Apple and AMD Hammer

[MisterMe]

Originally posted by jettredmont


No. Marklar, if it is ever released, will be on Apple-only hardware (tied to Apple MoBo's and ROMs), is what bryank1 is saying. That seems to be the consensus view on the matter.

....

Concensus view of whom? The entire history of NeXTSTEP, OpenSTEP, and later Apple's adoption of NeXT run counter to your consensus. Apple has abandoned its proprietary ToolBox ROM for economic reasons. By going with the open standard Open Firmware, which it codeveloped, Apple reduced its costs and time-to-market. A return to a proprietary ROM would mean that each new hardware revision would require a new version of the ROM. Each new software revision would require extensive testing on each version of the ROM. That means higher costs and lower profits.

Now for the notion of a proprietary Intel or AMD-based motherboard from Apple. This is just plain silly. The reason that Intel systems are cheaper is that everyone and his Aunt Edna buys interchangeable components--including software--from interchangeable vendors. Microsoft makes out big, but everyone else just gets by. Some vendors sell higher quality components than others, but not enough to change the business model of the industry. The point is that a lot of players are sharing the development costs--economies of scale.

If Apple were to develop its own proprietary Intel-based system, it would not enjoy the economies of scale of its Windows- and Linux-based competitors. To build a system of the same quality as current Apple systems, Apple would actually have to sell it at a higer price than its current PPC-based systems. Remember the power consumption and heat production of Intel processors pose a significant engineering challenge that Apple will have to bear alone.

You forget another important point. When NeXT ported OpenSTEP to Intel processors, it supported a subset of high-quality standard Intel-based PCs. When Apple shipped Rhapsody to developers, it ran on standard Intel-based PCs. Apple and its NeXT veterans have experience developing an OS that will run a standard Intel-based PC. Porting MacOS X to proprietary Intel-based Macs would mean that Apple would have to sacrifice its experience on the standard platform. The company would literally be starting over with a new hardware platform. That increases costs and time-to-market.

If Apple ports MacOS X to Intel, it will port it to machines that are interchangeable with Windows machines.

But, it ain't happening.

 

[Hattig]

Apple will not be using any x86 processor or variant in the near future.

What Apple will be using is HyperTransport, the AMD interconnect bus as used on the Hammer line of processors, and many chipsets that are coupled with the Hammer.

Apple will rebrand the bus, and this will be the 900MHz interconnect that the IBM 970 has 4 of on the processor (to allow scaling up to 16-way).

It is possible that Apple machines based upon the IBM 970 processor will utilise the same chipsets that Hammer uses, perhaps with some modifications for what Apple want in their system. For example, Apple might also be using the AMD 8131 PCI-X HyperTransport bridge in upcoming IBM 970 based XServes, or the AMD 8151 AGP8x bridge...

However this is all speculation.

 

[ffakr]

Originally posted by bryank1


Let's see... $600 for a Power PC chip or $200 for a top of the line AMD chip. No, saving $400 wouldn't be at all an interest to Apple.

First off... You are talking about volume prices for chips that don't exist yet. PPCs are currently expensive because of volume an because a good number of the chips that Motorola produces are junk. If you have a 50% failure rate, your price automatically doubles.

The problem is that if Apple were to make a Hammer based system, they would either make it proprietary to protect OS X and their hardware sales or they would use standard components to save cost. If you go proprietary, this means extra R&D.. extra Cost. There goes your price advantage.

If Apple makes an x86-64 machine that used off the shelf parts, the price advantage would screw the PPC over (and everyone with a PPC investment). It would also screw Apple over because long time PC makers will be able to beat apple on price anyway (as will home builders). You'll see Marklar burned and swapped on the net and Apple will not only loose hardware sales (and ripe margins from designing a lot of the hardware components), but they will also loose money on software licenses.

Right now, even if someone steals OS X for their Blue and White... At least Apple made money off that user when they bought the computer. If Apple makes a version of OS X for off the shelf components, then there will be switchers who don't buy anything from Apple.

Granted this has benefit... a larger user base drives development for the OS (hopefully for both architectures). It has a real negative side too though. There are lots of hard core Mac people like me out there that have no problems buying a PC. I have an athlon at home because I can play games on it, and because I got a smoking machine for $800. I'll eventually buy another Mac (my last non-work machine was a 6100!) but If I could run OS X at home on my PC, there'd be a hell of a lot more temptation to spend $200 every year to upgrade the PC that there would be to spend a couple grand every few years on a Mac. Even it it only ran on x86-64 (it was 64 bit clean), I'd be a lot more tempted to build an $800 Hammer than to spend a couple grand on a new Mac (it'd be a purely financial decision.. well, I'd be able to boot windows and play games too...)

 

[ffakr]

You forget another important point. When NeXT ported OpenSTEP to Intel processors, it supported a subset of high-quality standard Intel-based PCs. When Apple shipped Rhapsody to developers, it ran on standard Intel-based PCs. [/B]

another point I wanted to make ties nicely into this.
I had Rhapsody for Windows. I couldn't get it to boot on a pretty standard PC board because the VIA IDE controller wasn't supported. We had to put a 2940UW into it and a SCSI drive to get it to install.

if Apple doesn't support a very narrow, or intentionally proprietary set of hardware, they will have to do an enormous amount of driver work. You can bet vendors won't bother to learn how to write OS X device drivers until there is a compelling reason to... which won't appear until there are enough drivers so a lot of people can install it. Catch-22.

Apple doesn't have the resources to support a wide range of PC hardware... or even a reasonable variety of hardware.

 

[Booga]

x86-64

If this bombshell is dropped, I think it will happen at WWDC in May. Apple cannot release a x86-based machine without developer support, and there is no better place to make their case than the WWDC. Technically, it is not a difficult proposition for everything to be ported to x86-64 or any other ISA. If Apple provided a 100% compatible API, and simply gave Project Builder a switch for fat binary production, it would be pretty simple for third parties to port. It could definitely be set up for a Jan 2004 release, if not before.

I, for one, am all for it. When the best case scenario is a chip to be available in a year that doesn't quite match the competition's chips today, you have to worry. People keep talking about Marklar being maintained in case something terrible happens with the PowerPC. Well, it's happened. The PowerPC has really, really bad performance compared to x86 chips today. The sooner people admit that, the sooner we can move on.

There are no huge unknowns to dissipating the heat from these things. It's a $15 part from Fry's. And my PC is quieter than the latest "wind tunnel" Macs anyway. You won't see any nifty fanless models, but c'est la vie. On the other hand, MacOS X will scream on one of these motherboard/CPUs. I would love to be able to buy any Mac that matched the speed of my PC, let alone have it cost a reasonable amount.

How much would a 100% performance boost be worth to you? If Apple could offer it to you at that price, then why not? It's all about how easy it is for the developers and users-- if they can make it easy enough, they can make the transition cheap enough.

 

[nighthawk]

It would not be as hard for Apple to add support for an X86 architechure based machine (Apple branded or not). For example, the WINE project...

Wine is an implementation of the Windows Win32 and Win16 APIs on top of X and Unix. Think of Wine as a Windows compatibility layer. Wine provides both a development toolkit (Winelib) for porting Windows sources to Unix and a program loader, allowing many unmodified Windows 3.x/95/98/ME/NT/W2K/XP binaries to run under Intel Unixes. Wine works on most popular Intel Unixes, including Linux, FreeBSD, and Solaris.

Wine does not require Microsoft Windows, as it is a completely alternative implementation consisting of 100% Microsoft-free code, but it can optionally use native system DLLs if they are available. Wine comes with complete sources, documentation and examples and is freely redistributable. (The licensing terms are the GNU Lesser General Public License.)

source - [url]http://www.winehq.com/about/[/URL]

WINE would allow most of the current Windows software to run on OSX-X86, and Apple's experience and development resources would really benefit the WINE group in their mutual goals.

The second point is that nearly every pure Cocoa Application (I would say about 90%) would be able to be recompiled and run on OSX-X86. This was the intention from the beginning anyways with NeXT. In fact, Apple promised this years ago and never delivered (one framework library to run on Apple and NT machines).

Now, my opinion on the hardware -- I do not believe that Apple would ever develop X86 based hardware unless it co-existed with PowerPC hardware (like the 6100/66/PC computer 8 years ago).

I work with macs all-day everyday in the graphics and printing industry. I also work on PCs (to support our customers with PCs). While we do not have the top of the line machines, they are not older than a year. I work with large and complicated files, Postscript rips and image processing in Photoshop. There is a difference between the Macs and P4 Dell machine, but they are well balanced. When it comes to working on large files in Photoshop, they are equal -- and a lot of that has to do with limitations with the HD access. For whatever weird reason, the P4 is about 5x faster than the Mac using Acrobat Distiller 5. However, for production purposes, the Mac is equal to the P4 even though it is half the clock speed (and only single CPU).

Apple takes a lot of heat for having slower processors, but that doesn't stop people from buying them. And it is not just the old faithful, but all of the new switchers.

 

[Token]

Regarding the argument that making Marklar run on a 'standard' pc config would make Apple loose on hardware sales:

I think this only partly correct- Indeed, releasing Marklar for Intel/AMD will sell fewer copies of Macos X. But, first of all, Apple will probably, due to lower deveopment costs, make the roughly same profit from one copy of Marklar as from one iMac/Powermac. Second, they are more and more becoming a software & service provider: they charge for quicktime Pro &mpeg2, possibly for upgrades to iApps, for .Mac and for AppleCare. So, even if lots of people will use cracked copies of Marklar, it will only be part of the 'product' they steal. Third, cross platform digital devices like the iPod is another way to distribute sales to other products than computers.

Apple will become much less vulnerable to diminishing computer sales... And they could go on producing great looking computers (like the iMac/iBooks) that would sell very well.

 

[Phil Of Mac]

I think we're all forgetting something.

AMD's chips aren't x86 chips. They're AMD chips with an x86 frontend. If they can make a PPC frontend for the Hammer, none of these issues would occur, and the Mac world would continue as it is, except with faster processors

 

[alex_ant]

Originally posted by bryank1


Let's see... $600 for a Power PC chip or $200 for a top of the line AMD chip. No, saving $400 wouldn't be at all an interest to Apple.

No PPC chip costs half that.

 

[scaniasaab]

RE: AMD and OSX "Marklar"

please comment on the viability of the following scenario:

Apple and AMD should get in bed and market various x86-64 based SERVERS for SPECIFIC AND WELL-DEFINED MARKETS such as

(a) Entry and Mid-range Corporate Servers running "Marklar" on a supped AMD HAMMER Opteron
and
(b) catch this "HOME" Servers which will be nothing more than a supped up AMD Athlon 64 workstation

With an x86-64 "HOME" server packaged in a blade-type (1U) case ideally, Apple could entice PC users to switch to an Apple product through implementing an aggressive pricing structure for customers purchasing additional PPC-based "CLIENT" workstations while at the same time providing an easy migration path OUT of Windows eventually when hell freezes over. The key for the Apple marketing dept. is to figure out a way to attract either potential users to buy a "HOME" server to replace all of their existing x86 hardware AND to ultimately get them to follow up with purchases of PPC-based "CLIENT" workstations in the form of PowerMac Desktops or low end iBooks and everything in between those.

And you may ask how is that going to work... well here's the answer:

VNC (Virtual Network Computing) pioneered by AT&T R&D at Cambridge U.
VNC is freeware through GPL.

For applications not available in a PPC binary all you need to do is install and launch them off of your Apple branded x86-64 "HOME" server running the latest version of your garden variety Windows XP or whatever. There wouldn't be any ridiculous licensing fees from MS because Apple only needs to obtain a single OEM license per "HOME" server. The "HOME" server will act as a server for all connected PPC clients but will in reality only be simply another
x86-64 WORKSTATION. All of those PPC clients will of course be running OS X. That's the beauty of VNC's platform independence.

Furthermore, because of the flexibility of the VNC protocol multiple sessions can be launched via the "HOUSEHOLD's" Gigabit Ethernet or whatever ethernet provides the greatest bandwidth as a LAN.

Basically what I'm suggesting is that Apple can provide its current and potential users seamless access to a plethora of applications that exist in the x86 world of Windows which may never be recoded in a PPC binary for recompilation into OS X. Applications which Mac users need to use on a daily basis and just cannot function without for the sake of efficiency in the workplace/classroom/family room.

I stronly believe that what I've outlined above can be interwoven easily into Apple's Digital Lifestyle/HUB metaphor. Also, I truly believe that emulation technology is an inadequate dead end at best for wasting precious computing cycles. Why even bother with emulation or reinvent the Win API "wheel" with WINE when the solution is in VNC.

My logic argues for many assumptions to be taken for granted. First, Apple's business model needs to be refined and fine tuned with just as much innovation and precision as they are willing to put into their products. Second, Apple bigots must accept the fact that we do live in a world where x86 and Windows are firmly entrenched and fully proliferated. Nothing is going to change that I believe. Third, Mac users must coexist with their Windows brethen oops I meant heathen no matter how painful a process. What I've outlined is an innovative painkiller to beat MS and Intel at their own game.

For those of you who are confused by my logic please forgive an economics major's digression into the market dynamics and human behavior.

SAABS rule!!!
--scaniasaab

 

[lmalave]

Re: RE: AMD and OSX "Marklar"

Originally posted by scaniasaab
There wouldn't be any ridiculous licensing fees from MS because Apple only needs to obtain a single OEM license per "HOME" server. The "HOME" server will act as a server for all connected PPC clients but will in reality only be simply another
x86-64 WORKSTATION. All of those PPC clients will of course be running OS X. That's the beauty of VNC's platform independence.

While only one copy of Windows would be needed in this scenario, keep in mind that many license agreements for commercial PC software specifically specify that each separate "end user" needs to have a separate licence (specifically to cover PCAnywhere/Terminal Services/VNC type situations), so technically you'd be violating the license under any scenario that let more than one user use the software simultaneously (not that that's ever stopped anyone).

But the biggest problem is: I'm pretty sure that VNC is NOT multi-user, at least not on Windows. That is to say, even if multiple clients connected, they would all be connecting to and controlling the same desktop. So it would strictly be one user at a time (which I guess is the same thing as if the users had one shared PC machine that they actually had to sit down in front of).

And also, this home server had better be cheap! VNC is already available for free and actually quite easy to install and use on both the server and client side (I use it on a daily basis), and basic PC hardware is dirt cheap, so the only value that Apple will provide is packaging and marketing (unless they actually write their own VNC client that better than any of the freely available ones out there like - I use VNCThing). Also I believe that VNC is purely a display technology - so no sound, I believe.

I do agree this is a much higher performance solution than Virtual PC, though (albeit without sound, which VPC actually does, I think...) With a fast LAN connection VNC is actually very snappy. Since Virtual PC costs $300, if Apple could design a super low-budget (but still stylish on the outside) box for no more than $400 or $500 (trailing-edge AMD processor, 10 or 20 GB drive, 128MB RAM (2 DIMMS ), CD-ROM, integrated LAN, no monitor). I think there would be some market for such a product, but I myself am already building a cheap $175 PC (using 3 new parts but mostly parts scavenged from old machines).

 

[Phil Of Mac]

So you're saying people are going to purchase seperate client and server machines for their own homes? What the hell would they gain that way? More hardware cost? If they already have Windows machines, they don't have to buy a new Windows machine from Apple to run their old programs on.

Besides, you're all still stuck on this x86/Marklar fantasy, taking it as accepted fact when it's really just a rumor. Even if Apple keeps an OS X build on x86, it's probably just for research purposes and as a last resort, not because they have any imminent plans.

Any cooperation with AMD (in terms of processors) will be contingent upon AMD's ability to make a PPC, or adapt their current chips to a PPC instruction set. Switching processor platforms would be unspeakably stupid of Apple for the next few years at least, and making a "Home Server" would be a complete waste of good R&D money.

 

[lmalave]

Originally posted by Phil Of Mac
So you're saying people are going to purchase seperate client and server machines for their own homes? What the hell would they gain that way? More hardware cost? If they already have Windows machines, they don't have to buy a new Windows machine from Apple to run their old programs on.

Besides, you're all still stuck on this x86/Marklar fantasy, taking it as accepted fact when it's really just a rumor. Even if Apple keeps an OS X build on x86, it's probably just for research purposes and as a last resort, not because they have any imminent plans.

Any cooperation with AMD (in terms of processors) will be contingent upon AMD's ability to make a PPC, or adapt their current chips to a PPC instruction set. Switching processor platforms would be unspeakably stupid of Apple for the next few years at least, and making a "Home Server" would be a complete waste of good R&D money.

Well, with VNC it doesn't really make sense since it's really single-session, but if there was something like an OS X client for Windows Terminal Services that allowed multiple independent sessions on the same PC, then I could see it working for workgroup settings (or even large households with multiple computers). If someone needs to use a Windows program, they just connect with the Terminal Services client (which had their session saved already, so it'd be instantaneous), and they have a much snappier interface than if they were using Virtual PC.

 

[Phil Of Mac]

Now that situation makes sense. I could see that being done. At home, for one or two users? No.

And it would be better than VPC.

 

[lmalave]

Originally posted by Phil Of Mac
Now that situation makes sense. I could see that being done. At home, for one or two users? No.

And it would be better than VPC.

Yeah, if Apple built such an appliance I wonder if Microsoft would crack down on it, though, making sure that Apple couldn't get Windows licenses to install on these machines....

 

[Phil Of Mac]

MS sells licenses for VPC *shrugs*

(I have VPC, which is how I know.)

 

[lmalave]

Originally posted by Phil Of Mac
MS sells licenses for VPC *shrugs*

(I have VPC, which is how I know.)

Right, because VPC is "single user" (one per machine), and in order to use VPC you would need to buy a copy of Windows. So as far as Microsoft is concerned VPC is just like another real PC - it's just another installation of Windows.

Once you get into "sharing" a windows box among multiple users, that's when I think Microsoft might crack down...

 

[tkuechle]

Apple outsourcing?

Maybe, Apple has developed a great PPC processor design for us Mac computer users and AMD is going to FAB it. It could be some new fangled PPC chip that is cheap, fast and has low power consumption. There are probably many reasons why this would not happen, but it is one scenario that i have not seen mentioned yet.
Apple can design its own chips and get the new design made by some fab other than IBM or Moto, so this part seems possible.
If AMD can make PPC chips that are wicked fast and cheaper than Moto or IBM can produce in volume, then we would benefit greatly from this arragement as would AMD to some extent. It might help Apple that AMD is just up the road and doesn't have some bone to pick with Apples closing down the Clone Licensing thing some years ago. I have done some light reading on the AMD and G4 cpus. They have many similarities, beside the fact that they use the same materials and very similar processes. I haven't read anything yet about his scenario coming to light but it can be added to the rumor mill.
Enjoy.