Apple to Announce the Power Mac G5 at WWDC?

a.ameri writes "Apple Insider is reporting that Apple will announce computers based on IBM's 64 bit PPC 970 processor in the upcomming WWDC and will market them as G5. The new Power Mac G5s will sport a completely new motherboard design utilizing DDR 400 RAM as well as AGP 8x graphics, FireWire 800, and USB 2.0, sources said. "In the box" connectivity among the news systems is based on Hypertransport which provides 64-bit addressing and will replace Apple's multilevel bus architecture found in current systems. Initial offerings of the Power Mac G5 are said to boast 1.4 to 1.8GHz, single core PPC 970 processors, with the possibility of a dual 1.8GHz chips shortly thereafter."

Comments

Probably true but...

[iJed]

While it is probably true that Apple will launch a PowerMac G5 at WWDC the information given here is only from a rumor site. Many of the rumor sites cannot be trusted much (such as MacOSRumors) and a one or two are extremely accurate (ThinkSecret). AppleInsider is one of the oldest rumor sites and at one time was one of the best. Recently though it has been taken over and the general accuracy of its stories is now unknown. However this rumor seems to have enough other sites reporting generally the same thing to be true. Its not fact yet though!

Re:Probably true but...

[jpkunst]

However this rumor seems to have enough other sites reporting generally the same thing to be true.

I'd say that that doesn't mean much because rumor sites will probably copy from each other without attribution.

JP

Does the clock speed matter that much?

[gotr00t]

Apple themselves have made public demonstrations trying to debunk the myth that clock speed is processing power. Being known for sticking to "slower" processors, it seems that Apple is finally starting to cave into the demands of the consumers.

I have tried to use the Distributed.net client on an AMD Athlon 1600 XP running Linux 2.4.10 and a G4 864 Mhz using Mac OS X 10.2. It seems that in terms of raw processing power, the G4 was actually more powerful, at over 10,260,280 nodes/sec, while the Athlon was only at 8,160,200 nodes/sec, and that's with no backgrounds processes running (besides the OS)

Re:Does the clock speed matter that much?

[Anonymous Coward]

Because the dnetc is optimized for the altivec capabilities of the G4. The Athlon has no similar vector processing unit.

Re:Does the clock speed matter that much?

[phelddagrif]

I wouldn't say that they've caved into the Mhz myth, so much as the G4 processors have pretty much hit the top of their lifespan quite some time ago. This is just a natural progression, they needed a new processor, and Moto's offerings are pants, so they went with the 970, which is a darn good processor.

Furthermore, if they were really caving, they would have gone with a processor with a few more RPM.. 1.8Ghz? That's pretty weedy in the world of x86.. That's actually pathetic as far as "new" machinew would be concerned..

Re:Does the clock speed matter that much?

[d3faultus3r]

Remember that Apple has optimised its processors to run with their OS, while Intel and AMD have no such luxury. I think a better test would be the above both running Linux.

Re:Does the clock speed matter that much?

[Alan Partridge]

the dnetc RC5-64 and RC5-72 client for MacOSX both make extensive use of hand coded Altivec optimisations. All that your results prove is that Altivec is an extremely powerful vector unit - but we knew that already.

Look at the dnetc client comparison database and you'll find some spectacular results for the MP Macs - a 2 way 1.42 Ghz G4 scores like a 16 way 1.05 Ghz Sun Ultra SPARC III.

Re:Does the clock speed matter that much?

[default luser]

If it's a very small loop that can be kept entirely in cache, the G4 is going to own that sucker.

It has very short pipelines, perfect for a single loop. This hardly makes it "better" overall, it is simply the better choice for that particular task.

Re:Does the clock speed matter that much?

[Andy_R]

You have misunderstood what the parent post was claiming.

Linux certainly is optimised for x86, but no-one would claim that the x86 has been optimised specifically for Linux!

Re:Does the clock speed matter that much?

[akuma(x86)]

Please people.

Stop using "Distributed.net" to compare microprocessor performance. It's a highly skewed benchmark that really only tests the speed of the "Rotate" instruction (which is on the critical path of the program).

Altivec supplies a data-parallel version of the Rotate instruction so processors with altivec can do many rotates in parallel which is why a G4 will beat anything else (no other processors have this data-parallel instruction because it is completely useless with the rare exception of this app). That is to say that most other computer designers felt that adding this instruction would be a complete waste of die area and power, since no other ISA supports it (x86, SPARC, MIPS, POWER etc...)

Distributed.net ...

1) Does not test branch predictors because it's a simple loop that is very easily predictable by even the most trivial preditctors

2) Does not test the internal L1/L2 cache hierarchy because all of the data fits in the L1 of most processors

3) Does not test the memory system (DRAM/Front-side-bus/memory-controller) because, as mentioned in #2, all of the data fits in the L1 cache.

4) It does not test the instruction cache performance because all of the code fits in the L1 instruction cache.

Stop using it to compare general-purpose computer performance. It is only important if the only app you care about is distributed.net

Your Athlon 1600 will spank the G4 at most everything else.

Re:Does the clock speed matter that much?

[lewp]

The only people who need to do those things are terrorists and thieves. Would you like to confess now, or down at the station?

Re:Does the clock speed matter that much?

[RevAaron]

It is a good thing to know, acedemically, that Altivec is the reason dnetc is so fast on a G4 and that there is no equivalent on the Athlon or a P4. However, this does not change the fact that the G4 performs so much better than the x86 processors available- and isn't that end-of-the-day, real-world performance what matters? It's not like someone can say "The Athlon doesn't have a vector processing unit- so you have to take it out of the G4!" and expect some "more fair" comparison. The G4 is the G4 and the Athlon is the Athlon.

Re:Does the clock speed matter that much?

[RevAaron]

People may blow off Altivec G4 performance, thinking that it's something very few applications take advantage of. That was true with Mac OS 9- all but a few, specifically coded apps would be any faster on a G4 than a G3, given the same bus, RAM, CPU MHz. Without Altivec, a G4 is about 15% or so faster in raw FP ops than a G3 of the same speed. Int ops are about the same.

But most folks don't run OS 9 anymore as their primary OS, for a number of good reasons. People run OS X. And OS X has Altivec optimization throughout the core of the system- in libmath and others. Anyone who has used both has noticed a signifigant speed increase in OS X when moving from, say, a G3/500 to a G4/500 that doesn't exist in OS 9.

Re:"New!"

[pldms]

3... 2... 1...

Go!

Which wintel motherboards have fw 800 and hypertransport? I'd be interested.

Appleinsider is a rumour site, btw.

Re:No Gigabit Ethernet ?

[pschmerg]

My dual g4 450 that I got back in the fall of 2000 had gigabit ethernet, so I don't see why they'd remove it from the machine.

nTh Post!!!

[FosterKanig]

Anyway, what I found most interesting about the rumor/article was the inclusion of USB2.

They have long championed Firewire as superior (which it is, and is still included) but it is nice to see that they are willing to adapt and a more common USB2.
This acceptance of USB2 shows a willingness to accept standards, no matter how wrong they are.

Re:nTh Post!!!

[squiggleslash]

There's nothing particularly wrong with USB2 - it's seen as a competitor merely because the bus speeds of FW and USB2 are similar. USB2 is cheap and processor intensive. FW isn't either cheap or processor intensive. You'd want to use Firewire for most storage applications (ie the computer's permanently attached disks, etc) and processes where there'd be high processor involvement while those devices are in use, such as DV cameras.

But there's no reason for USB2 not to be used for a lot of the "rest of the stuff", such as portable storage devices, CD burning, cameras and MP3 players, etc. There's no reason to believe that while these devices are being used, the machine itself will need to do a lot of other work, and given the price difference, it seems reasonable.

You might liken it to IDE vs SCSI, except IDE was a real heap of crap so even when performance wasn't an issue, there were still good reasons to go with SCSI. USB2 on the other hand is a decent enough standard, has wide support, and shouldn't be treated with the snobbery it usually attracts.

Re:nTh Post!!!

[willtsmith]

Ummm,

While Firewire does stuff that USB2 cannot (because it's effectively dumb and requires the host CPU), they do compete in a LOT of markets.

Firewire only
Consumer digital video
Point to multi-point, hi-bandwidth networks

Firewire and USB2:
External storage connection - HD CD/DVD

Medium bandwidth imaging, scanners, printers and webcams

Point to Point networking - adhoc networking cables

Satellite peripherals - MP3 players, Handhelds(though I haven't seen a handheld to date that included firewire)

--

So they do compete in MANY markets. It's a shame because firewire was set to take over the high-bandwidth PC peripheral area. Intel announced USB2 and scared off a lot of motherboard manufacturers. USB2 came in late and buggy.

USB2 will never truly be able to compete in the digital video area. I have seen a few recent offerings that stream video over USB2.0, however I expect these to be absolute flops. Despite USB2's higher theoretical bandwidth, it's reliance on the CPU makes it subject to unpredictable dropouts and dropped frames.

Beyond that, the need for a host CPU controller makes it an unreasonable option for handheld and mobile devices. Their "hack" for this area "USB On The Go" is a two year old standard that to date is complete vapor-ware.

It's a real shame that Intel chose to intervene with USB2 and effectively sabotage large scale firewire deployment. The lack of critical mass has kept firewire prices high (with relation to USB) and kept firewire as a niche player in high-performance (low CPU utilization) scenarios.

One last comment, the Firewire folks DID screw up in a way. The 4-pin connector includes no power. That connector (due to it's smaller size) has showed up on all kinds of notebooks. The problem is that plugging a dependent peripheral into these 4-pin plugs does NOT provide power. This puts Firewire at a serious disadvantage when compared to USB. All the USB ports theoretically CAN include power. As much as Sony wanted the smaller connector, I think Apple should have insisted on a small form-factor plug that INCLUDED power capabilities. Certainly, they should have insisted on licensing policies that prevented computers and hubs from including integrated 4-pin, unpowered connectors.

Have no fear. Firewire is alive and well beyond the reach of any of Intel's machinations. The Consumer Electronics has embraced the standard as a replacement for RCA, S-Video, and component video cables. Furthermore, It's widely speculated that some derivative of firewire (firewire 800) will replace DVI (and ultimately RGB-15) as the method of interconnection between computers and digital monitors due to it's ease of fabrication, size (low pin-count) and a plethora of hardware from the TV industry which will run on firewire.

Beyond that, Firewire has some serious applications in the arena of cluster computing and storage area network. The firewire protocol itself is designed to run over CAT-5 and fiber to produce super-bandwidth networks.

So yes, Firewire does do a lot more then USB2. A lot of these things CAN be done on USB (poorly and inefficiently). But it would have been nice if USB2 hadn't butted into the desktop/laptop space and prevented a very, very QUICK introduction of firewire across EVERYTHING (including cell phones and PDAs).

Re:nTh Post!!!

[huckleup]

EXACTLY like firewire does

No, it's not *exactly* like FireWire. Actually it's not the same at all.

FireWire is a true peer-to-peer model and can work in a ring or star mode. USB uses a Host-Periphieral model where all data must go through the host and only operates in a tree mode with the host at the root. If it is both host and peripheral, it is a leaf on the peripheral end and the root of another tree on the host end.

In FireWire if you have three devices device A can send and receive data directly to/from device C. If it's in a star mode you need a hub, but that doesn't put any load on any of the other devices and is essentially just routing. In ring mode device A sends the data to device B but it just passes it through at the hardware level to device C. You can combine stars and rings, but that is just phyiscal and not logical, as the data is essentially still just passed from one device to the other with no software processing required by any of the intervening devices.

In USB you have a Host and a Peripheral. First off, the host must essentially 'poll' each peripheral to see if it has anything to say. A peripheral cannot initiate a transaction. The polling happens each frame, which is 1 msec in USB 1.x. Secondly, if you want to send from device A to device C you really have to tell the host that you want to send the data to C, then it asks C if it is OK, then the host essentially brokers all the transactions. All the data has to go into the host, get buffered and prioritized and repacketized and peeked and poked and then is turned around to device C, mostly all in software running on the host processor.

FireWire uses a collision avoidance scheme on the virtually shared wire similiar to the way ethernet works. There is no host required to poll peripherals or broker and process all the transactions.

Devices that have both a host and peripheral controller means it has to have 2 connectors since they are different physically. (There is that USB2Go thing, but that's really just a repackaging of the hardware, while all the same host-peripheral and sofware issues remain.) While it is a peripheral it is at the mercy of whatever the host is allowing on that side of the fence. You don't really get a star, you get a messy tree with a slew of idiocsyncracies, and delivery times that become very unpredictable.

If you want to be a host, then you have to essentially replicate what the major OS vendors have done as far as driver support and such. Host controller software is infintely more complicated to implement than peripheral software. It has to have drivers for all the possible peripherals that may be connected to it, and possibly support loading of drivers (at least for updates and such, if not to work with mfg. exclusive-class peripherals). It has to be able to a whole bunch of stuff, hard stuff like scheduling for all the peripherals. If this custom host is also to be a peripheral of say a computer or other host, it has to deal with bridging between the other host and the peripherals connected to it. It has to intervene on every transaction. If you want any kind of throughput you have to have a pretty heavy duty microcontroller to do all that work.

Then there is the issue of drivers. The host has to have native drivers for all the peripherals it is to support. When the peripheral is plugged in it has to negotiate with a driver that knows how to talk to it. The host can't ID a device that it doesn't have a driver for. So if you had a camera, a printer and some weird host in between, that host would have to support both devices with drivers just to pass data between them. Do you think the scanner and camera manufacturers are going to provide drivers for every propietary host OS? It's hard enough to get drivers for Mac/Win/Linux/Unix OSs. In FireWire only the two devices that are communicating need to support a common protocol, since any other device in the ring or star would just be passing around the raw data and doesn't have to support each device.

G5 a good name?

[freedom_leffo]

Is G5 a good marketing name though? I've understood that the whole Gx-line is thought of as slow and stagnant processors. Perhaps a name not associated with Motorola would be a better idea.

Thoughts, anyone?

Re:G5 a good name?

[Hawthorne01]

It will work right up until the G8 is released, which will not sell due to massive "spontaneous" protests that form around each machine with it inside.

Shadow and Substance.

[A_Non_Moose]

Panther (10.3) we know is coming, that is a given and that is the substance.

The "Shadow" is the G5 and even the most die-hard mac fan would most likely utter the phrase:

I will believe it when I see it.

IMO, apple needs to figure out if they are going to keep/dump metadata...and stick with it.
I find it quite half-assed you can generate previews of images, but not store them.
(with the exception of Internet Explorer, but only one at a time)

(won't someone think of the pr0n collections?)
.

Re:Shadow and Substance.

[Andre Breton]

"[...]and even the most die-hard mac fan would most likely utter the phrase:

I will believe it when I see it

Nah, the phrase we die-hard Mac fans use (when talking about the G5) is:

I want to believe

I'm really not trolling, but...

[niko9]

whats stopping IBM from making these chips available
with an appropriate motherboard for folks who would like to run linux/bsd/ on them?

Re:I'm really not trolling, but...

[RalphBNumbers]

Nothing.
In point of fact IBM announced they would be selling PPC970 based systems running linux months ago. The announcement concerned blades, but I'd be willing to bet they'll build "low end" (compared to Power4 systems) workstations around them too, finally phasing out their old PPC604 low end workstations.

Of course, I wouldn't count on them matching/beating Apple's price point. Historically IBM's PPC based stuff has been *much* more expensive.

Re:I'm really not trolling, but...

[doce]

As far as I know, the technology is licensed to Motorola which is actually manufacturing the chip and reselling it to Apple.

alas, no. IBM and Moto both manufacture PPC processors. Most G3's, for instance, were manufactured by IBM (IIRC, even the current iBooks sport IBM procs), but G4's are all from Motorola* because IBM refuses** to use the Altivec unit.

Anybody knows if Apple has been able to gain any market share in the rackmounted server market with its 1U servers? A 1U server build around the PPC970 may be a killer server.

well... before the XServe, Apple had 0% of this market. Selling just one would increase their market share. A friend of mine went to a Minnesota Wild hockey game and was telling me that the stadium luxury boxes are decked out with Apple hardware which you could use to watch instant replays and call up historical video of the various players. Supposedly the stadium has an XServe data center to host it all.

Re:Eh...

[displaced80]

Read again...

It lacks the 4 curved handles found on current units, but sports a single USB and a single FireWire 800 port on the front of the casing, with additional ports in the rear.

I think the real news...current Mac supports USB2

[adzoox]

FROM: "Following reports of a USB 2.0 chipset in some newer Power Macs, MacNN reader Steve Chung writes about unofficial USB 2.0 drivers for MDD Power Macs: "According to a Korean site Create Mac, it appears that the drivers for the USB 2.0 chipset in the MDD 1.25 and 1.42 Ghz are updated and confirmed that they do work. However, it also says that extreme caution should be taken and that only the two MDD models have been tested. (No liability for damaged systems.) Installation on other models could render the system inoperable."

Also the current Macs with PCI slots or ANY Mac that can run OSX with a PCI slot has been able to support USB 2.0 for almost a year.

Shenanigans

[Anonymous Coward]

I call shenanigans on both AppleInsider and Slashdot for being lame. I'll believe it when I see it.

No news really

[selderrr]

those rumors have been floating around for a few weeks, if not months on other sites. For the wannabe-mac fanatics among yuo : here are other rumor adresses :

macrumors (reliable, good forums)
macosrumors (unreliable, bloated, no forums)
looprumors(reliable, low traffic forums)
thinksecret(reliable, low traffic content, low traffic forums)
macwhispers (reliable, mostly hardware info, no forums)
macslash(slashdot for mac, mostly blahblah)
macbidouille(french, rather new, so reliability unconfirmed)
appleturns(100% reliable news by Steve Jobs's alter ego)

Re:No news really

[jpkunst]

appleturns(100% reliable news by Steve Jobs's alter ego)

Parent was probaly tongue in cheek, but for the uninformed: As the Apple Turns, now back after a long hiatus, is nothing like an actual rumor site. It is an Apple themed 'soap opera', very funny, by a guy with a great sense of humor and writing skills to match.

JP

Re:damn

[Alan Partridge]

I wouldn't worry about it, this is PURE speculation. Apple's hardly going to release SLOWER computers in the future, are they?

Re:Will they still be behind Intel ?

[boaworm]

When you buy a new car, do you look for the one with the engine featuring the highest RPM ? (thougth not)

The comparison is not as stupid as it sounds, your argument is actually a lot worse. Not trying to flame here, but seriously, do you really think the amount of Ghz is what really counts nowdays ?

Re:damn

[Anonymous Coward]

Well, when AMD switched to the Barton architecture for the Athlon XP, they boosted the performance rating to 3000 (and it truely was a 3Ghz equivalent) but dropped the clockspeed a bit (relative to the 2800)

Pet rock.

[Duncan3]

"These initial units will ship with Mac OS 10.2, and hence, will not be optimized for the 64 bit PPC 970 processor. Consumers who purchase these Power Mac G5s will receive a coupon for a free copy of Mac OS 10.3 (Panther), which will ship in September and will be optimized for the new 64 bit processor"

Translation: It's a pet rock until September, by which time production can be geared up and units will actually be available. Of course this will kill all sales until then, so announcing this early would be a very bad idea. So an announcement this early is unlikely.

But... we all know it's coming. Won't be cheap tho, but you get what you pay for in the OSX/Win/Linux world. Over 2 years and still not one crash on my Mac.

USB2 not really surprising

[Anonymous Coward]

A lot of you have been commenting that you're surprised they have USB 2 ports on them. I personally am not surprised to see them - the current MDD G4s have USB 2 ports on them, it's just the drivers in OS X make them into USB 1, you can actually replace the drivers and get nice 800Mbps ports. The fact that the hardware is there does not mean that they will be supported in the OS - it just means that USB 2 ports are cheeper to get hold of than USB 1 ports.

Bob

Re:USB2 not really surprising

[KefkaFloyd]

USB 2.0 is 480mbps. The 800mbps you're saying is probably confusion with FW 800.

Re:damn

[byolinux]

they should release a 23" powerbook, with 16 G4 processors, made entirely out of lead. i wouldnt buy it, but i'm sure Steve Jobs could tell me why I should...

Re:Why doesn't Apple switches to Windows ?

[clifyt]

This is a troll but I'll respond anyways.

Most of us LIKE Mac OSes. Up until Win2000, I wasn't satisfied with Windows OSes...I am a Wind'rs programmer by trade, but I always liked my Macs just a little more. Between the time os OS9 and 2000 -- I was a little torn over what was better...I think Win2k was a lot better than OS9 in retrospect.

BUT soon after that, I picked up the public beta of OSX and haven't been back to 9 since. I was right back in the Mac camp because it meant for once I didn't need to terminal into my Unix servers to get simple things done. On Windows, yeah, I have the Cyrix (err...is that it?) GNU Tools -- but it never felt right or integrated. The interface felt once again in the background to OSX.

Honestly, I wish Apple was a software only company -- The hardware is nice, but this is the area it seems to lag. I use to buy into the The MegaHertz Myth Is Wrong -- but as a programmer, I realized folks should not have to optimize their code for a specific base EACH AND EVERYTIME A PIECE OF HARDWARE CAME OUT. Some apps work with Altivec rather nicely...they can afford to optimize their code. Most of us want to write efficient portable code that can work anywhere. Of course, I do get pissed off when I hear friends talking about code I *KNOW* they've optimized for Windows and then left unoptimized for the Mac and then compare the two...I do a lot of work in the sound design industry and a lot of friends work at companies that make DSP solutions (both massmarket for consumers and the higher end for designing items that will not be of use to many others) -- and I see this all the time. Someone knows how to optimize to the SSE sets on Intel and have no problem tweaking the hell out of it and claiming benchmarks, but throwing the ported code to some monkey that knows only enough about the Mac to be dangerous -- and its embarassing because they then make outrageous statements about the relative speeds.

As for lack of Applications -- I don't know where folks get this. Numberwise -- yeah. Professional app to professional app, we have what we need. Anything in my field has an equivelent analog in both the Mac and the PC world -- with a lot of specialized apps actually being Mac Only (or at the least Mac First) because the creative market still looks at Mac Users as being more in this camp.

Again, I realize your response was a troll, but I felt like educating ya anyways :P

Re:damn

[Alan Partridge]

if you placed it on your lap, it would at least protect your family jewels from most forms of ionising radiation.

That's gotta be worth SOMETHING...

The Way Of Steve

[nwf]

Didn't Steve say some time last year that Apple will never introduce new hardware at WWDC?

Of course, the way of Steve is complex and sometimes contradictory.

Re:Will they still be behind Intel ?

[the eric conspiracy]

but at a certain point wouldn't you want clock speed over architecture ?

No. Intel has shown that you can sacrifice too much chasing clock speed in the case of the P4. Look at the Centrino - the same performance of the P4M at 2/3 the clock speed.

With the G5 we are talking about a 64 bit CPU with clock speeds in the 1.2 - 1.8 GHz range. This is in fact quite competitive just on a clock speed basis with current 64 bit designs from AMD and Intel.

Announcing early not unprecedented.

[phillymjs]

Who's to say Apple will announce it now, but not ship it until Panther debuts? Apple announced and demonstrated the original iMac (IIRC) in May 1998, but did not actually begin shipping until August of that year-- I may not have the dates exactly right, but there were certainly at least two months between announcement and availability. And that was not an instance of Jobs saying "This is available now," but product not shipping until weeks later because they couldn't ramp up production quickly enough. It was a stated two or three month delay from the start.

I think that this time, however, Apple would be doing the right thing to release the G5 ASAP-- that way the hardware will be available during back-to-school time, one of Apple's busiest sales periods. If they do the announce-and-wait thing this time, they'll miss the back-to-school sales. They'll also piss off a lot of people who just blew their wad in August on a G4 with significantly less computing power for about the same money that now buys a G5.

As long as everyone who buys a G5 gets a voucher in the box for a free upgrade to 10.3, I see no problem with shipping the hardware a few months before the OS that takes full advantage of it debuts.

~Philly

Re:No Gigabit Ethernet ?

[The AtomicPunk]

Yes. Despite having moved to it in all their current models, they've decided it's overkill for the consumer and are moving back to 4mbit token ring.

Re:Um, point of order...

[the eric conspiracy]

AAPL currently sits at a price it first set in 1987!

Yah, if you neglect the fact that AAPL split twice during this period of time, the price is the same. In actuality the value of a share of stock purchased in 1987 is 4 times higher today.

Re:Will they still be behind Intel ?

[BinxBolling]

You're misunderstanding the analogy. The point isn't that engine RPM doesn't matter, because the law limits how fast you can go anyways. The point is that engine RPM is only one factor out of many that determine a car's overall performance. Similarly, clock speed of the CPU is only one factor out of many that determine a computer's overall performance.

Re:Um, point of order...

[Lysander Luddite]

Some rumor sites claim the PPC prices will be lower than Motorola's G4. Who knows for sure? I would think IBM would offer the lowest prices possible to speed adoption of the chip. I have also read on the web that IBM's costs for G3s were lower than Motorola's because their production facilities were better. *shrug* If PPC970 truly is 25-35% lower than G4s its a no-brainer for Apple to use them.

If the prices are that low I wouldn't expect Apple to lower prices though. The pro models have been fairly consistent at their current price point for some time. I'm sure other costs have gone up with new features like Hypertransport, Firewire 800 and USB 2.0 (how much does Intel charge for that?). All o fthis si specualtion until the hardware actually comes out of course.

As for the stock comment, prices go up and down. I seem to recall a stock split a few years back. The stock has gone up recently with the iTunes sales announcements. However I think stock price is one of the least indicators about how well a company is doing - and certainly has nothing to do with the price of mac hardware.

Re:Appleturn by Jobs's alter ego?

[selderrr]

What do you mean by that? Is this site done by Steve or what? Or is it a joke? I don't understand it...

No, i was joking. It's not a site done by Jobs himself, but it is a well known fact that many apple employees (supposedl including his steveness) are fan of appleturns. Especially the Polls :-)

Re:damn

[cosmo7]

hey, 1999 called and it wants its troll back.

Re:No Gigabit Ethernet ?

[Anonymous Coward]

You know, the rumor sites haven't said anything about a keyboard, either! Does anybody know whether Apple plans to ship a keyboard with the G5?

Re:Will they still be behind Intel ?

[TheRaven64]

Intel has 3GHZ+ chips out these days. Thats double what the new Mac would have.

It gets worse. The new 3GHz P4 has a thermal output of almost 80W, around four times the thermal output of a . How will Apple cope with a computer which produces so little heat? The average consumer won't care that you can leave a Mac on longer in order to heat up their room, or that they will actually be able to hear their music over the fan noise. All they will see is that the Mac costs more and they have to buy a heater to keeps their house warm as well.

Re:Will they still be behind Intel ?

[drsmithy]

No. Intel has shown that you can sacrifice too much chasing clock speed in the case of the P4. Look at the Centrino - the same performance of the P4M at 2/3 the clock speed.

You seem to be missing the point - what good is the Centrino being just as fast at 2/3 the clockspeed if the P4 still has enough headroom to (say) quadruple it's clockspeed and the Centrino only has enough headroom to double it ?

CPU performance can be increased by (amongst other things) architectural improvements or by ramping clockspeed. Neither, in an of itself, is inherently superior to the other. A CPU that performs twice as fast per clock, but is only clocked at 1/3 the speed, is still slower.

Re:Um, point of order...

[the eric conspiracy]

You also have to take into account share dillution too.

Depends on your perspective. If you are interested in the market cap of the company, total number of shares x market price indicates the valuation of the company is 12x what it was in 1987. If you are interested in shareholder return, then the number is 4x.

Either way it shows that the original post is totally off-base.

Re:"New!"

[Distinguished Hero]

Which wintel motherboards have fw 800 and hypertransport? I'd be interested.

I wasn't really going to comment on this, but since it got modded up 5, Insightful...
nForce, nForce2
You might be further interested in knowing that hypertransport was primarily designed by AMD and is used in all of their Opteron systems, and will also be used in all Athlon64 systems. I guess that's not wintel per se... but it's a PC motherboard nonetheless.

Furthermore, if there is a demand for fw 800 on PCs, they will have it... a small upside of not being at the whim of a single company *cough*Apple*cough*.

Re:Will they still be behind Intel ?

[droleary]

In the computer world, there is no speed limit.

Yes there is: you. Everything external to the CPU limits the computer these days, and responding to human events is like idling at a stop light; your raw RPM doesn't make a big difference. According to procinfo and top, my computers are idle a good 90% of the time. Everyone chasing clock speed really needs to take a step back and instead design an architecture that meets the burst processing pattern that most people have.

The other part of the analogy is not about performance, it's about packaging. You don't buy a car on speed alone. There are styling and comfort factors, and suitability to a purpose. What's really amazing is that Apple is one of the few that understands that; you'd think PC builders would be more inclined to do that sort of thing in order to differentiate themselves from all the other clones that are on shelves.

Re:stock prices

[Oswald]

I'll reply to you, because you were polite, but it's intended as well for the geniuses who can't read a stock chart and want to tell me about stock splits.

First, my original post is correct: adjusted for splits, AAPL is stagnant since 1987. This chart clearly shows it. Notice that the blue line is upbroken. If it were not adjusted for splits (ALL charts are adjusted for splits, but apparently you don't know WTF you're talking about, so I'll inform you), the blue line would take have a break (down to half the preceding day's price) every time there was a split. So, nyah, nyah, nyah.

Second, in reply to this particular post, stock price is always indicative of a company's perceived prospects, relative to the other financial opportunities available. The fact that AAPL never makes any sustained headway is due to the fact that they have had NO sustained increase in earnings. Their profits rise and fall with their hardware cycles, but they never improve market share or margins for long.

I was trying to use a widely-accepted proxy for a business's past success and current prospects--its stock price--as a quick way to make my point, but apparently people just took it as a red herring. Let me be more specific: Apple Computers, Incorporated has been struggling for over a decade--poor profit picture, poor market share. I don't actually give a shit about Macs one way or the other. I was only pointing out that it's important for the company that these machines succeed (i.e. turn a profit and increase market share), and to do that they must be reasonably priced.

(To those who posted that the 970 may actually be cheaper than the Moto chips, I thank you for the information. We'll know shortly, I suspect.)

Rumors and more

[customjake]

Personally, there are a lot in the PC community who are crying "FOUL" I personally believe that the 970/980 will bring Apple back to the fastest computer title, and PC users don't like the fact that they're using a server processor to do it. I personally think it's great!

Yeah, it's a rumor, but this is a pretty substatiated rumor, that i think we can all agree is happening. One thing still in dispute, is if the new processor is gonna be called a G5 or not. I'm sure Apple wants to get away from the image that the motorola processor havee generated over the last few years.

I would expect to see the PPC 970 at WWDC, or shortly after, i.e. August. As for USB 2, it's coming. Apple has already started using USB 2 cards in its powermac lineup (just not supported by the OS). As for apple trying to catch up in the Mhz race, i don't see this. IBM is the one who's set the Mhz of the 970. I also agree with many rumor sites, stating that the 970 will not be any more expensive that the current G4 lineup. Apple is the only company getting anything based off the G4 motorola line, but IBM currently builds the 970 for it's own blade servers, thus they don't have to gear up just to make chips for apple.

Yeah, the P4 is up over 3Ghz, but looking back, crays are still uber fast, and they don't run ungodly mhz......

Also, i wouldn't count on Apple calling it a G5, as apple might go back to calling their chips by their developed name...ie-970

As for 10.3 and the 64-bit stuff, the 64 bit only comes into play when you start getting 64 bit software to run on the machine, that's why Panther is so big, it'll be a 64 bit OS. Also expect a 64-bit version of Project Builder to help move to 64-bit apps.

AMD is not the founder of hypertransport...They are part of a group who's developing it, and one of the last members to join if i recall...... And I don't think that the transition of an AMD chip is much more complicated than you make it out to be....

Personally, if the idea of a 970 makes your blood boil, wait until WWDC and make an informed choice...if you can't wait to buy a mac, but it now.....

How cany anyone say that the 970 is behind AMD/Intel? Last time i looked, IBMs own 970 is FASTER than the new opterons, aren't those supposed to be fast?

Sorry for the sarcasm, but i find that PC users bash what they don't understand. Apple is heavily imitated by the PC world, so the must be doing something right. Let's just all watch and see what happens at WWDC, and talk about it later. Gossiping about new Mac Hardware......$Free Writing Cocoa apps that screa.........$Free Showing your PC friends how must faster your PPC 970 is over their WINTEL box......$Priceless

Reminders of ALPHA hype

[JDizzy]

Back in 96' or 97' I can recall a bunch of hype in the public markets for the infamous DEC ALpha. I can recall banner adds here on slashdot for "64bit power" and other advertisments basically to the effect of "my processor is bigger than yours" type stuff. The difference now is that the market seems slightly more ready for 64 bit computing as more than 2 vendors are selling 64 bit systems. Intel (ia 64), IBM (ppc 970), Transmeta (128bit/2 core), MIPS, AMD, and I think you can still buy a new Alpha from HP still. I suspect the market still isnt' ready for 64 bit computing, but the saturation of vendors trying to be the one wwho actually makes penetration, like sperm on the egg of the consumer market. Apple is probably the most end-user'ish vendor on the market with very little server penetration, and this is promising news. Most of the other 64 platforms go the way of awsome servers. Apple has the chance to sell systems to mac-heads who would do anything to recapture their former elitness geek glory of years gone by. The onyl way 64 bit system will work ijs if they are compatible with the 32 bit software, and yes I mean the OS + user apps. This is why Apple, and AMD have an advantage. Intell seems to have the notion that since it is the market leader that it can simply force a new architecture down our necks, and the market has decided otherwise, and Intel hasn't lived up to its own expectations either. Time will tell is the IBM incarnation of the PPC is going to make it, and Apple has a history of over pricing their gear. If they could get their systems down to the average price of $1200 usd, then they would have a chance.

Re:Reminders of ALPHA hype

[multiplexo]

Back in 96' or 97' I can recall a bunch of hype in the public markets for the infamous DEC ALpha.

What was so infamous about the DEC Alpha? I worked for a large e-commerce company that used AlphaServers from the AS1000 up to the big 8400s and they were fast, solid boxes with great storage options. Having 64 bits available for databases was nice and the megahertz of these systems wasn't that bad either. Plus the fact that you got Tru64UNIX which despite some annoyances (most notably problems with AdvFS) had some nice features and was far more pleasant to work with than any variant of Slowlaris that I ever touched. The university where I worked also used a bunch of DEC hardware for number crunching, they were quite happy with them. As far as I can see the Alpha wasn't hyped, the 8400 with a bunch of Storageworks BA-370 arrays smoked everything on the market at the time. It's a pity that DEC's marketing department was run by the people who weren't smart enough to be in their engineering department, otherwise the Alpha architecture might still be alive instead of being discarded by HP.

what is the processor price point?

[The Lynxpro]

I've read several rumor sites myself, and I've read that the PPC970's manufacturing price point is actually cheaper than the existing chips Apple uses. So if this is true, it raises serious issues with Apple. 1. Apple needs higher clock speeds to remain competitive in the minds of Joe Consumer and Joe IT Worker (see #2). 2. If the PPC970 is cheaper to manufacture and consumes less power than the existing G3 and G4 chips Apple computers feature, then the PPC970 needs to be implemented immediately throughout the Mac line. 3. Abruptly phasing out all G3/G4 machines (#2) would kill sales of existing units on the shelves. 4. Apple would want to offer the PPC970 at the top end to enjoy large profit margins from early adopters before implementing the 970 throughout the entire Mac line. The greater good requires Apple to incur short-term losses (think existing G3's and G4's in the stores) in order to leapfrog the entire PC market by offering 64 bit solutions top-to-bottom in their product line. It is crucial Apple comes out ahead of AMD's consumer 64bit offerings. But because of #3 and #4, Apple will probably choose otherwise... If Apple were smart, they'd start off with a single 1.4 ghz PPC970 in the eMacs and iMacs, and then work their way up the PowerMac ladder with dual (or even quad) processors up to 1.8 ghz. Afterall, it would be easier for $7/hr. sales employee at Worst Buy explaining why Joe Consumer should pick a 1.4 ghz 64-bit PPC970 powered eMac over a 2.5 or 3.0 ghz P4 equipped PC than it would continuing to argue the merits of the G4 line...

Re:what is the processor price point?

[willtsmith]

Apple needs higher clock speeds to remain competitive in the minds of Joe Consumer and Joe IT Worker (see #2).

Apple doesn't have a prayer of competing in the Joe IT Worker space. Wintel rules business operations, Unix still has a mindshare in web services, Linux is making a LARGE dent in web services, Mac is a non-player.

The first job of any Mac today is to PRESERVE market share in it's core demographics, Elementary Ed, Higher Ed, Graphics Artists, Publication, Videography. BTW, did you know that many hollywood and HBO/Showtime movies are edited with Final Cut Pro????

A PPC970 machine would obviously be a higher end offering for high-end customers that need serious CPU horsepower. Most of the market doesn't NEED 64-bit and probably won't take advantage of it for quite some time.

In the higher ed space, I see that a 64-bit processor could become a great tool for researchers who want to do simulation work on their desktops (and notebooks).

This is especially relevant as IBM starts putting PPC970 CPUs into super-computer rigs running linux variants. It becomes feasible to run the same large scale programs (at a dialed down resolution) on a G5 OSX Mac.

simple

[chocolatetrumpet]

It's very easy.

The processors will continue from G5 to G9, and eventually to G9.2. The next processor after that will be GX 10.0.

Got it?

Re:No Gigabit Ethernet ? Yes

[Brett Johnson]

Not quite. Hypertransport is a high speed bus used to connect the CPU to the peripheral chips. There will still be an ethernet chip, and a firewire chip, but they will live on a Hypertransport bus rather than a PCI or PCI-X bus.

HyperTransport technology transfers data at 12.8 Gigabytes per second. It is designed to be approximately 48 times faster than PCI, 12 times faster than PCI X and 10 times faster than 4-channel Infiniband.

The current G4 suffers from a severe bus bandwidth bottleneck. This is an on-chip problem, so no fancy peripheral chips can rectify it. This is why the current DDR PowerMacs don't see the significant benefit that DDR technology should provide. In most current P4/Athlon/G4 performance comparisons, the G4's lagard performace can be much more attributed to its poor memory bandwidth than it's core clock speed.

Although initial 970 core clock speeds don't seem to be significantly greater than the current G4, its peripheral interface bandwidth is lightyears ahead. Hypertransport would help the 970 sing, significantly improving its throughput. Hypertransport would be wasted on a G4. It would be like having a superhighway run by your city, but your on/off ramps are potholed dirt tracks with metering lights.

Re:I didn't like it

[Dylan Zimmerman]

Well, by "intuitive", you probably mean "like Windows". Mac OS is far more intuitive to people who have never used computers before.

Similarly, you're used to having the Windows right-click. Apple thought about adding that and decided that it would make more sense to have a single mouse button and give it modifier key support. Think about the mouse as having a key instead of a button. I know, it doesn't seem to make much sense, but I find that the Windows way makes much less sense to new users.

Also, remember that the Mac OS has built-in support for something like twelve mouse buttons. You just have to get a mouse with more than one.

As for hosting a web site, surely you don't use IIS for that, do you? OSX has all sorts of great server software like Apache. You just have to install them and turn them on.

For surfing, OSX has easily the coolest browser that I've ever seen. Safari beats Mozilla hands down in speed and it's more standards compliant than IE. Essentially, it's everything that Mozilla Firebird is, but it's built by the people who made the OS.

Macs are widely acknowledged to be the best computers for all sorts of multimedia stuff. If you want to edit video, there's iMovie, CinePaint, Final Cut, After Effects, and loads of other tools. For stills that could be used on a website, you have all of the standard tools like Photoshop, Illustrator, InDesign, Page Maker, Painter, and CorelDRAW, and the myriad Macromedia applications and quite a few that I've never seen for Windows such as Combustion. For audio creation and editing, you can use Logic, Deck, Cubase, Peak, Reason, and Spark, just to name a few. For 3D modeling, Maya is the only one that I know about, but I don't exactly research that.

If you just meant playing multimedia, I have found that QuickTime and iTunes do a far better job of that than Windows Media Player.

I really wish that I had a Mac, but I'm WAY to cheap to actually buy one new.

Re:No Gigabit Ethernet ?

[pHDNgell]

so tell me, why do YOU need gigabit ethernet on a workstation?

Because my system acquires DV data far faster than I can transfer it over 100Mbps ethernet.

Re:No Gigabit Ethernet ?

[Zork the Almighty]

Actually the computer will come with one big button. Press this button and it will do everything.

Re:No Gigabit Ethernet ?

[Knife_Edge]

Well, they discontinued that particular model in January 2001. It seems they were testing the waters with that one, and had not yet fully made the transition (all models shipping at that time did not have it). Fast forward to the present - All currently shipping PowerMac G4s have built-in gigabit ethernet. I doubt it will ever go away now. Curiously, the Apple Store does not mention this as a prominent selling point, placing more emphasis on the modem and airport extreme.

Don't forget CARS...

[alispguru]

If you're going to mention As The Apple Turns, you must also mention its evil twin site Crazy Apple Rumors.

I visit them both daily.

Re:I didn't like it

[scrod]

if I could get one that ran Windows or RedHat (or Mandrake) I'd think about it

Damn, yeah, it's too bad you can't run mandrake on them.

Or SuSe.

Or Debian..

Fuck, they don't run anything, do they?

Re:No Gigabit Ethernet ?

[mjpaci]

A lot of Macs used to NOT ship with keyboards. That was a big bitch point of PC users. (Jeez, your Mac doesn't come with a keyboard? Apple is sucking you dry!) Then Apple started shipping every Mac with a keyboard and the PC users started razzing Mac users that since their machine came with a keyboard, they didn't have as much choice as a PC user building his own box from Chernobyl debris.

It just goes to show you that there are no winners in the Mac/PC debate, just whiners. (On both sides).

--Mike