Is the x86 chip market really the diaster I think it is?
I've been following the x86 websites for a few months now ever since the 970 was released.
From my lay reading, it seems like Apple and IBM will have no problem competing with x86 land for the next few years.
From what I've read:
Intel's Itanic is a complete disaster. It's a dead-end ISA. You know a product is in bad shape when it's company starts a PR campaign to demonstrate why the product isn't really a failure.
Intel's x86 chips are late and hot, and getting later and hotter. They seem to be scrambling to tack on 64 extensions.
AMD seems to be in better shape but their smaller market-share and financial troubles seem to be a cloud hanging over the company.
With the explosion of Linux, the lock the win side of wintel has had on the computer market is eroding everyday. IBM is rumored to be dumping or preparing to dump windows from all of their internal desktops. I'm sure IBM would rather have Linux on their own PPC chips than x86.
Of course I don't think the whole x86 world is going to just die anytime soon, even though that would be nice. I'm sure the x86 crowd would have a completely different view the situation.
From my lay reading, it seems like Apple and IBM will have no problem competing with x86 land for the next few years.
From what I've read:
Intel's Itanic is a complete disaster. It's a dead-end ISA. You know a product is in bad shape when it's company starts a PR campaign to demonstrate why the product isn't really a failure.
Intel's x86 chips are late and hot, and getting later and hotter. They seem to be scrambling to tack on 64 extensions.
AMD seems to be in better shape but their smaller market-share and financial troubles seem to be a cloud hanging over the company.
With the explosion of Linux, the lock the win side of wintel has had on the computer market is eroding everyday. IBM is rumored to be dumping or preparing to dump windows from all of their internal desktops. I'm sure IBM would rather have Linux on their own PPC chips than x86.
Of course I don't think the whole x86 world is going to just die anytime soon, even though that would be nice. I'm sure the x86 crowd would have a completely different view the situation.
Comments
Where X86 is strongest is at the low end where competitors have 2.8Ghz systems for $500. Why would the masses need to spend any more money? So they don't. Very few companies seem to be making money.
Companies like Alienware are so starved for Press they actually released a PR claiming to have bested the G5 based on PC Mags test using(ugggh) Premiere 6 and Word amongs other apps.
I'm bored with PCs. P2P isn't even that much of a draw because it's a fulltime job stripping the adware/spyware from your computer. Gaming isn't a big deal anymore because of low cost consoles.
What needs to happen is a software renaissance. I don't care about "more" apps I want to see apps that work so right it feels like mind control. Less bugs more integration.
From what I read Comdex was dead. Sounds like people are tired of talking about how fast their computers are or how much of x the computer has and actually start working with them. Ouch...they find out software hasn't exactly kept up.
Linux is alive and bumping.
As long as Linux runs on the x86 platform then there is no need for people like me to invest in PPC hardware. A Unix-like operating system is all I need. Untill PPC is leaps and bounds ahead of x86 and the price is right I will stick with the x86 platform for my future investments.
It's funny but even when I do buy a new Mac I will end up running linux on it. Just like the iMac I have now.
Anywho, I don't think we will see the end of x86 for a longgg time. Intel has the cash and the knowledge to keep it alive forever. New instruction sets will get phased in, older ones phased out, but the beast will live. Now if they start implementing DRM on motherboard BIOS for x86 chips...
That's a pretty big assumption. It sounds a lot like what people have been saying about Microsoft for a long time when discussing Linux.
The 970/980 roadmap looks like nothing but clear sailing for Apple/IBM for the next couple of years at least.
Intel looks like they are having major problems. Not just the usual glitches here and there.
Down the road I would imagine Apple would want a Linux on PPC market to take off based upon cheap PPC commodity motherboards. Boards that have none of the legacy gunk from the old x86 world and none of the bios/drm monkey business I think MS is looking at to to stem the Linux momentum.
Both Apple and IBM probably want commodity PPC boards to take off. Cheap PPC motherboards to run Linux straight from the Apple Store? Plug them right into your existing beige x86 boxes. Seamless integration with OS X and Linux.
So where are we now? AMD knows what Intel cannot admit to itself yet: That what the market wants is Yet Another Feature Bolted On to x86. Legacy is always a bitch, but it's a real bitch when it's held a commanding share of an immense market for years. So AMD rolls out x86-64, which to be fair is one of the better-conceived extensions that has been glued on to the toy CPU that IBM engineers first put into their toy PC because it was cheap. It's outselling Itanium by something between six and ten times, which puts its sales numbers somewhere below the Tablet PC. As of yet there is no compelling reason to go 64-bit, and no single vendor can force the issue in a commodity market, where legacy and the lowest common denominator rule with an iron fist.
I expect to see x86 linger for a very long time in various forms. It's a lot bigger than Intel is at this point, and the longer they fail to acknowledge that, the more trouble they'll find themselves in. That's not a death knell by any means, though. Even in the pathologically unlikely case where they really have thrown all their eggs in the VLIW basket, they have enough money to coast for years before they turn their ship around. And they will.
I agree that the x86 market is something of a shambles right now. But that's not because x86 is weak. It's because x86 is so strong that all attempts by individual companies to steer it away from its current incarnation - or kill it outright - will be laughed out of the marketplace. And this is a shambles because the x86 is a weak architecture intrinsically. It was never supposed to do what it has been asked to do, or scale as far as it's been asked to scale, and that shows: Look at how much money and talent Intel throws at it every year.
On that note, moving to General Discussion.
Do you really think that Joe Schmoe cares that the architecture is old? All he wants is a cheap box, just like nearly all the buyers for large corps. They could care less about the 970's more advanced instruction set or bus or whatever.
And there's no point in even mentioning the Itanium, since it is not an X86 chip.
Edit: I didn't notice this from Amorph, since Itanium discussion is well underway anyway:
So where are we now? AMD knows what Intel cannot admit to itself yet: That what the market wants is Yet Another Feature Bolted On to x86
It's a good point, and it's incredibly galling for Intel. But that doesn't mean that The Itanium will always be such a bit player. Intel has an awful lot of those billions available for marketing, and it fired the first shot today. DOn't count the Itanium out just yet, especially if Montecito lives up to its advanced billing. That is one monster proc.
Tuttle wrote:
Both Apple and IBM probably want commodity PPC boards to take off. Cheap PPC motherboards to run Linux straight from the Apple Store? Plug them right into your existing beige x86 boxes. Seamless integration with OS X and Linux.
Why would Apple want this? It would detract from Mac sales, and if there is one thing that Apple has proved since Jobs came back it is that it is more interested in selling small amounts of premium gear than large amounts of cheap gear.
What do you do for a living to have such an informed perspective on the "computer-chip" industry? (if that's not too personal a question to ask in these forums)
Today, I'd have to pay around 100 euros for a Athlon XP 2600+, over three times the performance. I think only graphic work and acting as a server can demand more performance than that from a computer, at least for the next couple years to come.
I don't have to speak to my processor; compilers do that for me. So it's no big deal if the architecture is weird or old. I don't understand why people who are not assembly programmers would care about the architecture and not the results (speed).
Weird that no one mentioned this already, but since the introduction of Pentium M, mobile x86 looks good compared to mobile PowerPC. Performance and efficiency is excellent, AFAIK prices are reasonable if you're content with less than the top model. Their 2004 roadmap is promising as well.
Originally posted by hmurchison
What needs to happen is a software renaissance. I don't care about "more" apps I want to see apps that work so right it feels like mind control. Less bugs more integration.
I think the way we get there is changing the style of software development. Open interfaces that let other programs efficiently harness the functionality in a particular program, separation of program functionality and UI, easily modified UI using XSL etc.
All this is beneficial to a "hacking" community, which generates innovation, which can be packaged by companies into polished solutions.
The component approach is also good because you don't have to upgrade any more than you need new features. If functionality x is already good and bug-free, you can keep the exact same binary.
Originally posted by Gon
I write this on a 800MHz Duron. It's almost enough for me.
Today, I'd have to pay around 100 euros for a Athlon XP 2600+, over three times the performance. I think only graphic work and acting as a server can demand more performance than that from a computer, at least for the next couple years to come.
I don't have to speak to my processor; compilers do that for me. So it's no big deal if the architecture is weird or old. I don't understand why people who are not assembly programmers would care about the architecture and not the results (speed).
Weird that no one mentioned this already, but since the introduction of Pentium M, mobile x86 looks good compared to mobile PowerPC. Performance and efficiency is excellent, AFAIK prices are reasonable if you're content with less than the top model. Their 2004 roadmap is promising as well.
The problem of the X86 architecture is not for today but for tomorrow. The future of X86 is unclear. As Amorph stated, the X86 is an almost universal standart and an obsolete ISA.
This obsolescence will carry more on more problems. Intel and AMD have suceed to tweaks the X86 code with sucess, but as chips gets more complicated, the X86 isa penalty will become more important.
I don't believe in the future of the X86 after 2010.
PS : my G4 533 is sufficiant for my current duties too.
Both Apple and IBM probably want commodity PPC boards to take off. Cheap PPC motherboards to run Linux straight from the Apple Store? Plug them right into your existing beige x86 boxes. Seamless integration with OS X and Linux.
jouster replied:
Why would Apple want this? It would detract from Mac sales, and if there is one thing that Apple has proved since Jobs came back it is that it is more interested in selling small amounts of premium gear than large amounts of cheap gear.
The hell it would! When Apple does this it will change the market completely. Imagine if Apple were selling G5 XServes that support Linux out of the box! This would take away my ONLY objection to Apple's upcomming XServes by IT. Wait and see...
Originally posted by Mike
Imagine if Apple were selling G5 XServes that support Linux out of the box! This would take away my ONLY objection to Apple's upcomming XServes by IT. Wait and see...
Terra Soft are an Apple-approved reseller that sell Apple hardware with Yellow Dog Linux pre-installed.
http://www.yellowdoglinux.com/products/
Originally posted by Mike
Tuttle wrote:
Both Apple and IBM probably want commodity PPC boards to take off. Cheap PPC motherboards to run Linux straight from the Apple Store? Plug them right into your existing beige x86 boxes. Seamless integration with OS X and Linux.
jouster replied:
Why would Apple want this? It would detract from Mac sales, and if there is one thing that Apple has proved since Jobs came back it is that it is more interested in selling small amounts of premium gear than large amounts of cheap gear.
The hell it would! When Apple does this it will change the market completely. Imagine if Apple were selling G5 XServes that support Linux out of the box! This would take away my ONLY objection to Apple's upcomming XServes by IT. Wait and see...
Then why don't they do it?
How would it 'change the market completely?' Apple wants you to buy its OS and run it on its hardware. Why would it want you to run Linux, from which it will make nothing?
Are you saying that you think IT buyers would move to XServes solely because running LInux was feasible?
I can't buy that, but whatever.
AMD isn't even selling large numbers of their 64 bit chips. Nothing seems to even be on their roadmap with which to compete with IBM's upcoming 90 nm. 980 chip. If the rumors are true and IBM is producing 2.6 GHz G5 chips which Apple is preparing for an early 2004 release, the bottom may just drop out of AMD's 64 bit hopes.
Intel's successor to the Itanium looks to be a monster of a chip. However, it is an insane chip. How on earth Intel expects to compete in price to a comparable 980 or even a Power5, I have no idea. The amount of cache memory they are proposing for the chip must be quite costly.
Sometimes, one has to wonder about Intel. They have a real winner of a chip in the Centrino, but then go and cripple it by using the 802.11a standard instead of 802.11g. I know it must chap them to see how Apple has driven this market, but this has to take the cake. 802.11g is now the new standard. 802.11a equip is quite costly and incompatible with the b specification.
I guess it is hard for Intel to face the fact that they no longer drive the computer hardware market. They can join the ranks of Microsoft who also have nearly lost control of the software market.
If either Intel or Microsoft feel that DRM is the way to regain their control of the market, they are sorely mistaken. Such moves would accelerate the adoption of alternative platforms.
As fast as Apple, IBM and Linux are moving, they are going to complicate the future software and hardware plans of Microsoft, Intel and AMD.
I am still amazed by the fact that both Intel and Microsoft passed up the chance to purchase the likes of NeXT, Be, Apple, and Palm. Back in 1997, Palm may have been out of the question, but the other three were ripe for the picking.
Both Intel and Microsoft have a mortal enemy. Unfortunately, they were so blinded by their excesses. This time, however, IBM is going to haunt both of them.
I know both are big companies and have billions in the bank. IBM is yet a bigger player than either of them. Given IBM's presence, I'm not at all sure that either Microsoft or Intel will maintain any long term viability.
Both Intel and Microsoft need to make a major paradigm shift. Yet, they can't. Apple and IBM are driving the paradigm shift and offering true alternatives.
At this point, Wintel cannot compete in offering an alternative to the x86/Win32 market. They must figure out a way to advance x86/Win32. This is the only way to maintain dominance by leveraging legacy investment.
Intel has some smart people. If they see x86 as a dead end, why are the majority of tech authors clamoring for AMD's solution? The way I see it, AMD's chip only offers the opportunity for 64 bit memory addressing. It really doesn't improve on what is available as far as what currently existing 32 bit x86 processors offer for Win32.
Intel's refusal to budge from IA 64 is very telling folks. x86 is hitting a performance wall. Tacking on 64 bit capabilities isn't a very compelling feature for what the market currently desires.
People are wanting more overall performance. Not increases in performance for a few highly specialized 64 bit applications. In order to get it, the market will have to migrate off of x86 and onto an alternative. Intel desires to blaze the way with its IA 64 chip. However, they are failing.
Apple and IBM are blazing the way. It won't be long before Apple and IBM have commanding control. The question is whether they continue to move computing forward or become complacent and rest on their laurels like the current state of Wintel.
Only time will tell. The only thing that I am sure of is that 32 bit x86 computing will be coming to an end. I don't care how much legacy there is, the installed base or commoditization of the market. It is at a dead end. Even AMD's 64 bit chips have virtually no advantage to current 32 bit x86 chips in 32 bit computing.
Contrast this to the G5 in running 32 bit PowerPC code compared to the G4.
AMD's plan to move the market further with x86 64 will be an abject failure. There is no way they can keep up with what Apple and IBM will bring to BOTH 32 AND 64 bit PPC computing. Intel's plan to move the market to IA 64 is complicated by the slowness of their 32 bit emulation environment. To make matters worse, there is virtually no development work for native IA 64 applications.
Intel is powerless to advance IA 64 while Apple writes compelling PowerPC appications that are driving the adoption of the platform. Intel does have its billions. So why can't they hire a team of software engineers to bring a Final Cut Pro or even iTunes to the Itanium? They can, but are afraid to sink money into a project that would be extremely risky. The way I see it, they have nothing to lose after investing millions into the chip's development already.
Why on earth does HP stand by idly? They were the ones that got Intel into this mess with EPIC in the first place. HP could have bought up Eazel. They could add immensely to KDE or gnome development. They could have developed their own GUI for Linux, Tru64, or even HP/UX. Such moves would have been a wiser investment of their funds than the purchase of Compaq. Now all they can do is imitate Dell, IBM or Apple.
I can say one thing. Imitators are never as good as the real thing.
Wasn't it HP that turned down Jobs' and Wozniak's idea of the personal computer in the first place? HP may make decent printers but are incompetent in every other way and it shows. A company of their size should be doing more to aid Intel in the advancement of IA 64.
It will be interesting whether Microsoft will be able to leverage their dominance of the 32 bit x86 market and bring that over to the PowerPC. It would seem that the 970 and 980 chips might emulate x86-32 "good enough". How Microsoft takes advantage of it, however, remains to be seen.
The only way that this might work is if MS ports Longhorn to the PowerPC. However, Microsoft would come to a level playing field. x86-32 would be emulated whether using Longhorn or OS X. The OS itself and 64 bit applications would then become the primary reasons for adopting the specific platforms. I am not sure Microsoft would fare very well. One thing is for certain, they have their work cut out.
The longer MS, Intel, and HP delay in putting together a decent IA 64 alternative, the more they lose the market. They have to ignore AMD and forge ahead.
But hey, it doesn't bother me. If MS wants to fund a port of Longhorn to the Opteron instead of developing compelling seftware appications to drive the adoption of IA 64, so be it.
I never thought it would take this long for the PowerPC to truly exert its dominance. I remember the heady days of the 601, 603 and 604 processors. I remember the G3's dominance over the Pentium II. Were it not for Motorola's incompetence, the PowerPC should not have lost its lead. However, in the hands of IBM, there is now no looking back.
Finally, the x86 is close to being put out to pasture. Again, the fact the Intel is closing the book on the x86 should be very telling folks. It's time to forget the AMD hype. Computers willing be migrating off of x86 very soon.
OSX, Linux and the PowerPC are the only real compelling alternatives there are for the time being. Unless, MS, Intel, HP and Dell produce a viable alternative, they will watch the market slip through their fingers.
Does anyone know if there is any chance of a manufacturer resurrecting the alpha chip? Could MS bring Longhorn over to the chip in a reasonable time frame?
One has to wonder if the purchase of VPC has deeper implications. Especially with the next XBox moving over to the PowerPC. Perhaps MS realizes that porting NT or Longhorn to PPC would be just too difficult and needs the emulation environment to make things work. Only time will tell.
Perhaps this will also for once and for all put an end to the OS X on x86 rumors.
Its because they have no idea where computers are going, hardware wise, and software wise.
Originally posted by herbivore
I remember the heady days of the 601, 603 and 604 processors. I remember the G3's dominance over the Pentium II.
And I remember Apple losing money hand over fist during that time period.
Let's face it, the general public neither understands nor cares about this stuff. Intel could come out with something that had the performance of a Mac Plus and they would buy it because Intel is big, and Intel is what everyone else uses.
During the days of the 604e, when the PPC was kicking the Pentium's ass to a degree that was almost humorous, PC people were still saying "Macs are too ssssslllllooooowwwww" because that's what their preconceptions were. Today, people are still saying Macs crash all the time, even though OS X has been out for 3 years. These things aren't going to change, because there's still going to be schools with Performa 6200's running 9.0.4 that are slow and do crash a lot, and that's the only introduction that many people get to Macs.
Actual quality doesn't matter. What matters is market share and the public opinion, and when people make up their minds about something, it's near impossible to get them to reconsider their opinions. No matter what god machines Apple comes up with in the future, it's not going to make a whit of difference to the average user from when they were still only making G4 machines.
Originally posted by CharlesS
--snip--
because there's still going to be schools with Performa 6200's running 9.0.4 that are slow and do crash a lot, and that's the only introduction that many people get to Macs.
--snip--
I own a 6200CD and yes it is sssllllooowww... even when running OS7.5.