Freescale's 90nm PowerPC G4 chip destine for Apple laptops - Page 3

*sigh* at this stage asking me to drop altivec is like cutting off my left testicle \
give us time, my friend, just a bit of time, maybe 'till the end of the year...

I'm just guessing that Apple doesn't like the VPU in the Pentium-M. Since Apple has been developing OS X for x86 for five years, any Velocity Engine code in OS X PPC probably has SSE counterparts. Yonah is supposed to get an improved VPU that's the equivalent to what's in the Pentium 4. IOW, not the equivalent of Altivec but good enough.

How hard would it be for intel to make apple special chips with an altivec co-processor?

T, FTFY

We don't want 7448's.
We want 8641's (for iBooks) and 8641D's (for Powerbooks).

Then maybe Apple could compete with current PC-world mobile chips (Pentium-M 700 series).

More info here.

Couldn't agree more - but this is the reason Apple has dumped PPC - its always behind the Intel curve - both Moto and IBM are unable to deliver on sustained incremental improvements in processor development. So much for the mythical low power verison of the 970, and despite the 8641's potential, Apple are not going to be allowed to be strung along anymore.

WOW!!!! These cpus would be perfect for Apple's laptops.....
ABOUT 4 YEARS AGO!

Why bother. I bet hindsight's a killer about now. If they knew Apple would actually abandon the powerpc camp they'd have kept up with the x86 side.

Well I would like to think that that was the case, that is that Freescale and IBM where driven by customer needs, but obviously that is not the case. As far as Freescale goes they dropped out of the desktop processor market a few years ago, Apple only uses the processors because they have nothing else to put into their portables. Apple didn't switch to IBM for nothing you know. Freescale simply decided not to play in that market, that is the high performance host processor market. As far as Freescale goes they may have made the right decision for them, unfortunately that leaves a customer hanging.

IBM on the other hand seems to have resorted to familiar short sighted management practices and decided to milk one customer instead of trying to expand their customer base. IBM seems to have failed Apple in a different way, that is greed and a failure to invest. It does make one wonder if IBM will have any manufacturing capability 5 years down the road.

As far as Apple goes, I think they are screwed. At least with PPC they did have slight advantages with certain applications. On Intel they will be trying to maintain parity. Worst yet there are many good UNIX options available for Intel hardware so Apple will need more than a handfully of multimedia applications to get any creed in the market. Does X86 really need another OS when there is so much to choose from already.

Thanks
Dave

Yes but on to what?

Lets face it, for many AltVec was the only reason to own Apple hardware. OS/X simply doesn't offer as much as some imagine - iTunes does not make an OS for most people.

Dave

Yeah, you're right, who cares about a 10% performance advantage. I guess it means nothing when your 200% behind the competition.

I'm not sure why drop-in support is important. If Freescale could provide competitive laptop CPUs to Apple, then Apple would probably stick with PPC, in which case they would happily redesign a mobo to accept a bitchin' processor. I guess if you mean for now, then yeah, Apple needs a drop-in CPU to tide them over for a few months. But Freescale still sucks.

To what? A good question that assumes Altivec made a difference in the first place. For those who require vector operations (maths, media, graphics, etc), Altivec is the bee's knees and did make such a difference. I'm quite happy with the thought that many would only use Altivec in encoding mp3s...it made a difference but only in limited circumstances. It certainly didn't make the same difference to overall system responsiveness that ongoing revisions to the Darwin kernel have made in the past 4 years. A debateable point I know, but one I'm happy with.

One of the previous selling points for Mac Hardware was to 'think different' while using RISC hardware. Over the past 15 years, that has become increasingly questionable, with the transition from IBM and Freescale CPUs the latest and maybe last of these transitions.

When I 'switched' 1.5 years ago, i was sucked in by 'megahertz myths' and Altivec's 128 bit registers, just like any other mug brought up on the notion that more must mean better. Apple did well to entice me to make the change given that the hardware wasn't as good as a comparable Wintel solution. However, the security of the OS (being halfway through a PhD) gave me the confidence that blowing a large wad of cash on an untried solution was an acceptable risk.

Back to the first point - This doesn't mean, of course, that everybody didn't catch up in other ways with the huge advance that Altivec was in 1999. It was good in 1999 but would other CPUs and their comparative architectures stand still while Moto's version of SIMD hit the stage?

Just looking at Pentium M is enough to suggest that much of the breastbeating with regards to Altivec has much to do with the untapped potential of this VPU and the aging CPU it was attatched to. When the G5 came out, everybody saw that Altivec and RISC would continue and thrive. Now after being duped/outflanked by a combination of Apple and IBM decisions, people who put a large amount of practical and emotional cache in this aspect of Apple are now reaping the whirlwind.

BTW, Is anybody mourning the death of the G4 or G5s double precision FPU? As Yonah doesn't do this as well as even a G4, should we all be wondering dazed along the steets?

...Back to marking undergrad essays...

I vehemently disagree. While I will certainly miss having the more elegant, and in some specialized ways, more capable technology (read AltiVec) under the hood, what use is that if PowerPC cannot compete in the long run with the financial and technical capabilities of Intels R&D-machine.

Apple sees the future as a Performance/Watt-equation and that forced them to go with Intel and through yet another transition. Short term pain, long term gain.

To quote Steve Jobs: "The soul of the Mac is the Operating System". This is true and cannot be ephasized enough. Going with Intel will allow Apple to focus more on what it does best: Software Development. No more uncertainty whether or not Apple's CPU-supplier can supply Apple with enough processors for them to develop and ship new machines--new machines that Apple needs to sell to stay alive.

While it might be true that for certain specialized markets this could mean the further decimation of Apple's presence (would Virginia Tech have chosen PowerMacs/Xserves had they run on Intel, for instance). But these are small niche markets and the potential loss of these are outweighed by the many advantages that joining the Intel-juggarnaut brings.

Besides, I wouldn't rule out the possibility of future PowerPC-based machines from Cupertino beyond the end of the Intel-transition, either. If something interesting came out of the RISC-architecture Apple could exploit it at will, since I certainly don't forsee them removing that compiling check-box from Xcode anytime soon. And really, why should they?

i think one thing that sticks out at me like a dead pixel is the fact that I swear to Gawd macromedia's flash player for mac os has never EVER used altivec optimisation in their code. because it has always been like 5x slower than their PC counterparts in playing Flash movies.

macromedia developers, if you're in da house, i respect ya work, but this point still puzzles me:

laah dee daaah for altivec, what's the point if software doesn't use it?

as (i think PowerDoc?) pointed out, abstracting that Altivec/SIMD/SSE whatever whatever routines is only a good thing in the long run. For more abstracted, more modular, more efficient, more cross-platform code.

That at least is the way I console myself with the loss of Altivec, which like i mentioned, i have built an emotional attachment to buy(sic) being loyal to apple for the past 4 years.

?

Firstly it is OS X, and I'm assuming here that you've never used it.

What is Windows XP then? A device for playing things in Windows Media Player? No, it also comes with a webbrowser (IE, versus Safari), an Email client (Outlook Express, versus iMail), a video player (WMP, versus Quicktime), built-in virus/worm/spyware vulnerability, Windows Movie Maker (versus iMovie HD) and Chat (MSN, versus iChat). Where's iDVD? Where's iLife? Where's Automator? Where's Dashboard? Where's iCal? I presume it has an iPhoto style application, but I never found it. I like it's system-wide export to PDF ... oh, wait!

Never mind the fact that OS X actually lets you get work done. I haven't been using it for long, but I think I am twice as productive per unit of time than in Windows. It simply doesn't get in the way.

Not to mention that Xcode is free and included, and whilst it isn't a Visual Studio it is fairly powerful still.

Could I, in one day, set up Apache, MySQL, WebDav, a CVS server, then still have enough time to learn about SSI and create a small website and deploy it on Windows? No. But I did that yesterday on this Apple. Thanks to the built-in Unix subsystem anyway. This is a proper developer's machine.

I first used it last Friday. Before that I very rarely had access to Macs, I generally used FreeBSD, Linux or Windows XP at home. The only thing I hate is the Apple mouse, not because it only has one button, but because it seems quite unresponsive.

UNIX is not a subsystem. It's a System. There may be subsystems in UNIX, but UNIX itself is not a subsystem.

Are you kidding me? Have you used a real Unix before? Solaris, FreeBSD, NetBSD, etc etc??? These are no where near comprehendable to an average consumer. No consumer in their right mind is going to install a unix option over windows or Mac os X.

Then we have linux (not unix), have you used linux in a productive environment very long? Linux still has BIG BIG BIG issues. I work on it in a daily basis for a fairly large QA department. I test on Mandrake, Gentoo, Debian, Suse, RedHat Enterprise, and Fedora.

Each one of these is a nightmare to install on each system. Not all the time but about 50% of the time when you go to a different system these distributions act chaotic! Again no consumer in their right mind is going to install any of these options and mantain it. Perhaps a few years down the road when they get it really going. But again what software is their for Linux??? Not much! Graphics artists don't get illustrator or photoshop (gimp is not as powerful, I don't care what anyone says). It goes without saying that software is MUCH better in windows and mac os x.

Your theory is extremely flawed. If that was the case no one would have ever bought into the idea of OS X. Everyone would have just bought a cheap pc and ran linux or unix.

Architecture design (not counting duals, dual core, etc) doesn't mean a lot to consumers... they want 2 things: cheap price and high performance.

People will switch... theirs no doubt about it. Macs will be cheaper and much faster. Thousands of PC users wanted to switch before the announcement of moving to intel because OS X is so much better than windows in many people's eyes. Nothing has changed except the fact that macs will be cheaper and quicker.

As far as areal Unix'es it has been awhile but yes have used "real" Unix. As to currently I use Linux esclusively on all my PC's. You seem to categorize Linux as a non-UNIX platform but I see that as a convient distinction.

You seem to be a bit out of touch here, Linux has issues just like any other OS. As far as useing it though frankly in one variation or another I've been running Linux since RedHat 5 came out and experimentally before that.

The reality is that this mean very little in the context of the discussion. The issue now is that Apple will have to compete with multiple OS'es on the same archtecture. It will not be easy to explain away performance differrences base on the underlying CPU. Simply put I don't know at this point if OS/X can compete on Intel hardware. On the other hand I'm sure we will see interesting reports in the short term.

This I reject totally - I am a consumer as are the majority of Linux users!! Linux is certainly easier to maintain than Windows in all its variants. In any event it is totally uncalled for to suggest people are out ot their minds if they choose something other than OS/X.

Well if you want to believe that then go ahead! If Apple really wanted to compee with MS they would look seriously at implementing a software repository like Fedora or Debian has. Then they might have a chance.

That is what most people did, so my "theory" isn't that flawed. So the question then becomes how does Apple maintian its business model of selling average computers at high prices. This in a nut shell is the problem as I see it. Apple can't survive on low priced commodity hardware, so they will need to maintain margines on their products. Whitout other selling points related to hardware will OS/X be enough to attract new customers? I don't think so!

A combination that Apple doesn't always mix into a product. Granted at times Apple has had good performance with a new product but often the value part of the equation is a joke.

Well you are convinced of one thing I'm not convinced of, that is that Macs will be cheaper. I'm not convinced at all that that will be the case. As to quicker that seems to be the general wish, but again we will have to see just how well OS/X performs relative to the alternatives. People will only switch if they see OS/X as being a considerable value to justfy high hardware prices.

Dave

What Hattig said above. Hell, I'd run Mac OS X on an generic PC if I could, because you know what? It just frigging works. Sure snazzy hardware is nice to look at, but give me the environment to work in not some candy-assed up Windows 98 (read: XP) any day! Jeesh.

Mac OS X is Unix, it's got the power of Unix underneath (appealling to geeks and IT) and the Ease of Use of Apple.

I can't think of a way that migrating to x86 will not get MORE business for Apple, assuming that the prices come down accordingly. There are lots of people who want to use Mac OS X but refuse to pay Apple prices for hardware - the issue being that even though Apple prices have been slowly coming down, they aren't as low as the Intel prices for more powerful hardware. Whilst I expect the Apple hardware to be more expensive still, I don't expect it to be as high as it is now.

Also the risk is lower - if you don't like Mac OS X you can always install Windows.

One of the advantages of OS X is that it runs only on hardware Apple chooses and supports. Install it on a generic PC, and you'll see if it 'just works' or if it 'works if I get all the drivers, that is, if there are any'.

Why do so many seem to assume that MacIntels will be lower in price than the current PPC fare?

The incremental transition will make it difficult for Apple to lower prices without putting kinks in their product hierarchy (although starting at the low end does make it somewhat more feasible). The cost of Intel CPUs is typically not much different from PPC CPUs, and is definitely not significantly lower. About the only savings Apple will see is in R&D costs for motherboard chipsets.

Did Apple lower or raise prices significantly when the Powermacs switched from the G4 to the G5? I don't remember much of a change.

The primary reason a Dell costs less than a Mac isn't component costs, it's profit margins. Dell sells high volumes at razor thin margins, and Apple of course deals with low volumes at margins that average ~30%. So if Apple is ever to lower their prices significantly, sales volume must expand.

Of course MacIntels might be $50 cheaper or so, but to become competitive, Apple would need to shave a couple hundred bucks off of some Macs.

Dell's price is based on hardware costs plus WinXP cost plus "razor thin" margin.
Apple's price is hardware costs plus ~30% margin.... OR put another way....
Apple's price is hardware costs plus OS development cost plus R&D plus "razor thin" margin.

It stuns me that people think OSX costs Apple nothing.

I refuse to believe that Apple won't have negotiated a much lower price to use Intel's processors than their list price. They are a high profile company to have as a customer. I wonder if they got the price down to the price that Dell pays for processors?

Oh boy. Lets bet, we will put 2 machines side to side, with top of the line memory , same graphics card, one runs 1800 mhz P4 (or Mobile!) one runs 1800 Mhz G5. Forget benchmarks, lets run a good coded game as Unreal (Carmack should fix mem.access method bug on Doom), lets act like joe sixpack, comparing FPS!

I was wondering for years why people bitch about "zealots" and "hopeless fanatics" in Apple world, I got my answer after this intel crap.

Intel lovers, you got better sites to watch, like Toms Hardware.

Lets discuss this new PPC CPU and possibility of expansion card makers including it on new cards.

As apple is now another Intel based OEM PC vendor with custom locked OS, I am not very interested in Apple news anymore.

Microsoft with generation 9 (10 when Mactel ships) frameworks like DirectX will use more of Intel than whatever OS X or unix deriative you ship your machine with.

Lets start telling the facts if we are all moving to Intel. BeOS V2 is all I say.

Its costing nothing if you compare it to Windows XP from Redmond.

Those poor evil guys has to support every freaking chip taiwanese invents or copies with every kind of quality. They just package couple of chips with cheapest electronics available to a PCI case, code a amazing memory flooder driver (getting win32/64 SDK) knowing if it gets popular enough, Microsoft will take driver development from their hands.

You go and make a OEM PC at your own home, not even caring about what brand you put in, you plug XP CD and it installs. Next, you are just installing system updates, install directX, install any game of choice. It sounds normal as we are used to it.

If you think about it, that simple thing is result of BILLIONS of dollars millions of lines of code with thousands of tests.

Apple could ship a OS with broken modem driver (Microdash) which we STILL wait for updates (more than month?) to fix it. Its NOT possible for Microsoft to do it. They simply don't have those "fanatics" who forgot they are customer of a brand, being "more Steve Jobs" than the guy himself.

Apple has guts to tell "G5 sux" while still selling G5s, thanks to amazing ignorant, zealot so called community. I believe they spend too much, they should give Pentium 2 to such community, they will find a way to defend it.

What are you talking about?

Windows XP is nasty, the only excuse to use it is if you want to play a PC game.

Mac OS X is so far ahead of Windows XP in many respect it is quite embarrassing. What this move will do is combine the market leading hardware with Mac OS X, which in my eyes is a win-win situation.

People just ignore the difference between margin and gross margin.
Especially on this board

You're misreading my post. I agree with you, Apple needs to have fat profit margins to offset their R&D costs - that's the whole point!. What I'm saying is that because of Apple's need for high profit margins, MacIntels will not be appreciably cheaper than PPC-based Macs. Apple has to cover their R&D costs regardless of what chips are in their computers, because, as you pointed out, software makes up a large part of Apple's R&D budget.

Of course. The original poster was talking about using the 'drop in' replacement feature of the 7448 over the 7447A. It's only drop in if he can do SMT soldering.

I'd doubt very much a 7448 on a daughter card running a 200Mhz FSB would work at all never mind be reliable.

The point was that a 1280x768 12.1" screen would be nice to have, not that we we're running it on that Samsung. Personally I don't have a problem with a screen that small and that high a resolution. If you're eyesight isn't so hot, buy something else.

Also has a built in DDR2 memory controller, built in PCI Express IO, built in Gigabit ethernet, TCP/IP acceleration and runs at about 25W so potentially it's better than Yonah in some respects.

But, Freescale only quote speeds of > 1.5Ghz and mention it will scale above 2Ghz. And there was talk that the IO is more tailored to running it in a network switch or router than as a general purpose CPU.

Bearing in mind the current Pentium-M is hitting 2.13Ghz, Freescale need to get the clock speed up to compete. Yonah also has an interesting shared cache so running on 1 core it can use the cache from both and I'd bet on their power consumption being more battery friendly than Freescale's.

I suspect Apple has done the comparison already and worked out that going for a slower 8641(D) and having to redesign a whole motherboard rather than take the faster Intel chip and their 'platformization' would have just given a slower, more expensive laptop. And I presume Apple have seen roadmaps further ahead than 8641D and Yonah and know where the future lies.

end of an era of freescale+apple. 7448 will be the last freescale chip used by apple is my guess.

Well first off I do understand that Apple won't be paying list. However don't expect them to get a deal far better than anyone else. The reality is that Intel has to be very careful with respect to anti trust issues. For a specifc volume I don't expect Apple to get a deal that is any better than any body else shiping that many units.

Dell being the big dog as far as shipments go is likely getting a very good deal. Probally the best deal Intel has to offer.

The reality is though that hardware price probally isn't even a concern. Apples pricing structure seems to be oriented to sqeezing maximum profit from middling hardware. Heck I would love to see a change with respect to that issue but frankly I don't think Apple wants anything less than the margins it has now. If anything the switch to Intel could result in higher priced hardware.


Thanks
Dave

somehow i doubt apple will break the 30+ year generic trend of declining computer prices and increasing power of the PC industry -- im not sure why this came up though...

While I'd like to think that sound reasoning was behind the move to Intel I just have this bad feeling that reason had nothing to do with it. I fully believe that no matter what Freescale came up with Apple had alread made the decison to stop doing business with them. Material attributed to Mr. Jobs indicate that he had no desire at all to continue any relationship with Freescale beyond what he was forced into.

IBM's stumbling with respect to functional PPC hardware was likely the cause of the switch. Even then though I have to think, from what we know know, that Jobs heart wasn't with PPC at all.

As to Freescales chips I do wonder how the recent seminar went. It would be nice to know where they stand performance wise. It is to bad that a vendor of low cost PPC hardware doesn't exist anymore. That is hardware for running user applications. As it is Freescale still has trouble with e600 in the sense that integer performance still sucks. So even if they hit 2.5GHz they would still have a hard time competeing with some codes.

Dave

nevermind i see why. i thought we were on that ram tread. :-)

I'm surprised that this hasn't been thought of as a reason why the jump was made. Jobs, whether you care what he says or not, was quite specific when he said that the transition from PowerPC to Intel was driven by the future roadmap.

Now, to my mind, teh Steve wasn't talking about the PowerPC chips that are going to come out in the next 6 months...most likely to be e600 cores (7448 and 8641D?) from Freescale in laptops and Mini's and 970 MPs (Antares 970GX?) in desktops.

He was talking about what happens after this.

There has been no comment that I have seen regarding what IBM or Freescale were going to do after these processors. Freescale has a e700 on their site...some basic info (3Ghz, 64/32 bit) but no substance and a lot of history that says that these chips would be late and targeted at the embedded market. And at IBM, bugger all. Potential revision to the Power5 but thats it...

Thats where people's analysis should be...not on the near future but 2 years out. The movement to Intel is slightly more understandable in light of the chirping crickets and mangroves out there.

your right a widescreen pb 14 would be ideal for me and most, the 15 widescreen is bulky specially at meetings and tray tables. also i want thin and light with longer real life battery life. this is the year of laptops, most want them and the market shifted this year--more laptops sold than desktops.

compare a 14 widescreen to a standard 15 (it was done somewhere in the forums)