Rumored Timeline: G4, 970 and Beyond

[pudrik]

If Apple is able to beat Intel to the 800MHz FSB, it would be a huge marketing windfall, even for the uneducated crowd. (Yeah, their processor runs faster, but our whole machine runs faster.)

Personally, though, I will always have a PC desktop for gaming, but I need a laptop for work/school, so I especially hope the 970 Powerbook rumors are true, although they will probably have a slower FSB.

Apple seems to finally be putting it together... The agony is having to wait for it to happen.

 

[IVIIVI4ck3y27]

Re: What if..

Originally posted by ryanweb
Remember how everybody expected Chimera to become the new Apple browser, and then, boom, we've got Safari based on KDE?
Why everybody is so sure that Apple will even use PPC 970?
Why not Opteron from AMD? I've read an interview with head of AMD and he sounded very confident about this new chip, like he already have a very big customer to sell it to.
Just a thought.

Simple.

Because Apple is "JUST NOW" getting things ported to PowerPC using a myriad of Carbon/Cocoa. Cocoa = very portable. Carbon = "pretty portable", but in that same sense... look at how long it's taking Quark to get to PPC via Carbon in OS X?

Moving to AMD would require "EVERYONE" to rewrite. Trust me... I doubt Adobe would jump at the chance, Microsoft didn't even want to support Cocoa, and supported Carbon because "It was supposed to be easy", but judging by Macromedia, Adobe, and ultimately Quark taking forever to get to market with their applications... it's not as easy as it would seem...

PPC 970 = backwards compatible with other PowerPC's. That means ::gasping:: Doesn't have to rewrite the OS, doesn't have to make any changes to applications to work (and eventually you can optimize with the next generation version of your apps. to take full advantage, and... ::gasping:: Panther could take advantage of the 64-bit nature of the chipset too through optimization), and can focus on making the machine work better on the myriad of PPC's, from G3-G4-970. It's not the processor speed... it's the inefficiency of the OS. Also known as "bloat". If Apple can increase the threadedness of OS X sufficiently... a low-end G3 would be more than adequate to be responsive and productive. 970's would scream.

Oh, and if Apple wasn't going IBM... why would IBM even bother to include an AltiVec compatible SIMD? That was one of the major reasons they balked on the G4 roadmap according to Motorola, sold the Somerset manufacturing plant to Motorola, and was set to focus on their own needs. Apple wanted SIMD, IBM had little reason to develop it (Linux and AIX don't support it, and there's little logic to support it if you understand that the beauty of Unix OS's that are based on a monolithic kernal, is that a simple port of a POSIX application is easier when you're not writing for all sorts of specific hardware to tie it to; consider that IBM's PowerPC and Power# efforts are tied to Enterprise where it's not a graphics intensive area where vector-processing is a major focus... and you'll understand ever more that SIMD in a chipset isn't a priority to "IBM" for "IBM's Machines") and didn't want SIMD because their machines "DIDN'T" need it. Therefore, IBM continued to sell uprated G3's to Apple, while developing "NEW" G3 models to suit their needs that were faster in the areas they needed for Enterprise. Many of which were 64-bit and suitable for their own Enterprise needs.

Yet in the process their revenues from selling PowerPC's shrank, and the R&D costs increased because they weren't getting bucketloads of $ from Apple, nor were they focusing on the more lucrative PowerPC embedded market that Motorola spends 99% of their PowerPC development on. BTW, embedded "DOESN'T" include PowerPC desktops, where it's quite obvious Motorola simply doesn't give a damn, and only improves things when they feel like it.

Fact is, after Apple killed off the clone industry, and put a "permanently parked" sticker on Motorolas G3 PowerPC Platform (PPCP; previously known as CHRP for Common Hardware Reference Platform; an open-design spec for hardware that multiple OS's including MS Windows NT supported) PowerMacs. Ever since then... Motorola has been bitter. My ex-girlfriends (still friends) mom worked at Motorola. It wasn't very long after that, that Motorola sold off or gave away "EVERY" single piece of Mac hardware they had, switching over to Windows-based PC's.

So for those wanting AMD... not unless AMD builds PowerPC's with SIMD that make the 970 look sick and can run Jaguar with some minor patches/updates/tweaks.

🙂

 

[IVIIVI4ck3y27]

As far as the PPC 970 being suitable for laptop duty... actually it's between the current micron process high end and the low end G4 in heat. Since the current "Low end" is considered hot enough to fry eggs and been something that many have b!tched about for some time... it's not likely that the "SAME" desktop G4 that makes up the high end will be used in Powerbooks. It's just waaaaaaaay too hot, so I expect a smaller micron sized G4 for Powerbooks 'til a future PPC 9xx processor is suitable for the application. It's also safe to say since the PPC 970 is hotter than the current laptop range processors, it won't see application in any Powerbooks until the smaller die and more efficient (based on rumors) PPC 980 ships.

It's also not logical for Apple to launch every new machine at one time when processor yields on first gen processors tend to be sketchy. Betting the farm by launching "EVERYTHING' on it is sort of like betting the farm on a 50/50. A political coup in the light of what happened with the first generation Pentium with it's floating point flaw stands to mind as something Apple would struggle with far more than Intel in the bad PR it'd bring.

What machines can deal with excessive heat, and need the most extreme power?

Pro Desktops

Pro Servers

They're the first for early adopters because they're the top of the heap; the machines that people that want "POWER" buy. Expect 970's in both of these, and then if the rumormill is correct... expect Power# based IBM chips in Apple's mid-high end Enterprise stuff, as they try to transition from a consumer-pegged company into something much grander in all directions (consumer electronics, consumer computers, workstations, SOHO servers, and Enterprise). Apple won't beat Sun and IBM and HP and the like at their own game over night (much less Sony and Panasonic and Best Buy and Musicland), but give them time and they could become a player... and then work on maximizing to all reaches. After all... Enterprise is the cash cow. Sub $700 PC's are the things that you do only when you're making $ hand over fist to prompt a loss leader; and even then... when you're competing against Koreans building the equivelant of a $900 PC for $500, it's hard to beat them (eMachines) and has killed or maimed countless competitors (even hurt Acer who is Korean, killed off Packard Bell, and crippled Compaq enough to let HP take them over).

Which Apple can't afford.

Anything new goes in the above 2 first... it'll also take the greatest and swiftest changes as necessary in hardware because it turns the largest turnaround per item in profit, in terms of margins. Apple easily makes more on a single XServe than it does on a iMac or iPod or eMac. Given Enterprise tendency to charge for services on a much grander scale than even their hardware profits... imagine buying a machine at a premium in terms of profit margin to Apple, and in turn paying a fee yearly to provide support for those that need it. At a yearly pay scale, much like Sun and HP and the others provide... this is a "MAJOR" cash cow without even giving a piece of hardware for that expenditure. Selling Enterprise hardware is like a gift that keeps on giving, especially if a company doesn't have it's own IT department.

It's just common sense...

I doubt we'll see new eMacs, they just shipped with new architectured motherboards. Likewise, I think the iMac will shore up on the high-end of the G4 lineup and continue. What I do expect is to see the PowerMac G4 and XServe G4 processors make their way into the eMac and iMac the very minute the PPC 970 launches. This might require some revamped I/O technology to make them match the PRO predescessors... but it's only logical that this is the way it'll go.

As for the Powerbooks...

G4. It'll be this way until the more efficient PPC 9xx chips are available that'll not burn your lap to a crisp or require you to wear liquid cooled knickers inside a Nomex fire suit. The current high-end G4's are not suitable for laptops, the PPC 970's are too hot, and right now only the low-watt G4 and extremely low-watt G3 in the iBook are. Since Apple has just released the 12" and 17" Powerbook, I expect Apple to eek the line out as is for a bit, then uprate the Powerbook line, replacing the Ti 15" with an Al 15.x" and then uprating the processors across the line to a new smaller die size, more efficient G4 model that matches or beats the current G4 desktops in power.

iBook...

G3. Why? Cheap, cheap, cheap. Apple's 12" Powerbook is a relative bargain, and moving the iBook to G4 would seriously put it in competition against the G4 Powerbooks. Not very logical. When the Powerbook goes PPC 9xx, then and only then will we likely see a revamped and uprated Powerbook motherboard being slid inside the iBook casing with a G4 processor in the Ziff. Like to see a faster G3, and knowing the great efficiencies of the G3, which is coming in at a whopping 7w of power consumption (very little), pushing the G3 up a bit in power, perhaps sacrificing a "teeny-tiny" bit of the efficiency for more grunt could make the G3 a decent competitor 'til it's time to move to G4, and then... to a low-end PPC 9xx once everything else is screaming along on 2nd or 3rd generation 9xx-based processors.

PPC 9xx in everything right off the bat is downright moronic. Why? Pricepoints. If all of your machines are the same processor... then what you're basically saying is that there's no needs for a desktop PowerMac because the Powerbook does everything you need. The fact is a PowerMac doesn't need to worry about battery life or heat in the same vain that a Powerbook does...

You also have pricing...

Why buy a high-end PowerMac Pro Desktop "workstation" when you can get a consumer iMac that is the same speed? Or hey... consumer eMac. Even if the processor speeds between one or the other is "close", the differences are there in terms of I/O and expandability, bus, cache sizes, etc. etc. that the Pro line is still significantly faster, and as a result... more expensive. Yet, you get what you pay for....

Out of the box, for IBM to produce dozens of variations just for Apple's needs is fruitless and time consuming and likely would come down to the consumer in terms of jacked up prices (plus, mind you... it's been said before that IBM will use these processors themselves, so I half-expect IBM to take up a good body of production for their own needs as well). Wait 'til the machines go through a generation of teething, and the developers find ways to make them more cost effective... as well as more efficient in energy consumption and battery usage, and in generating less heat, and you'll find them in lots more applications than launch.

Only other place I expect to see PPC 970's outside of Apple Pro Desktops and Apple Servers once they're launched (less a *IF* but a *WHEN*)?

IBM Servers.

Yeah, they don't need SIMD... but a 64-bit low cost processor that they can use in multiple segments and sell to customers while also using? Are you kidding... Apple's investment in this helps subsidize and offset the costs of development of a processor that IBM can maximize to their needs. Granted, Altivec-style SIMD is along for the ride as a passenger, not a driver... but IBM can maximize the PPC 970's benefits to them, and ignore the rest they don't need... or yet... they can even consider building a variant that is sans SIMD if they want.

 

[dongmin]

Re: Re: is apple beating a dead horse . . .

Originally posted by arn
Besides... I see everyone's getting their hopes up... and getting a little greedy... personally, if they _demo_ the 970 at WWDC, I'll be pleased. You'll be lucky if "only" the PowerMacs get the 970 as early as June.

arn

Ahh, the voice of reason--rather rare on this forum.

I have to think that moving the date and place of the WWDC has to be significant in some way. There has to be more than Panther. (Jaguar was a bigger upgrade and it didn't demand a date switch and bigger venue.) There has to be some sort of 970 announcement.

For my money, I think Apple will announce 970 powermacs, publicly, and ship them within 6 weeks. And within that same time frame, they'll announce faster G4 PowerBooks, just to quash any false hopes of 970-powered PB debuting any time soon.

So by the end of summer, we should see:

Powermacs: 970, up to dual 1.8 ghz
PowerBooks: G4, up to 1.25 ghz (7457s?)
iMacs: G4, up to 1.25 ghz
iBooks: G3, up to 1 ghz

I'll be happy with these announcements (especially since I'm waiting to buy a Powermac), but I'm sure a lot of people on this forum will be bitchin and moanin.

 

[ktlx]

Why would I buy Apple for the enterprise?

Originally posted by IVIIVI4ck3y27
Expect 970's in both of these, and then if the rumormill is correct... expect Power# based IBM chips in Apple's mid-high end Enterprise stuff, as they try to transition from a consumer-pegged company into something much grander in all directions (consumer electronics, consumer computers, workstations, SOHO servers, and Enterprise). Apple won't beat Sun and IBM and HP and the like at their own game over night (much less Sony and Panasonic and Best Buy and Musicland), but give them time and they could become a player... and then work on maximizing to all reaches.

People keep saying Apple is going to take on the enterprise and will be a player but for the life of me, I cannot understand why.

I am in a large company's CIO organization and have purchases millions of dollars of enterprise class servers over the past few years. I cannot see any situation in which Apple servers would be even remotely interesting.

An enterprise server has four variables: hardware, software, reliability and managability and services support.

Apple cannot really distinguish itself in the hardware area because the main differentiator is the CPU and Apple will be doing nothing more than reselling IBM CPUs. An Xserve 970 is nothing more than a 1U IBM server with a prettier faceplate and ATA drives instead of SCSI. BFD.

Apple's Mac OS X actually is a hindrance in my opinion because even though it is built upon FreeBSD, Apple has munged enough of the UNIXness of the system that it would require retraining sysadmins already familiar with thing like Linux, Solaris and AIX. If Apple is doing nothing more than reselling IBM CPUs, I may as well buy the servers from IBM and run AIX or Linux and forego the retraining costs.

Apple's reliability does not seem to be any better with their Xserve than any other 1U server. Given their reliance on ATA drives, it is probably worse over the long run. While Apple's management software may be better for managing one or two servers (and even that is open to some debate), enterprise customers have dozens or even hundreds of servers to manage. And not all of them are the same platform. In this case you are managing your systems with third party applications anyway.

Apple does not provide services in the same sense that IBM, HP and Sun do.

I guess Apple could change a lot and try to go after this market, but it just seems to make no sense to me. Apple really brings nothing unique to the table in this market space and it is not even close to parity in a lot of the important enterprise qualities.

Apple seems to me to be too smart of a company to throw good money down the toilet so I would be very suprised if they tried to be anything more than a niche player. I could see them going after the back offices of content creation houses. I find it hard to see them targetting someone like me who purchases generic UNIX servers to run databases, Web, file and e-mail servers.

 

[rjwill246]

Re: These PPC970 rumors are plain silly

Originally posted by fpnc
Well, it is my opinion that these rumors about soon-to-be-released PPC970 Macs are entirely wrong. I don't expect that we will see PPC970 based Mac desktop systems until VERY late this year or more likely in early 2004. PowerBooks probably won't appear until well into year 2004.

Well, mate I think you are spot on. Macbidouille has certainly put its reputation on the line and maybe they are somewhat right but... If so, Apple would have to have their product line with 970s ready to go on the day of the announcement, or else, not ONE Mac will sell from that moment on. Now, Apple could weather that potential disaster by offering the most amazing firesale on all extant computers in stock and/or offer a substantial inducement for people to lay out money immediately on the 970s in order to keep up the cash flow. I can't see any of this happening.
Still, let's hope that Apple can pull the rabbit out of the hat.

 

[wms121]

the high end will move on...

let's examine a few more issues here..

Apple's so-called "new enterprise approach" with the 970 may make sense..if we assume they mean the "home-business compatible" server. If a POWER5 based box can make it in anyone's locality ..and subjected to consumer pricing points..Apple still has a sale that IBM can't make yet.

Secondly...beyond Infiniband 32X..and 256 bit processors ...and global
grid resourcing..what else is on tap for the high end?

Artificial Intelligence...the REAL STUFF.

It needs more than a TeraFLop of processing power..and terabytes of memory to work properly. What applications...ALL
applications if sentience could be demonstrated...even collective expert systems would be trainable for "home-based operations"..
business or otherwise.

Check out the requirements they are evaluating for Internet2 and beyond...some sentience is assumed..some implied. After system
bandwidths go beyond the Infiniband 32X metric...only geeks will truly know how much real power some of these machines may have. I used to hear crazy stories about overclocked 25 MHz Intel
80386's running at 80-100 MHz. Something similar may happen to a.i. . They have never managed to make full use of the compression-decompression algorithms out there..1000:1 data rates are possible in neural pattern sets.

IBM optical chips in Power Macs? Yep.

Internet3-4 capability by 2010 for Apple..yes...they are already pushing beyond the Internet2 specs.

A.I. machines for rent via the Apple Store? Shared teraflops with IBM? Ask IBM...they have wanted their "sexy Apple friend with the nice consumer spread" much closer to them for a long time.

Moto may be back doing 2N20's if AMD starts doing PPC's

<--wants a 980 laptop

 

[nuckinfutz]

PPC 970 systems will be coming way before 2004. Come on people. Even IBM's pdf docs on the 970 says "VOLUME" production 2H 2003. And this was shortly after the Microproccesor forum in 2002. You have to truly pessimistic to believe it will take Apple until 2004 to release these systems.

Where the hell are they going to go? 130nm G4's from Moto aren't due until Q4 2003. The current 1.42 G4's have to be costing Apple a FORTUNE...I bet yields suck on that thing.

I don't see Apple going "Enteprise" for years. Shipping a 1U Server doesn't make you Enteprise yet. They'll continue to improve. Apple said they "humbly" enter the Server space. It'll take time.

Powerbooks may or may not get 970s this year. The 1.1volt 1.2Ghz chip dissipates 19watts and could possibly be shoehorned in. Or Apple could wait until the 970 is 90nano making it an easy fit.

Lots of option coming.

 

[noel4r]

no 970 in iMacs until next year? damn, i bet intel will be up to 4Ghz by then....

 

[Shaktai]

My two cents worth.

• 970s this summer with July Availablity. PowerMacs only, followed by X-serve. Initial units will run a patched version of Jaguar until Panther is released, with a free Panther upgrade for the 970 models. Top end premiers at dual 1.8 ghz. Faster models will follow when faster chips are available in quantity. (Yes faster 970 exist now, but not enough to meet demand)
• No 970 powerbooks yet - as much as I would like to see that. Speedbumps? Maybe. Faster GPU? Good chance. Wait until 2004.
• Speedbumps only for iMacs. Nothing major in changes until MWSF 2004, then maybe a redesign to coincide with 970.
• No changes to iBook until Fall, then speedbumps with the new G3 chips. No G4.
• Apple's Enterprise multi-chip solutions will actually be modified IBM server solutions optimized for OS-X and intended specifically for the creative and certain scientific markets. They will be marketed in conjunction with IBM who will provide support, but not before next year.
• Next PowerMac will be called the G6, because the G5 was aborted by Motorola.

I would really like to see a 970 in a Powerbook this summer, but don't believe it will happen. I could imagine a dual 1.2 ghz 970, 17inch powerbook but think it will have to wait.

 

[fourthtunz]

Originally posted by noel4r
no 970 in iMacs until next year? damn, i bet intel will be up to 4Ghz by then....

Yeah but it will still be like the time I had a
55' Chevy with a 327, lots of power but bad mpg and basically good just for raw speed. I sold it, I like my 93 corolla alot better
😀

 

[RC23]

hmm.. if mac is switching to IBM cpu's.. would it be dumb to get a g4 now?

 

[beatle888]

it would be if you can wait.

 

[NeXTDev]

Re: Re: What if..

Originally posted by IVIIVI4ck3y27
Simple.

Because Apple is "JUST NOW" getting things ported to PowerPC using a myriad of Carbon/Cocoa. Cocoa = very portable. Carbon = "pretty portable", but in that same sense... look at how long it's taking Quark to get to PPC via Carbon in OS X?

Moving to AMD would require "EVERYONE" to rewrite.

A re-write isn't necessary. Most of the big apps are architected for portability anyways. The endianess issue is already dealt with for most apps since a substantial amount of the code base must work on Windows. As long as the underlying APIs stay almost the same, it would be a trivial port. The difficulty is in re-factoring code for new API's where the fundamental assumptions may differ significantly. In this case, we're talking about the same API's, just ported to a new chip architecture. Most code will move easily, but anything written in assembly would have to change - most likely that is a small portion of anyone's code base.

Of course, Cocoa is designed to be very portable, but even then the same issues of making sure you bit swap for bit fields and making sure you don't make stack assumptions are important - Cocoa just isn't architected around specific machine or platform issues except for the possible 32 bit assumptions in some places. Even that should be a relatively minor cleanup.

But there are many other good reasons to not go to the x86 side, including things like losing a near full speed Classic environment, the lock step with the rest of the PC world, the lack of mystique, and the need/want to support all sorts of various x86 configurations and peripherals. SGI tried to go the x86 route, and it has been a disaster except possibly where they can leverage their NUMA experience in making big compute servers.

 

[ddtlm]

macrumors12345:

The 970 will also have a superior FSB to the P4 (450 mhz double pumped vs. 200 mhz quad quad pumped). It should match up very well and beat the P4 based machines by a substantial margin in many tasks.

But its more complex than "look at the mhz" and "low pumpedness is good". Remember that the 970's bus has 1 byte in 9 overhead, and remember that it is made of 2 unidirectional 32-bit busses vs a single 64-bit bidirectional bus. I don't think anyone here is qualified to talk about which is better, so its time to stop making baseless pro-Mac statements.

herr_neumann:

Apple has no choice but to have the 970 shipping by august. If they go any later than that, they miss the back to school purchases, which is a big market. There is no way Apple wont target this market. I would like to think they have learned a thing or two from their last fizzuckups.

Back to school doesn't mean squat if the chip's arent ready.

Silencio:

I assume all those benchmarks use Altivec, and, if they're accurate, they really show the true power of the technology and how much it was hamstrung on the G4 by glacial bus speeds.

Don't you think that if the G4 was held back by bus speeds then someone could provide benchmarks showing this? In particular, I would love it if someone would link me some benchmarks where the 17" AluBook defeats the 15" TiBook in processor tasks. I also can't remember any PMac benchmarks that clearly showed the 25% faster FSB and RAM doing and good, but the PBooks are particularly good test subjects because the CPU and L3 did not change at all.

Frobozz:

The photoshop actions test took a Dual 1.42 G4 72 seconds. A dual 1.8 970 took 24 seconds.... and a single 3.06 P4 took 58 seconds.

Photoshop!?! It is forever tainted because of the G4 performance myth (lie?) that it perpetuated for so long. Lets see something new, something credible.

pudrik:

If Apple is able to beat Intel to the 800MHz FSB, it would be a huge marketing windfall, even for the uneducated crowd. (Yeah, their processor runs faster, but our whole machine runs faster.)

Intel chips with the 800mhz FSB have been available for sale for weeks I think. People have already overclocked the FSB into the 1ghz+ range.

ktlx:

People keep saying Apple is going to take on the enterprise and will be a player but for the life of me, I cannot understand why.

Nor can I.

 

[solvs]

Wow, when did you get so cynical?

All I can say is, if the 970 is available in the next couple of months, with all the info I've heard of it (substantiated details, not rumors) it may not beat the pants off of the fastest PCs, even the ones shipping now, but it should at least bring the Macs closer to parity. AMD like. Over the next year things look interesting in the price/performance areas already.

I hope they do come out, and I'll be disappointed if they don't, but the world won't end.