AMD64/IA32e to be intentionally unsupported by tech industry?

[MadRat]

Windows XP/64-bit will support IA64 and AMD64/IA32e, right? So will Windows 2003, right?

Just wondering. The world seems to think that IA64 is going the way of the dodo now that Intel has saw the market for IA32e. Oracle just announced that they won't port their 32-bit x86 products over to IA32e, but will rather stick with the IA64 gameplan. This during the time that AMD has partner after partner publically announcing how easy the porting work can be done for AMD64 compatibility! Call me silly, but that sounds like the big technology partners in the industry (Intel and HPaq) are going to hang on to IA64 at all costs. Oracle is also recently tilted towards grid computing, though, which makes me wonder why they wouldn't want to include full AMD64/IA32e compatibility to this project. (The decision seems to contradict their 5 principles slogan of "Globalization. Simplification. Standardization. Automation. Innovation." unless we're missing part of the message somewheres...)

Could it be Intel will hunker down and buy their way out of the IA32e hole by paying firms not to port for it?

 

[n0cmonkey]

I see ia32e as absolute garbage that someone will be tricked into purchasing for midrange or lowend servers and maybe desktop/workstations. IA64 will be in the high end servers. If Oracle supports ia64, it won't die anytime soon. The government will probably start moving their alpha and hppa machines over to ia64 instead of sparc4u. It's a shame really.

From what I have read so far, AMD64 is the way to go. I will be ignoring ia32e.

 

[Nothinman]

Intel and HP will support IA64 all the way to the grave, they've poured so much money into it over the years they can't afford to switch to something else.

 

[drag]

Originally posted by: Nothinman
Intel and HP will support IA64 all the way to the grave, they've poured so much money into it over the years they can't afford to switch to something else.

If they do that they deserve to die. They aren't even aware that they are violating one of the fundamental rules of business.

They are commiting the sunk cost fallacy. You can't throw good money after bad and expect to be successfull.

I assume that most people already know it, but since I like to talk I'll give a example.

Say you bought a car for 5000 dollars. It's a lemon and it seized up after 200 miles after puking it's oil all over the highway and you have had it rebuilt for 2000 dollars. It worked for a little bit, and now the tranny is going out. You take it to the mechanic and he tells you that the dealership did the sawdust trick on you (a noisy tranny can be made to seem to work very well after dumping a couple of handfulls of fine sawdust into it). It's going to cost another 1500 dollars to get it fixed. Plus the engine is running rough and the mechanic thinks that it's warped from the overheating and didn't get corrected in the low-buck engine rebuild you paid for recently. It may or may not be a problem. Plus the bearings in the wheels need repacking, maybe even a replacement spindle assembly, and there is rust in the gas tank.

The mechanic (who is completely trustworthy) has a car that a customer didn't pay for it's repairs and he will sell it to you for 3000 dollars to square the cost with the previous customer (within his legal rights). He says it's in great mechanical condition, even if it's a bit scruffy looking and he is whilling to stand behind his work and fix it for cost-of-parts if anything big mechanical happens in the next year or so.

So now the math is 7000 + 1500 = 8500 for a working car, that may fvck up OR a 7000 + 3000 = 10000 for a car that may still mess up, but is much less likely too, plus your going to be spending 10000 for a 3000-dollar ugly car. Right?

Wrong. Of course the orginal 7000 dollars doesn't exist anymore as far as your concerned. It's gone, you can't NOT waste it because it's already been blown, gone, and done for. Unless you can travel back in time there is no way to recover the loss.

Your real choice is do you spend 1500 on a car that is probably going to break again, or do you cut your losses and go with a reliable 3000 one that maybe isn't as cool looking?


The intellegent choice is to ignore past costs and go with the 3000 dollar car.

So is intel/hp going with the intanic in hopes that all the market trends in the world for the past few years will suddenly reverse and they create the next "Alpha" chip that everyone will drool over (and pay big bucks for)? Or is Intel going to bow to AMD this time and let them win and hope that they can out-improve AMD/IBM on AMD/IBM's own design and go with a much lower profit potentially?

Tough choice for intel.

Oh, BTW. HP is now going to start selling Opteron-based servers.

 

[jhu]

your argument is invalid because in the market segment that itanium is competing in, itanium does quite well in terms of price.

yes, intel did sink quite a bit of money. but it costs money to bring a new architecture to market. i'm sure hp spent quite a lot to develop pa-risc. dec did the same with alpha, and now look where they are. how much did ibm/motorola/apple spend to develop powerpc?

 

[drag]

Originally posted by: jhu
your argument is invalid because in the market segment that itanium is competing in, itanium does quite well in terms of price.

yes, intel did sink quite a bit of money. but it costs money to bring a new architecture to market. i'm sure hp spent quite a lot to develop pa-risc. dec did the same with alpha, and now look where they are. how much did ibm/motorola/apple spend to develop powerpc?

It's not a arguement its a observation.

The point was that NOBODY CARES ABOUT HOW MUCH MONEY WAS SPENT ON DEVELOPING THE PROCCESSOR.

That's gone and done for. You can't recover costs, once you spend money on something it's gone forever and shouldn't enter into business calculations if you want to be successfull. What? Are you going to spend a extra 2 billion dollars on something just because you already wasted 32 billion dollars on something?

The only people stupid enough to use that logic and get away with it are people in governments because they aren't spending their own money and people vote for people that make them feel good rather then people that do their jobs well.

What matters in business is how money your GOING to spend vs how much money your GOING to make. You only care about what you SPENT on something when the sht*t hits the fan and it's time to start firing people (or selling stock).

The fact is that IBM is making bucket loads off of selling PowerPC proccessors. Power3's made them money, Power4's are making them money, Power4+'s are making them money, Power970's are making them money, and Power5's will begin making their money in a couple years.

And were do you suppose AMD got much of the technology for opteron? From IBM research for the Power line of proccessors.

It doesn't matter if Intaniums are priced decently in comparision to Power4's or Sparc III's and IV's. Because if you don't sell any you aren't going to make money.

And Intel isn't selling many. Last figure I saw was something like 4700 in server sales this in one quarter vs Opterons that sold 10,000+ on their very first quarter. And how long has Itanium been around?

It's not a healthy ecomonic forcast for any tech company right now. People don't want to spend much money on new high end servers. And even if they did a company that owned IBM is going to buy IBM.. A company that owns suns machines will generally buy sun machines, A company that buys HP will generally buy HP. Change is bad, it makes things expensive.

Plus Intel is not a company of infinate resources. How many lines of proccessors will they have?

Pentium 4ee (or eeee or e? whatever), Xeons, itaniums, mobile pentiums, celerons (which blow donkey nuts), and now the IA32e which at best is a AMD64 clone? And a slightly inferior and more expensive one at that(so far it seems)?

Seems like intel has streached themselves kinda thin here. The only real one that isn't that profitable is the itanium.

 

[NightCrawler]

AMD64 will be successful only when M$ gets the OS out for it. Now that Intel has decided to support AMD64 it looks a lot better for AMD but M$ seems in no hurry to get Windows XP 64 version in people's hands to fast. It's going to require a lot of testing.

 

[drag]

Originally posted by: NightCrawler
AMD64 will be successful only when M$ gets the OS out for it. Now that Intel has decided to support AMD64 it looks a lot better for AMD but M$ seems in no hurry to get Windows XP 64 version in people's hands to fast. It's going to require a lot of testing.

Ya, that's one of the things that suck about MS controlling around 90% of the desktop arena. Since MS windows (as far as consumers are concerned) only runs on one archatecture any thing new coming along gets a bit screwed over.

Look at for instance the power970 platform. From a technical stand point it's superior to x86 cpu's. It is powerfull enough, it has 64bit registers and the PPC ISA is much more flexable then the x86 ISA.

Other then that the big advantage that the power970 has over it's x86 rivals is that it is a miser when it comes to electricity. At 2ghz the PPC-based chip uses less then a 4th of the wattage of a AMD or Intel counterpart and it's not realy that expensive to build.

It would be perfect for a quiet computer. People talk about undervolting unclocking cpu's, spending money on big powersupplies with the quietest fan possible, big aluminum cases and expensive passive cooling heatpipes and all sorts of gimmacry to get a quiet computer.

Instead they could just get a PPC computer and put a few resistors here and there to slow fans down and they would have a quiet, cool running computer at a much lower price with 80-90% of the performance of a high-end x86.

But you can't do that, because Windows doesn't run on PPC. You can't just go buy a PPC motherboard and cpu from newegg, because their's no market.

AMD stuff is still expensive because Windows doesn't support AMD64 bit yet.

The situation sucks in terms of a hardware guy's viewpoint. Escentually what you have is Window's limitations dictating the hardware market.

 

[MadRat]

Sounds like you think the Power970 platform should have found its way into the PC+Windows market. IBM could essentially bring that to bear with their power over the market, but it won't happen. IBM is a Linux shop now. What people out in the world need is a form of Linux that uses MS's file structure. From what I've heard from person after person is that Linux sucks because its file structure looks like hell. Its intimidating at first, but its really pretty simple, no more complex than using the MAC's file structure. People just got used to the C: drive and a "Windows" folder for the OS, I guess...

 

[drag]

Haha. That's funny.

Windows file structure is crap. It makes no sense at all. They just put stuff were ever they feel like and it's up to everyone to remember to obscure naming sceme that follows no standards and is fragile to touch.

The whole point of *nix is based around it's filing stuff.

/boot is were boot files are.
/etc is wer system wide configs are kept.
/var is for various things like log files.
/home is for home files.
/proc is a place to find information collected from the operation of the kernel
/tmp is a place to stick temporary files.
/bin is for the most basic executables needed by the average user.
/sbin is the most basic system apps need to get everything running
/usr is the place to keep all the programs and related docs that people use in the OS.
/dev provides a standard file-like interface with your systems hardware so that it's easy to access.
/lib is a place to keep the basic lib files needed for the apps in /bin and /sbin

Most *nix systems follow this sceme. Not exactly of course each OS has a slightly different idea of how to arrange everything. But for the most part it's standard enough that a person used to Solaris can go to AIX then goto Linux then goto FreeBSD without feeling to alienated. OS X is the only *nix OS that doesn't seem to give a damn about it and pretty much raped it. The vestages of it is still their mostly intact but with a awfull lot of extra crap tagged on. What you see when you click on the harddrive is only a small percentage of the filing system that's on the drive, it's a bit hidden. You can still go to the /etc/ folder if you specify it in the gui for example and it still has most of the system wide configs in it. But to realy see it you need to open up a terminal and look around.

You try to ruin all that by forcing everything to get jumbled up to fit a flat filing sceme (i.e. Microsoft C: drive) created to work with f*cked up assumptions about a partitioning sceme from a 15 obsolete operating system about 7 year obsolete (x86 has come a long way since then) archetecture and you will turn the average Linux distro into a unworkable pile of goo.

 

[Monoman]

Originally posted by: MadRat
Sounds like you think the Power970 platform should have found its way into the PC+Windows market. IBM could essentially bring that to bear with their power over the market, but it won't happen. IBM is a Linux shop now. What people out in the world need is a form of Linux that uses MS's file structure. From what I've heard from person after person is that Linux sucks because its file structure looks like hell. Its intimidating at first, but its really pretty simple, no more complex than using the MAC's file structure. People just got used to the C: drive and a "Windows" folder for the OS, I guess...

Hahaha.... if Linux used MS's file system.. blah.. LOL

anyway... if people pulled their head out of the MS hole and thought logically for a sec.. they would see that Linux's file system is actually easier to use. Even my computer lame wife agrees with me on this and she know how to get on the internet

 

[jhu]

The fact is that IBM is making bucket loads off of selling PowerPC proccessors. Power3's made them money, Power4's are making them money, Power4+'s are making them money, Power970's are making them money, and Power5's will begin making their money in a couple years.
last i heard (from business week), ibm's processor operations aren't the ones making the big bucks but rather their other tech services.

Ya, that's one of the things that suck about MS controlling around 90% of the desktop arena. Since MS windows (as far as consumers are concerned) only runs on one archatecture any thing new coming along gets a bit screwed over.
windows nt did run on multiple platforms. but like you said, the other ones weren't profitable so they were dropped.

Look at for instance the power970 platform. From a technical stand point it's superior to x86 cpu's. It is powerfull enough, it has 64bit registers and the PPC ISA is much more flexable then the x86 ISA.
hey, so's itanium. powerpc has the advantage that ibm uses them in their servers and apple used it instead of some other architecture. would alpha be alive today if apple had chosen it instead? probably

But you can't do that, because Windows doesn't run on PPC. You can't just go buy a PPC motherboard and cpu from newegg, because their's no market.
oh there's a market. and there are manufacturers that are selling g4+motherboard combos for around $700. it's just that x86 hardware has been so commoditized for 20+ years that they can't compete. any new architecture that comes in must be low enough in price to compete with all this "legacy" equipment.

The situation sucks in terms of a hardware guy's viewpoint. Escentually what you have is Window's limitations dictating the hardware market.
as i said before, you can buy any type of cpu+motherboard combo right now: powerpc, alpha, sparc. the question is, are you willing to spend extra for it?

 

[drag]

Originally posted by: jhu

The situation sucks in terms of a hardware guy's viewpoint. Escentually what you have is Window's limitations dictating the hardware market.
as i said before, you can buy any type of cpu+motherboard combo right now: powerpc, alpha, sparc. the question is, are you willing to spend extra for it?

Thats silly. Of course not. Nobody is going to go out and buy a 700 dollar motherboard combo for commodity hardware. (were can you find one anyways, just curious?)

Sure people use it for embedded stuff and special purposes. They are low power, cool proccessors that fit into niches that x86 can't fill.

But there is no real market, not like there is for x86.

The reason x86 is so cheap compared to other architectures isn't because of some technical brillance or even cutrate manufacturing (although that does happen with some cheap crap you can find). They are cheap because of economy of scale.

Look at x86 motherboards. You can get decent ones for less then fifty bucks. A fabulous one is usually below 200 dollars.

Look at all the stuff that comes on them, the special plastic socets and slots. The capacitors, the packaged transistors, all that stuff. Cheap as dirt.

They can get away with selling them that cheap because they sell so freaking many. It's called the economy of scale. The cost of manufacture is spread out as much as possible.

You can take a motherboard, make and sell only 500 of them, you would have to sell them for several hundred dollars apiece just to hope to come close to breaking even. However, if you sell 50,000 of them then everything automagicly becomes much cheaper per unit. You don't even have to cut corners or make a inferior design or anything, it's just that once you get to a certain production size then everything becomes much cheaper.

That's what I mean by there is currently no market for PPC machines. A few people selling them to individuals to make amiga machines doesn't constitute a viable consumer market.

Oh, ya. (sarcasm) I am also gratefull for MS to only support one platform. I am sure their are others in the industry that are perfectly happy about it too. I am sure that IBM or whoever is happy and content to only be players in the mid-high end unix markets and would just hate to build hardware for the average person. After all, in this world there is one thing we don't need, and that's choice.

Don't get me wrong. I do not think that x86 = crap and the ISA is outdated and obsolete. I know the x86 ISA is just a hardware abstraction layer nowadays to give people a common thing to program for, while the hardware underneath it is constantly evolving, changing and improving. With AMD's extensions and modifications to the x86 ISA we could have a dependable, fast, and cheap platform for several more years to come.

Until recently their was no reason to care about ppc setups. Now there is. And more then likely my next computer is going to be PPC, a Apple of course. Probably next year around this time I will go out and order myself one of those ibooks to replace my crappy gateway (damn crappy). The only other one I would consider would be a IBM laptop(x86 obviously), but those command as much and even more of a price premium then Apple's!

 

[n0cmonkey]

You can get PPC hardware from pegasos, but it's broken.

 

[RadBrad]

Originally posted by: MadRat
Windows XP/64-bit will support IA64 and AMD64/IA32e, right? So will Windows 2003, right?

Just wondering. The world seems to think that IA64 is going the way of the dodo now that Intel has saw the market for IA32e. Oracle just announced that they won't port their 32-bit x86 products over to IA32e, but will rather stick with the IA64 gameplan. This during the time that AMD has partner after partner publically announcing how easy the porting work can be done for AMD64 compatibility! Call me silly, but that sounds like the big technology partners in the industry (Intel and HPaq) are going to hang on to IA64 at all costs. Oracle is also recently tilted towards grid computing, though, which makes me wonder why they wouldn't want to include full AMD64/IA32e compatibility to this project. (The decision seems to contradict their 5 principles slogan of "Globalization. Simplification. Standardization. Automation. Innovation." unless we're missing part of the message somewheres...)

Could it be Intel will hunker down and buy their way out of the IA32e hole by paying firms not to port for it?

You can all talk about who doin what to who all you wan,t but Intel, Amd, Hp ,Sunmicro, Ibm and all I forget to mention
all in bed together having a big orgy on our dime, and all any of us are goin to get out of it is a sniff of the sheets.


And oh yea Diddo's on the GO HUSKERS Thing from Wellfleet

 

[n0cmonkey]

Originally posted by: RadBrad

Originally posted by: MadRat
Windows XP/64-bit will support IA64 and AMD64/IA32e, right? So will Windows 2003, right?

Just wondering. The world seems to think that IA64 is going the way of the dodo now that Intel has saw the market for IA32e. Oracle just announced that they won't port their 32-bit x86 products over to IA32e, but will rather stick with the IA64 gameplan. This during the time that AMD has partner after partner publically announcing how easy the porting work can be done for AMD64 compatibility! Call me silly, but that sounds like the big technology partners in the industry (Intel and HPaq) are going to hang on to IA64 at all costs. Oracle is also recently tilted towards grid computing, though, which makes me wonder why they wouldn't want to include full AMD64/IA32e compatibility to this project. (The decision seems to contradict their 5 principles slogan of "Globalization. Simplification. Standardization. Automation. Innovation." unless we're missing part of the message somewheres...)

Could it be Intel will hunker down and buy their way out of the IA32e hole by paying firms not to port for it?

You can all talk about who doin what to who all you wan,t but Intel, Amd, Hp ,Sunmicro, Ibm and all I forget to mention
all in bed together having a big orgy on our dime, and all any of us are goin to get out of it is a sniff of the sheets.


And oh yea Diddo's on the GO HUSKERS Thing from Wellfleet

And your proposed solution is?

 

[OffTopic1]

Originally posted by: NightCrawler
AMD64 will be successful only when M$ gets the OS out for it. Now that Intel has decided to support AMD64 it looks a lot better for AMD but M$ seems in no hurry to get Windows XP 64 version in people's hands to fast. It's going to require a lot of testing.

It might seem so at the moment, but Linux64 could be the kick start that MS needed to get their arse in gear.

 

[RadBrad]

Originally posted by: n0cmonkey

Originally posted by: RadBrad

Originally posted by: MadRat
Windows XP/64-bit will support IA64 and AMD64/IA32e, right? So will Windows 2003, right?

Just wondering. The world seems to think that IA64 is going the way of the dodo now that Intel has saw the market for IA32e. Oracle just announced that they won't port their 32-bit x86 products over to IA32e, but will rather stick with the IA64 gameplan. This during the time that AMD has partner after partner publically announcing how easy the porting work can be done for AMD64 compatibility! Call me silly, but that sounds like the big technology partners in the industry (Intel and HPaq) are going to hang on to IA64 at all costs. Oracle is also recently tilted towards grid computing, though, which makes me wonder why they wouldn't want to include full AMD64/IA32e compatibility to this project. (The decision seems to contradict their 5 principles slogan of "Globalization. Simplification. Standardization. Automation. Innovation." unless we're missing part of the message somewheres...)

Could it be Intel will hunker down and buy their way out of the IA32e hole by paying firms not to port for it?

You can all talk about who doin what to who all you wan,t but Intel, Amd, Hp ,Sunmicro, Ibm and all I forget to mention
all in bed together having a big orgy on our dime, and all any of us are goin to get out of it is a sniff of the sheets.


And oh yea Diddo's on the GO HUSKERS Thing from Wellfleet

And your proposed solution is?

My proposed solution would be to move this thread to a forumn where it belongs because the way you are approaching it is pure speculation, political by nature, and other than its titles has nothing to do with operating systems.

 

[n0cmonkey]

Originally posted by: RadBrad

Originally posted by: n0cmonkey

Originally posted by: RadBrad

Originally posted by: MadRat
Windows XP/64-bit will support IA64 and AMD64/IA32e, right? So will Windows 2003, right?

Just wondering. The world seems to think that IA64 is going the way of the dodo now that Intel has saw the market for IA32e. Oracle just announced that they won't port their 32-bit x86 products over to IA32e, but will rather stick with the IA64 gameplan. This during the time that AMD has partner after partner publically announcing how easy the porting work can be done for AMD64 compatibility! Call me silly, but that sounds like the big technology partners in the industry (Intel and HPaq) are going to hang on to IA64 at all costs. Oracle is also recently tilted towards grid computing, though, which makes me wonder why they wouldn't want to include full AMD64/IA32e compatibility to this project. (The decision seems to contradict their 5 principles slogan of "Globalization. Simplification. Standardization. Automation. Innovation." unless we're missing part of the message somewheres...)

Could it be Intel will hunker down and buy their way out of the IA32e hole by paying firms not to port for it?

You can all talk about who doin what to who all you wan,t but Intel, Amd, Hp ,Sunmicro, Ibm and all I forget to mention
all in bed together having a big orgy on our dime, and all any of us are goin to get out of it is a sniff of the sheets.


And oh yea Diddo's on the GO HUSKERS Thing from Wellfleet

And your proposed solution is?

My proposed solution would be to move this thread to a forumn where it belongs because the way you are approaching it is pure speculation, political by nature, and other than its titles has nothing to do with operating systems.

PM a mod. And what does the forum have to do with your previous post about how all of these companies are fleecing us?

 

[CTho9305]

Regarding the "sunk cost" debate - I doubt they really are making that mistake. Intel killed the Alph EV8 after spending 3 years developing it... that's a pretty big investment that they dumped.

 

[jhu]

of course they also dumped their i860 years before

 

[drag]

There is no question about the sunk cost fallacy.

You can't afford to spend money to save money you already spent. It's illogical. Once money is gone it doesn't matter any more, all you have to deal with is the reality of your situation.

One good example I read once is about buying a ticket to a concert.

Say you buy a ticket to a concert for 50 bucks. It's non-refundable and you think it's going to rock. But someone else has been to one in a different state and you trust his judgement in music and he says that the concert realy realy sucks. It's a rock band with little talent, but they have fun concerts normally, but they got on a artistic bent and it just sucks now.

So do you go to the concert anyways just because you spent the 50 bucks, but be miserable for the entire time your there? Or do you go off and hang with your freinds and have fun?

Well the money doesn't enter into it anymore since you understang the concept of sunk-cost and your only 2 real choices are have fun with your friends or be miserable at the concert.

The choice is much easier and your going to make the correct one, which is to go have fun and forget about your 50 bucks.

Now I am not saying that Intel is going to dump Intanium, it would be kinda stupid for them to abandon it complety for a bunch of different reasons pther then just the money they spent.

You say they spent 3 billion dollars on intanium developement (or whatever, I am making up numbers), but it's not that hot. So they should spend more money on it? How about another 500 million?

What it seems at first is that they are going to spend 500 million to help recover the 3 billion already spent.

But in reality it would be that they are continuing to spend 500 million on a project that has no sign of profitability. A white elephant.

Ignoring the sunk-cost fallacy of in business is like trying to shovel out a basement for a house in quicksand. The faster you shovel the more work you do, the less gets done, and at the end of the day you end up with a broken back and no hole.

Now I don't know if itanium is selling enough to make enough of a profit to justify it's existance to intel. It's not going anywere soon anyways, even if it was a money pit. High-end buyers are long range buyers, dumping a product that you promised to support for years and years is a good way to loose business to competators and NEVER get it back.

Normally it would be stupid for a company to dump a product line if it's making money. Profit is profit after all even if it's small. However in computer hardware if you don't spend your money to make yourself money fast (fast turnaround), then you can't spend it on stuff like research. Which is vital to staying ahead and staying alive.

 

[CTho9305]

drag - What I'm saying is Intel knows that sunk costs are gone, and if they are continuing it is only because they believe it will be profitable.

edit:

Originally posted by: drag
Haha. That's funny.

Windows file structure is crap. It makes no sense at all. They just put stuff were ever they feel like and it's up to everyone to remember to obscure naming sceme that follows no standards and is fragile to touch.

Hmm... Windows-related stuff is in C:\Windows. Windows is you operating system - better not touch that. Programs are in C:\Program Files. Your documents and settings are somewhere in C:\Documents And Settings\username\. Seems pretty clear to me.

The whole point of *nix is based around it's filing stuff.

/boot is were boot files are.
/etc is wer system wide configs are kept.

"etc"? In what language does "etc" mean "config files"?

/var is for various things like log files.

Why not /log?

/home is for home files.
/proc is a place to find information collected from the operation of the kernel

What's a kernel? Why do I want to see this?

/tmp is a place to stick temporary files.
/bin is for the most basic executables needed by the average user.

What's a bin?

/sbin is the most basic system apps need to get everything running

Different bins? Now I'm even more confused.

/usr is the place to keep all the programs and related docs that people use in the OS.

The second place I would expect to find home directories, after /home is something named /usr. Yet for some reason /usr isn't even writable for non-root....

/dev provides a standard file-like interface with your systems hardware so that it's easy to access.

Why do I want to access my hardware?

/lib is a place to keep the basic lib files needed for the apps in /bin and /sbin

What are libs?

Of course, I know the answers to all of the above, and understand the reasoning. But you'd have to be smoking crack to think your average user knows or cares about the differences between /sbin, /bin, /usr/bin, /usr/sbin, and /usr/local/bin. Windows' system is pretty simple: \windows: don't touch it. \program files: install your programs here. \documents and settings\username: put your documents and settings here. There are some standards to it, and you get the advantage of having all your system stuff (/sbin, most of /boot, system-related parts of /etc) in the windows directory, where noobs don't go.

Most *nix systems follow this sceme. Not exactly of course each OS has a slightly different idea of how to arrange everything. But for the most part it's standard enough that a person used to Solaris can go to AIX then goto Linux then goto FreeBSD without feeling to alienated.

And you can go between 95/98/ME or 2k/XP without any trouble, and even between the two groups the differences are minor (specifically, where the user stores his/her stuff).

You try to ruin all that by forcing everything to get jumbled up to fit a flat filing sceme (i.e. Microsoft C: drive) created to work with f*cked up assumptions about a partitioning sceme from a 15 obsolete operating system about 7 year obsolete (x86 has come a long way since then) archetecture and you will turn the average Linux distro into a unworkable pile of goo.

You need to provide some evidence for this statement. I would propose the following:

/Programs
__/Appname
____/Version
______/Bin
______/Conf
__/man (maybe call it "/doc")
/Users
__/Username
____/Documents
____/Appname (maybe have a separate "program setttings" directory containing all per-user app configs)
______/Conf
/.System (in here you can use scary unix-style naming conventions if you want)
__/Bin
__/Devices
__/Etc
__/include
__/Libraries

x86 sucks. Why do we keep it around? Legacy.
The current unix directory structure is not ideal for a user-friendly desktop. Why do we keep it around? Legacy (and stubbornness - unix geeks insist the current structure is so wonderful yet rarely provide REAL evidence, other than "I'm used to it").

I tried using my setup in a linuxfromscratch build and gave up, because SOOO many poorly-written apps don't like being moved around. However, a linux distribution called "GoboLinux" does use a relatively sane directory structure.

My apologies for an almost-completely-off-the-thread-topic post.

 

[spyordie007]

Long thread and I read it quickly so I may be missing this, but there are a lot of people complaining about Microsoft not releasing OSes for AMD64/Opteron and this is simply not the case. Windows XP 64 bit for AMD64/Opteron and Windows Server 2003 for AMD/Opteron are both at the public beta stage; it wont be that long before they are RTM.

http://www.microsoft.com/windowsxp/64bit/downloads/upgrade.asp
http://www.microsoft.com/windowsserver2003/64bit/extended/trial/default.mspx

And BTW Drag, I'm going to have to side with Ctho on the file naming conventions. Windows file naming conventions is no more confusing than *nix for any intermediate to advanced user; and for your basic run of the mill end-users (read 95+% of the population) they could care less.

 

[drag]

Originally posted by: spyordie007
Long thread and I read it quickly so I may be missing this, but there are a lot of people complaining about Microsoft not releasing OSes for AMD64/Opteron and this is simply not the case. Windows XP 64 bit for AMD64/Opteron and Windows Server 2003 for AMD/Opteron are both at the public beta stage; it wont be that long before they are RTM.

http://www.microsoft.com/windowsxp/64bit/downloads/upgrade.asp
http://www.microsoft.com/windowsserver2003/64bit/extended/trial/default.mspx

And BTW Drag, I'm going to have to side with Ctho on the file naming conventions. Windows file naming conventions is no more confusing than *nix for any intermediate to advanced user; and for your basic run of the mill end-users (read 95+% of the population) they could care less.

I know they are releasing it, but it's WAY behind scedual.


And the difference between Windows and Linux filing systems is that one is accessable and modifiable and one is not. It's about utility, each subdirectory exists for a different purpose.


The different directories corrispond to the different purpose of the files inside each directory. Their is rhyme and reason to their placement , and you can tell the purpose of a file just by it's location in the directory system.

In windows when you install a program, it's DLL files can end up in twenty or thirty different locations. What if I need to uninstall a program for a newer version and the old uninstall program is broken? Were do I go to find files that may conflict with the newer version of a program?

The only thing that makes windows remotely easy to deal with is that you have to memorize the different 3 letter combinations that are at the end of most files.

Anyways if 95% people don't care about /dev then that's irrelevent. The filing system doesn't make it any different or more difficult then anything else.

Now if the C: d: and E: is so wonderfull then tell me which drive letter gets assigned to what partition when I add a 3rd harddrive. And also point out to me how much easier this is to the average user to understand vs the Linux answer of "it gets mounted wherever you want."

Also tell me if I run a service (a http or ftp or SMB or whatever) were does the log files end up incase the system crashes and I need to boot it up with a boot disk and troubleshoot to find out what happened.