Tuesday, November 28, 2006

Roads Not Taken

Lately Apple and Intel both seem to be doing fairly well. However both of these companies have overlooked potential opportunities. Choices like these will eventually make the difference between success and failure.

Apple has been quite prominent on tv lately with their coffee house neo hippie Mac against the dull and untalented PC. It's cute I suppose but it certainly doesn't show Apple's reality. Apple has been boosted tremendously by both its iPod and iTunes website. Apple managed to get these into the market when no one else was doing it and has been successful. Unfortunately, both of these are doomed to follow in the footsteps of Sony's Walkman and become commodity consumer items. To put it simply, neither iPod nor iTunes can stand up to real competition. It is extremely unlikely that Apple will be able to come up with another consumer success like the iPod. One wonders how Apple will fair when its computers have to stand on their own against increasing competition.

Apple's big problem is that it is heavy in desktop systems but light in servers. Apple's natural ally would have been Sun. Sun is very heavy in servers but much lighter in desktop systems. Sun/Mac would have been a perfect match with Apple providing the experience and hardware for desktop/client and Sun providing the muscle and support for workstation/server. With MacOS now based on unix this could have worked well with a merger of sorts between MacOs and Solaris. This would have been an ideal situation for Apple where its Macs fitted seamlessly into heavy duty Sun backend server environments.

In terms of software, a Sun alliance would be even more profound. In spite of the hokey antics on Apple's commercial, the PC is not the Mac's true competitor. Apple's true competitor is a company that Apple has no chance of beating. This company is Microsoft. Microsoft is all too aware that Macs don't run Windows. However, Apple is all too aware that Macs are tied to Microsoft applications. This is why Apple does not and will not sell a version of MacOs for PC's. The retaliation from Microsoft would be epic and would probably include suspension of any further upgrades or releases for Mac versions of Microsoft Office. This is not something that Apple can afford.

However, an alliance with Sun would include Sun's Star Office suite as well as a major contributer to Open Office. With much less pain, Apple could have removed itself completely from Microsoft's control and opened up a true market for its products including MacOS for PC. That this did not happen could be due to a lack of vision among Apple executives or perhaps Apple just didn't like sharing the spotlight with Sun. Of course it could be that Sun was not open to such an alliance. Whatever the reason, this alliance was pretty much Apple's only ticket into the future. Alliances with HP, Dell, and Gateway are out of the question as they are direct competitors. Apple's deal with Intel now makes a Sun alliance nearly impossible along with any deal with Cray. And, Apple has already dropped IBM. The only name left that I can see is SGI. This wouldn't be as good as a Sun deal but with a bit more effort I could see it working. I have serious doubts though that the upper management at Apple can see this. But, without it, I would say that Apple's days are numbered as a desktop player. The desktop will continue to become more competitive as will mobile. Without a heavy share in servers Apple will eventually become marginalized. If this idea comes as a shock to anyone who really likes Macs then just consider the same goofy commercial if support were discussed.

Hello. I'm a PC and I'm supported by nearly all of the top electronics manufacturers and system builders.

I'm a Mac and I'm supported by, well . . . uh, just Apple.



Intel had a similar blunder with Transmeta. In fact, it is staggering to think that out of all of the losing businesses that Intel picked up like lint on a sticky lollipop they failed to see the perfect match. The Transmeta Crusoe was designed as a VLIW processor. It is true that this technology has been slow but this is primarily because the software layer translates X86 instructions into very disimilar VLIW instructions. However, this would not have been the case with Itanium. Crusoe was very much like a stripped down version of Itanium and and could have executed most of the Itanium instructions with microcode and only used the software layer for more complex translations. In 2000 the Transmeta company could have been purchased by Intel for what would have amounted to spare change. This would have given Intel an inexpensive, low power processor that was completely compatible with Itanium.

Intel could have used this technology immediately for embedded processors, palmtops, and thin and light notebooks. With a bit of effort I'm certain Intel could have beefed up the Crusoe and built a very nice second generation chip that would have been suitable as a general mobile processor. This could have allowed Intel to move to dual core earlier for more power. The design would have been good because it would have been able to run X86 better than the primative translator on Itanium but would have run much faster with native Itanium code. This technology would have meshed very well with Microsoft's .NET environment for even better performance. Proof that this could have been done is evident from the fact that AMD did this very thing when it bought Geode in 2003.

This foothold at the bottom of the market would have made it much easier for Itanium to hold its ground. It is even possible that by using the Transmeta technology in Itanium that Itanium could have run faster with X86 code. By almost any standard this would have been an inexpensive purchase that could have yielded substantial benefits for Intel. Or, in other words, it was a good risk. I still have no idea why Intel didn't take advantage of this opportunity. The almost certain result will be that Itanium will continue to be pushed out of markets by Opteron and by newer C2D based Xeons and will be pushed out of the market altogether in the next few years.

23 comments:

Anonymous said...

Analysts have been demising the iPod's doom for as long as it's existed, citing competing hardware and software incompatibility. 5 years and still kickin'. What is 'real' competition? Zune tried..:P I can see Apple pulling another huge success with iPhone but then we really don't know much about it:D

..So what advantage would Sun have over Intel in 'muscle' and support? Solaris+Mac OS? >.< Hasn't there enough pain from OS 9 to OS X?

OpenOffice works fine for me on XP. I can't think of a single Microsoft app necessary for Apple. Sorry, I see no reason why Apple would go with Sun:/

I don't see how not playing in the server market now as opposed to not playing much in the past is going to affect future desktop market? Please connect the two:)

I'm not sure where you're going with Apple being just supported by Apple. It's not like Gateway users call the actual manufacturer of their LCD as opposed to Gateway.

http://arstechnica.com/news.ars/post/20060307-6335.html
Longtime Ars readers will probably recall that the combination of VLIW + x86 emulation has been tried before, and fairly unsucessfully. The technical similarities between the Itanium-Transitive deal and Transmeta's Crusoe are hard to escape, but there are some fundamental differences with respect to goals and circumstances that make such a comparison unfair. The key difference is that Crusoe relied on binary translation for everything, whereas Intel will be using it selectively and (they probably hope) phasing it out as Itanium-specific ports become available. The lesson of Transmeta's Crusoe is that binary translation is not a long-term plan from a performance/watt perspective, but the lesson of Apple's Rosetta is that it is a good temporary solution for easing the pain of a platform transition.
That last bit of your article is extremely speculative and would only have worked if Transmeta were to execute. Oh yeah, is Geode executing? You're essentially saying that Intel should've flipped a five sided coin so that Itanium could be magically dominant.

Scientia from AMDZone said...

what advantage would Sun have over Intel in 'muscle' and support?

Intel is a cpu vendor; not a server company. Why are you comparing them with Sun?

Servers are higher margin than desktop. As I said in my article, the competition in the desktop and mobile markets will increase which will make it harder for Apple to remain profitable.

One could make a list of all of the motherboard, graphics, and peripheral makers that support Apple along with all of the software vendors. There simply is not as much support for Apple systems. PC's in contrast are supported by HP, Dell, Gateway, TigerDirect, VoodooPC, Alienware, and eMachines. PC based servers are supported by IBM, Sun, HP, Dell, and Gateway and several others. There is no competition.

The arstechnica article doesn't contradict anything I said. Why did you quote from it?

Again, you don't seem to grasp what I was saying about Transmeta. Transmeta + plus Itanium would have been very good for Intel. And, yes, Geode has been good for AMD and is still working for AMD.

Anonymous said...

I'm comparing them because Apple is depending on Intel as it would depend on Sun if it went down that path.

They're increasing share with $249 PCs everywhere. And Dell and HP have seen the destruction of PC prices and as a result have tried not so much to focus on price, so what is this increasing competition?

I always see Mac support for the tons of printers I buy and I'd assume that's true for most other stuff. Try looking for 3rd party hardware on the Store at apple.com.

I'm confused how you went from Apple to suffer as a result of not being strong in server to resulting in a weak desktop and randomly remind us of Apple's lack of motherboard choice. And what do you mean no competition? >.< Total confusion:/

Anonymous said...

I was under the impression with the cancellation of the PIC and lack of enthusiasm for OLPC..:)What's it been good for?

VLIW, Transmeta, brain fart:/ Surely Intel could've adapted Itaniumized Core or Corized Itanium by themselves for vastly better performance and good wattage compared to Crusoe? But I see no point as Itanium is aimed for a vastly different market than x86, no? I'm sure there is good IP in most companies but that doesn't mean it's good enough to snatch up any company that has some innovation. And considering the way Transmeta has turned out as a patent troller..

So..back to today:) Care to predict any other far fetched partnerships.. only about today:D

Anonymous said...

Thanks Scientia for the very well written and thorough analysis and comments. I enjoy your articles.

Why do you think the Apple/Sun merger couldn't happen now or sometime in the future. I don't see Intel being an obstacle. Wouldn't it be quite easy for Apple to switch to AMD? Since Sun is in the process of making the switch, Apple using AMD would make the merger more attractive.

Also, may I request some more technical articles about AMD's architecture? How it compares to Intel's. Do you think the new AMD architecture due out next year will put AMD on top in the Desktops? Notebooks? or Servers? How do you see AMD competing with Intel in each of these markets?

Thanks again for the informative articles.

Scientia from AMDZone said...

Red, I can't write an entire tutorial for you. I'm sorry if you can't understand what I write about.

Apple needs to be part of a bigger environment to survive in the future. A merger with Sun would have given them that.

However, Sun is in close collaboration with AMD while Apple is now in close collaboration with Intel. So, a merger with Sun is now very unlikely. It seems like SGI would be possible because they too are closely collaborating with Intel and it seems like they too could benefit from a partnership.

These seem like common sense to me both from a point of view of technology and market position so I'm not sure why Red characterizes them as far fetched.

Geode technology is still active for AMD. PIC is now in a different form.

Itanium was originally in workstations and was intended to be a desktop system. Transmeta used to have processors in servers. To suggest that these were vastly different is incorrect. Itanium today has been pushed out of workstations, 2-way and 4-way servers, and mostly out of HPC. However, this was not Intel's original intent.

Scientia from AMDZone said...

Wouldn't it be quite easy for Apple to switch to AMD?

My understanding was that Intel provided programmers to Apple to help them switch to X86. I don't know the terms but I'm certain that Apple has some kind of commitment.

Since Sun is in the process of making the switch,

I don't know what you mean. Sun has everything from single cpu to 8-way servers using Opterons. Sun is building an Opteron based supercomputer.

Yes, I can write more about AMD's archtecture. I see Intel growing but AMD growing faster during 2007 and 2008.

Scientia from AMDZone said...

No, Microsoft wouldn't kill Apple. It would be painful that is true. But, if Apple merged development of MacOs with Solaris and fully embraced Star Office they could weather any retaliation from MS.

Sun is currently going its way without MS and this would be a big help for Apple. SGI could work in a similar fashion although not quite as well as Sun. It would be tough for awhile but once converted Apple could begin offering MacOs in direct competition to Windows which is something that it can't do now.

hyc said...

But what does SGI have to gain from a partnership with Apple? SGI has already ventured into the x86-based workstation market and failed miserably, why would they go there again? These two companies have practically zero relevance to each other - SGI is about HPC and InfiniteStorage, Apple is diametrically opposite. Neither company's reputations or strengths would help them in the other's space. SGI is IRIX (dying) and Linux, there's no opportunity for software synergy here either.

I guess a port from one Unix to another (FreeBSD to Solaris) may not be the most painful transition in the world, but it would still be annoying. How many times can Apple force their users to jump through that hoop, in how short a time? It's not like there's any ABI compatibility between BSD and Solaris to ease the way. So a merge with Sun doesn't make a ton of sense either.

I guess I can see that Apple freeing itself of its dependency on Microsoft (Office) would be a good thing, and having a partner with deeper pockets to weather the transition period would be good too. But neither SGI nor Sun strike me as being the most robust of choices at the moment.

Anonymous said...

1/12/06 9:32 PM
Basically, everything in your post is based on extreme speculation on the past. Lack of good AMD news/bad Intel news?:D Tell me why they need to be into servers to survive when they're still here and PC prices have bottomed. I consider your points far fetched because it'll take quite some time for us to even see if they'll pan out and..it's extreme speculation;) If you are so sure that things would've changed if they were different in the past, why not predict something that actually matters, like something today?

Erlindo, considering that Apple put a mobile chip in their desktop, I'm 100% sure they would not bother with that noisy juice sucking box:D Oh yeah, predictions on the past are totally facts;) Have you no argument than to label me a fanboy?
http://www.extremetech.com/article2/0,1697,2065503,00.asp

bb.. And Scientia, please do write about the architecture:) And not resort to AMD is 'as fast as Intel except SSE';)
http://i17.tinypic.com/2eq51k5.jpg
http://i16.tinypic.com/4d6wjmo.jpg

Again, thanks for repeating yourself:/ Tell me in what ways Geode is active. And if you're gonna say PIC is in a different form, wanna specify?

And what exactly is wrong with OpenOffice besides that they don't have pretty templates and stuff? I think new Office versions are just rehashes anyhow, not like Apple will die if they can't make Powerpoints with a ribbon.

Anonymous said...

don't know what you mean. Sun has everything from single cpu to 8-way servers using Opterons. Sun is building an Opteron based supercomputer.

Note, by mistake my first comment was under a user named bb. Google really screwed up the login for blogger.

I was under the assumption Sun still used Sparc for many of their servers. If so, my comment was implying that Sun is switching from Sparc to AMD64.

Anonymous said...

http://dailytech.com/article.aspx?newsid=3363
I remember this debate some weeks ago about Via's presence..And it seems Intel really wants that 1%:D..So are there any actual numbers of how well Geode is doing for AMD? And a simple search on Google News for OLPC does not look good.

http://www.microsoft.com/presspass/press/2006/jan06/01-10Macworld2006PR.mspx
To avoid getting sued, Microsoft will have to make Mac Office anyhow 'til 2011. And I'm confused about StarOffice..There is mention of OpenOffice Mac on their site, but no StarOffice? But at least if they're working on the Mac version of OO, SO should come eventually..

http://www.techworld.com/opsys/features/index.cfm?FeatureID=1247
83 per cent of graphic designers, 77 per cent of corporate design departments and 65 per cent of advertising agencies rely on Macintosh computers.
If anything, Apple relies on Adobe. I really don't see all the fuss about productivity suites.

http://www.dailytech.com/article.aspx?newsid=4037
Google around for more confirmation. And I'm talking desktop, not workstation.

I don't disagree that Apple in servers would be good but find the notion that they'll be any more 'marginalized' than in the past silly, considering they've been this way forever.

I missed that SGI part. He still said that Apple would still be too silly to see how he sees it. I'd like for him to tell us a company that Apple[which I guess he discusses because they're with Intel, as 180 does with Dell and AMD] or Intel would seriously benefit from. Benefit in a way that supposedly Sun/Transmeta would've saved Apple from future marginalization and Itanium from being pushed out of the market. Or are both companies too stupid?

Anonymous said...

Apple has been boosted tremendously by both its iPod and iTunes website. Apple managed to get these into the market when no one else was doing it and has been successful

You just started your blog with an incorrect statement. the iPod started late into the mp3 player market. But its the iPod that made mp3 players into the mass market conciousness.

Sony's walkman was always cool and hip until their reluctance to embrace the mp3 format where they lost market share to just about everyone else who can play mp3s.

You obviously do not know anything about Mac's user base. The people that buy Mac hate the PC. To that effect Apple has a very loyal user base who always wanted to see their systems to be hip and unique from the messy PC environment. Sorry, Sun isn't hip. and so is AMD. Steve Jobs had to take so much flak from its user base just by switching to Intel because it takes away its uniqueness.
Apple is in a much stronger position that it was years ago boefore the ipod and it never died then. Theres noting Microsoft nor the PC industry can to do crush this computer segment even if they try giving away their PCs for free.

Another incorrect assumption you have is Apples desire to take significant market share from Microsoft. You're mistaking Apple's niche 'hip and cool' market strategy to HP or Dell's mass market strategy. Similar to Versace/Prada againts Gap/Banana Republic target market.
This basically moots the entire point of your blog.

Scientia from AMDZone said...

Roborat, your statements would be interesting if there was any truth to them.

I never said that Walkman disappeared. Walkman went from being the primary portable to just one of many. I can guarantee that the margins on Walkmans dropped.

Again, you are confused about what I said about iPod. No one else had a website for downloads and a player when iPod/iTunes was introduced. These items will lose margin as the field becomes commoditized.

Your view of Apple is also seriously skewed. Apparently you are getting your demographics from Apple's own commercials. In reality, the majority of Mac users are over 50 and couldn't care less about being hip or cool.

The PC market doesn't have to do anything to try to crush Apple. Apple will encounter its own problems as its technology loses margin. It won't happen tomorrow but it will happen.

You are also incorrect about what Apple needs to do avoid being pushed out. Making its operating system available for PC's would be a reasonable move. However, Apple is prevented from doing this by factors beyond its control like Microsoft and its ability to support a wider variety of hardware. Unable is not the same as not wanting to.

Similar to Versace/Prada againts Gap/Banana Republic target market.

This statement all by itself shows just how far off the mark your views about Apple are. I'm quite familiar with Apple's margins since they first started selling the Apple II. I have similar knowledge of the margins that DEC and Data General had.

It is the essence of lunacy to compare the margins for Apple with the margins for Prada. If you really want a comparison with Prada, Alienware and VoodooPC would be closer. Apple is not a boutique manufacturer or vendor by any means. Apple's market depends on a large volume and Apple has no means to step away from maintaining competitive prices.

Wise lnvestor said...

It's a know fact that Apple's GPU's are 2 - 3 cycles behind PC. Now that Intel controls the CPU and chipset. You might see all the ATI lineup disappear.

There are prices to pay to go with Intel. I just hope Steve knows what he's doing...

As a consumer, I would not buy an Ipod unless they include FM recording. But that seen impossible as the likes of SONY and Apple will never take this into consideration. They are NOT consumer centric.

Sony had gone downhill cause they release DRM infected CDs. Secondly ,their network walkman had BIG softwares problems. Connect Player Fiasco

I know this cause I use to own a NWA608V for 4 days.

In a sense, Apple had pick up some shares when other big companies had fail to deliver. It's only natural.

As for Apple's competitor. There are Creative, Iriver, Sony(if they could ever get their ass together!)and ETC... Panasonic could be a player in this. Though I wonder why they show no effort after SHOCKWAVE mp3-CD player...

Anonymous said...

Scientia, could you give some comments about Z-RAM Gen2. How likely is it for AMD to use this technology? 2007? 2008? ever?

http://www.us.design-reuse.com/news/news14802.html

Anonymous said...

Scientia,

This is off topic, but I don't know how else to ask you.

Can you comment here or in a new article about Z-RAM Gen2. How likely is it that AMD will use it in 2007? 2008? ever?

http://www.us.design-reuse.com/news/news14802.html

Thanks

Anonymous said...

Sorry for the double post. Blogger makes new users have a Google account to log in and it messes things up.

Anonymous said...

Hi Scientia

You've probably noticed Toms review of Clovertown :

http://www.tomshardware.com/2006/12/06/quad-core-xeon-clovertown-rolls-into-dp-servers/index.html

I would like to be gentle with this one but I just can't help saying it:
"What a piece of junk!"

And for every one saying that Intel's processors aren't FSB limited this is the ultimate proof that the FSB is the biggest bottleneck there can possibly be for more than 4 cores ...

Compared to AMD's 4x4 launch where Toms suggested not to buy it, amazingly for Clovertown it's says "buy it" with some exceptions ofc but the first thing the recommend is to buy it ... They are the biggest bastards in the whole universe!!! Money driven scoundrels!!!

Anonymous said...

Ofc they don't compare it to an AMD system. The AMD 2P system runs almost head to with 2P Woodcrest and that would have made Clovertown look bad by comparison

Anonymous said...

I haven't seen anyone say that the FSB scales all the way to the sky.. No inflammatory comment about people making up arguments on the other side to start their own:D

The 5160 has a 12.8% clock lead over the X5355. With the exception of the MainConcept bench, the quad is not exactly lagging behind. I seriously doubt that it is the FSB's fault that there were no gains on consumer apps from two more cores. With Octo FX coming up, I suppose AMD will push for mega multicore support in apps:)

Quad FX vs QX6700
Quad FX is slower and hotter.

Clovertown vs Others
They say it is a good buy if it fits your needs. And there are certain situations where Clovertown shines.
http://www.2cpu.com/review.php?id=114&page=1
Here, they compare price to price and bench it in not silly stuff like DivX and 4x2.33 beats 2x3.0 a majority of the times.

Anonymous said...

"Quad FX vs QX6700"
Quote: Quad FX is slower and hotter.

1. QX6700 is 65nm process while Quad FX (FX-70 series) is 90nm process. In terms of leakage (forgot the term... sorry), 65nm is better than 90nm.
2. Quad FX is based on a three-year-old microarchitecture and Core 2 Quad is not (one-year-old only, oh I forgot it is variant of Pentium M.... sorry again). Core 2 have wider SIMD units and Athlon FX don't have it, and is of course slower.
3. Two FX-70 CPUs of TDP 120W compared to one ~80W Core 2 Quad QX6700 is of course hotter (and consumes more power).

Please forgive me for all those nonsense talks, and have a nice day.

ck

Anonymous said...

Pop Catalin said that some say that the FSB is unlimited which was what I responded too.

I agree that Clovertown review was not good.. It pains me to admit with Sir Grumps but I have lost my trust in THG recently:p
http://i16.tinypic.com/4d6wjmo.jpg
But here is a bench straight from AMD. Is that real world enough?:)

Last time I checked, 4P is a lot more expensive. According to AMD's price list, the slowest HE 4P is as expensive as the fastest 2P.