim
Monday, May 03, 2004

Microsoft has always bet on hardware, and Intel (now AMD) has thanked them for it.  However, the tax laws in the US are not forcing most companies to depreciate hardware over 36 to 48 month cycles.  Sure we developers get blazing machines, but that machine they just took off your desk, probably went to an end user.

They are going to need something more than it's the next OS, to convince companies that need to replace 10,000+ machines that it is worth the investment.  As someone pointed out, they have not even done that with XP yet.  Of the percentage quoted, I would even bet most of it came from new machines, where a new OS is always an easier sell.

MSHack
Monday, May 03, 2004

I have to agree with Prakash. Most people buying Longhorn will be getting it already installed with their new computers, and those specs will be the mainstream desktop systems sold by Dell, etc in a couple of years.  Older systems that don't meet the spec will be starting to fail whether they're fully depreciated or not.

Tom H
Monday, May 03, 2004

Microsoft has issued what their goals are for both "minimum" and "recommended". These so-called "analysts" (which is usually a code-word for "I can't cut it at a real job, so I'll just sit around and make predictions) have taken Microsofts recommended specs, jacked them up even further, and then called them "minimum".

They are, to put it bluntly, a bunch of idiots. The actual minimum and recommended specs will not be truly known until we're close to release. If these analysts have been using the VERY incomplete and VERY unoptimized alpha version of Longhorn, and making future guesses based on that, then they're VERY incompetent. Simple as that.

Brad Wilson (dotnetguy.techieswithcats.com)
Monday, May 03, 2004

Am I the only one that thinks the speculation in this article is way off.

I run a W2K DC on a P75 with 64MB of RAM and it works very well. I am running W2K3 Server on a P233 with 128MB for some non-critical stuff and again it works very well for our small office. While Longhorn won't be released until at least 2007 so hardware standards will certainly have moved forward considerably. And some of the new features will certainly need something of a jump in power. I would be completely amazed if the minimum requirements changed so dramatically.

But if the article is accurate then I'd have to agree with the OP, the day after those minimum requirements are announced we'll be working on Linux versions of all our products.

Eudoxus
Monday, May 03, 2004

Obviously, I'm not the only one who thinks so.

Eudoxus
Monday, May 03, 2004

"Analysts expect minimum requirements to start with "

Now, lessee...

"a 4 GHz CPU"

I bought a PC in January, with a 3.2 GHz. I hear the 3.4 is now at the price of the 3.2 I bought in January. So, I guess in 2006-07 we should have no problem in meeting this.

Next.

"and 1 GB of RAM"

Hum.. This is what I put in my PC in 01/2004.

Next.

"built-in wireless"

And, may I ask, why is this required? Will Longhorn choke if someone attempts to install it in a machine without wireless support?

Next.

"and a DirectX 9-capable 3D accelerator with 128 MB of RAM or more"

I don't know about the DirectX 9-capable, but my 3D graphic card has 128 MB, and is, by no means, a state of the art. It was rather cheap, actually. we may even have graphic cards more powerful than this embedded on the motherboard in 2006-07.

So, what's the big deal?

Long-in-the-coming-horn won't grace us with his presence for another couple of years, at best. I really dislike MS FUD, but this poor attempt anti-MS FUD is just as idiotic.

At least, try to make it slightly believeable.

Paulo Caetano
Monday, May 03, 2004

> And, may I ask, why is this required? Will Longhorn choke if someone attempts to install it in a machine without wireless support?

So that it can phone home without ou being able to pull the wire out of the wall. You read it here first ;-)


Monday, May 03, 2004

Same ol' Same ol' "I don't see why you would ever need more than 640k" debate.

Get over it, guys . . . technology is NEVER done.

Steve
Monday, May 03, 2004

No, it is different this time.  First, people are upgrading a lot less these days.  The typical Windows box uses about 25% of the performance available to it, if that.  Even developers don't feel the need to upgrade any more.  Personally, I found 1GHz to be the sweet spot.  After that, I can only tell the difference for a few select applications, especially video compression.  But now video compression is being moved to hardware.

Second, notebooks and other portable computers are the way of the future.  4GHz processor?  Not in a portable device.  The power consumption and cooling problems with current high-end CPUs are already bad enough.  You have to slow down any notebook to get decent battery life.

Junkster
Monday, May 03, 2004

All of you/Most of you come from computer background. i come from a typical manufacturing background.
In fact Windows 95 will serve our needs very well ! Its not 600K problem. I do agree that after sometime M$ would hook most people onto it. But at the same time, Linux thin clients are going to be a rage, i think because of Longhorn system requirement

Dont think in terms of whether Avalon looks cool. Think like
1)Can the hard disk capacity of the computer fit my data?
2)Can the processor speed give me reports fast?

If 1) and 2) is a "Yes", M$ can show a system built by martians and still not many would fall for it.

Karthik
Monday, May 03, 2004

I don't see it as that big a deal.  Computers of almost those specs are available *now*.  The only thing current top end computers lack by that list is 4 GHz, but they really aren't that far off.  This for an OS that is scheduled to come out in 2-3 years. 

Windows XP was released in late 2001.  3 years prior to that, a 400 MHz machine was state of the art.  Consumer graphics cards were just starting to embrace 3D acceleration.  Systems had in the range of 64 MB of RAM.  What are the minimum listed specs for XP?  Minimum 233MHz processor and 64 MB RAM, 300 MHz and 128MB recommended.

Now considering that like XP, Longhorn will probably have a long phase-in period, and that initially it will mostly be on brand new computers which far exceed those specs, it really doesn't seem like a big deal.

MikeMcNertney
Monday, May 03, 2004

"4GHz processor?  Not in a portable device.  The power consumption and cooling problems with current high-end CPUs are already bad enough"

Those "expected minimum specs" are just projecting current technology forward.  Intel is planning for future chips to be slower but have better IPC, similar to Athlons vs P4s.  There will certainly be chips "equivalent to a 4GHz P4" which run at a much lower actual clockrate.

MikeMcNertney
Monday, May 03, 2004

4 GHz CPU and 1 GB of RAM, built-in wireless, and a DirectX 9-capable 3D accelerator with 128 MB of RAM or more

I would believe this if the OS was coded in gw-basic.

someone
Monday, May 03, 2004

Anyone know what Longhorn will do with all that computing power ?

The trends according to BillG (correct me if I am wrong) are seamless operation (all devices, programs and people have easy time working together) and security. Both of these sound more like standardization issues to me.

Then there is 3D UIs, voice recognition and other new age technologies that yet have to prove themselves in desktop environment, but at least they are require new hardware and look marketable.

And companies sure don't upgrade like they used to, mine (~3000 people non-IT) is just switching from Windows NT to XP and that is simply because MS will stop making security etc. updates to NT.

dmann
Monday, May 03, 2004

<i>There will certainly be chips "equivalent to a 4GHz P4" which run at a much lower actual clockrate.</i>

Correct, but we're still nowhere near anything that is comfortable to use in a notebook. You get lower power consumption with things like the Pentium-M or a Transmeta processor, and they get lower power consumption by cutting both clockrate AND transistor count.

I think the Longhorn processor requirement is almost certainly incorrect.  The primary change is having a flashy UI, but at the same time if a good 3D video card is required--one that does all transformations and clipping on the GPU--then all of this work is offloaded from the main processor.  In fact, it happens in PARALLEL with the main CPU.  In fact, this would result in Longhorn being faster than Windows XP on current systems.  But what is Microsoft doing that ups the minimum spec from 300 MHz (which is what it is for XP, admittedly too low), to over 10x higher?  I think this is just a rumor.

Junkster
Monday, May 03, 2004

Come to think of it, it must be wrong.

<Theory>
MS plans to use LH as a means of getting wide spread adoption of several new technologies. Adoption of .net, avalon, XAML and what ever else theyre shouting about is important to MS because they are the corner stones of a strategy for maintaining dominance.
If LH only gets distributed with new machines adoption will go slow. They are not likely to take that risk.
Thus I bet LH will run fine on a 1ghz/512 meg box.
(Memory might be more important than CPU speed though, and memory is really getting quite cheap.)
</Theory>

Eric Debois
Monday, May 03, 2004

Karthik:

I've worked in manufacturing, too. In 2006-7, most manufacturing plants will still be using Windows 95 or 98. So I don't see a problem :)

Rob VH
Monday, May 03, 2004

How come there was only one (1) post that mentioned the word "disk"?
I installed Win 2000 on a Pentium 100Mhz with 32 MB (PC of my grandpa, he just won't upgrade before he dies :-). It runs fine when using Word and stuff, but bigger apps are just too slow, because of... you guessed it - the disk. Same story with XP on some older machines.
I wonder how many gigs you'll need to install Longhorn. And how much of that it'll read when booting, or starting IE (or whatever app is fully integrated into the system). In my opinion, not CPU or RAM is important, but it's the disk that does most of the work in modern OSses. And guess what? Disk speed increases only very slowly over the years. That's the real point of concern. IMVHO anyway.

YoHeresDaRealAnalysT
Monday, May 03, 2004

As for whether the requirements are accurate or reasonable, I can't comment.  However, it is interesting to note how companies like IBM are reacting to some of these Microsoft strategies that often come off to users as alienating.

excerpt:
"By the end of 2004, Microsoft is expected to discontinue support for the Windows NT operating system and discontinue the availability of security patches, which will require up to two million customers to develop a migration strategy. As customers around the globe make the decision to migrate to a new platform, many will consider Linux, [.....] Linux is a top choice of customers and Business Partners who choose to migrate from a Microsoft operating system [.....] IBM will offer free Windows NT-to-Linux-migration classes to qualified IBM business partners, delivering education and training to its 90,000 Business Partners located around the world. The classes, many of which will be held in IBM Linux Centers, will provide a hands-on environment featuring IBM software and hardware for partners to learn how to plan and deliver a successful migration. "

(full article: http://www.in-sourced.com/article/articleview/1194/1/1/ )

Things like disowning previous OS versions that are still widely used in critical functions, and jacking up hardware requirements, are not exactly making Microsoft more popular.  IT managers will be forced to budgets.  If their people can get their work done on Linux boxes, and they can show millions in savings on the bottom line, any guesses as to what they will do?  The business world revolves around support.  If companies like RedHat and IBM can get Linux up to a supported and secure level, and MS keeps cutting off support to older and more affordable versions of their OS, many companies proverbial hands will be tied.

Clay Whipkey
Monday, May 03, 2004

I myself think Longhorn will not get the adoption M$ expects (people falling on each other to get it). Linux thin clients will almost certainly increase and people will take another look at Java.

An organization-most organizations has a finite amount of data that is growing. They access this data by a set of reports/insert update etc. Additionally, they generate reports. For an organization to adopt new technology one or more of these three conditions must be met

1)The present technology is simply not sufficient. This is certainly false. You can do maybe 90-95% of the thing with VB6 or VB.NET /Windows XP. Where does Longhorn fit the bill here?. Dont show me XAML. Its cool But thats about it

2)Can it get my reports faster?. A faster processor helps. But another operating system with humungous requirements?. No.

3)Is it cost effective?. In Longhorns case it will most probably require a hardware upgrade too.

The folks who adopt it will be the IT types. Not people who use spreadsheets or end users in typical offices. Microsoft has ertainly alienated them. We will see desktop linux/linux thin clients grow. Longhorn will help, not hinder desktop linux. You cant convince manufacturing organizations with some stuff like XAML. Its  cool. But they can do without it.

Karthik
Monday, May 03, 2004

Well i work in the SME support/programming sphere and IMO the stakeholders in that market couldn't give a rat's proverbial about OS's and hardware. 

They just want their LOB programs to run.  Whatever OS their LOB program runs, that's what they'll get.  Whatever hardware that OS requires, they'll get it.

These people typically lag the bleeding edge by a good 2 years and consequently won't be put out by launch day requirements.

As an aside, I used to own a PC retail shop.  We used to scale our main offerings to the OS du jour as clients generally wanted a PC to run "winders".  Other than the odd geek, they cared not for less demanding OSs be they Linux, OS/2 or whatever as they wanted "games to run".  Until the alternatives can run consumer LOB apps - games - I don't see much retail slippage either.

Motown (AU)
Monday, May 03, 2004

Missed my main point.

Until the LOB availability picture changes, MS knows they'll get you in the end.  LOB trumps all.

Motown (AU)
Monday, May 03, 2004

What is LOB?

Karthik
Monday, May 03, 2004

LOB=Line of business

eg Scripting software for doctors, job tracking, CRM etc etc

and games for kids ;-)

Motown (AU)
Tuesday, May 04, 2004

I think the LOB picture will change because of LongHorn. Even Macromedia is in peril because they have to recode + compete with M$ now.

Karthik
Tuesday, May 04, 2004

That's pure out of my backside guessing Karthik ;-)

I can't predict 6 months down the track, yet you're predicting secondary effects in an unknown future market based on an unreleased OS's hardware requirements in 2006-7.

While we're in the field of wild speculation, I'd appreciate the winners at Moonee Valley this weekend.  That's less than a week away, so you shouldn't have any trouble with that ;-)

j/k

Motown (AU)
Tuesday, May 04, 2004

I admit its my backside guessing too. But i work in a very typical manufacturing setup where something as disruptive as this is unimaginable. !.

I dont think even the port from 16-32 bit was so disruptive as longhorn is.

Karthik
Tuesday, May 04, 2004

> Pentium 100Mhz with 32 MB ... runs ... too slow, because of... the disk.

> In my opinion, not CPU or RAM is important, but it's the disk that does most of the work in modern OSses.

You know, if you had more than 32 MB of RAM, then more of the work would be done using RAM instead of using disk ... and because RAM is much faster than disk, so the overall computer would be much faster.

Christopher Wells
Tuesday, May 04, 2004

They are, in the words of Wayne Gretzky, "Skating to where the puck is going to be". Or at least where they *think* it will be in 2006/2007/whenever.

Keith Moore
Tuesday, May 04, 2004

Ya know, a few years ago, I read in a sci-fi book to be, a comment made by one of the lead characters to another, "I haven't bought anything since Microsoft folded."

At the time I chuckled, now I say, Hhhhuuuuuummmmmmmm.

The Patrick
Thursday, May 27, 2004

*  Recent Topics

*  Fog Creek Home