[Cipher13]

...even the chips we have now are very fast. Dual 1.25GHz G4's - very nice.

I don't see why everyone's looking forward to the 970 so much. Have people considered the fact that Apple REALLY need to get their act together to make a motherboard thats gonna be able to feed this thing?? They can't even feed the current machines...

Seriously, the mobo is a bigger problem than the damn processor.

[Superchicken]

yes but it's more fun to bash motorola cause they won't give us high numbers... and you can brag about numbers... can't brag about mobos... no one nkows what those do!

[MindFad]

Won't the bus be 900MHz per processor, PLUS DDR support? This is my understanding, but please correct me if I'm wrong. A bus effectively at the processor speed? Isn't that a good thing? And I completely agree -- dual 1.25s are monsters -- but they would be even greater if they had the buses on them to make them more effctive. But isn't that Moto's fault? Either way, I'm still excited about a PPC970 in my next Power Mac. Apple needs to optimize and get every kink out of OS X first. You know how I love to whine about scrolling.

[suhail]

Originally posted by Cipher13:
...even the chips we have now are very fast. Dual 1.25GHz G4's - very nice.

I don't see why everyone's looking forward to the 970 so much. Have people considered the fact that Apple REALLY need to get their act together to make a motherboard thats gonna be able to feed this thing?? They can't even feed the current machines...

Seriously, the mobo is a bigger problem than the damn processor.

Wait a minute�
I thought that the current mobo is much better than the previous one, but the G4 cannot take advantage of its MHz. It almost seems that the current mobo is designed for a different processor.

[passmaster16]

Originally posted by Cipher13:
...even the chips we have now are very fast. Dual 1.25GHz G4's - very nice.

I don't see why everyone's looking forward to the 970 so much. Have people considered the fact that Apple REALLY need to get their act together to make a motherboard thats gonna be able to feed this thing?? They can't even feed the current machines...

Seriously, the mobo is a bigger problem than the damn processor.

I agree. That may be an issue. If apple can't correctly engineer the board, they should hire somebody who can. I would hope though with the 970 that they work with IBM to implement a decent mobo design.

[superlarry]

Originally posted by suhail:


Wait a minute�
I thought that the current mobo is much better than the previous one, but the G4 cannot take advantage of its MHz. It almost seems that the current mobo is designed for a different processor.

no, there's definitely a bottleneck in the motherboard, i think in the controller between the memory and the processor. it's bad enough to eliminate the advantages of DDR memory. so the new processors are running idle while waiting for something to chomp on.

[Eug]

Originally posted by superlarry:


no, there's definitely a bottleneck in the motherboard, i think in the controller between the memory and the processor. it's bad enough to eliminate the advantages of DDR memory. so the new processors are running idle while waiting for something to chomp on.

Dammit Jim, I'm not an engineer, but the limitation is a function of the Motorola 7455 chip.

Regardless of the mobo design, the current G4 cannot utilize the DDR bus.

As for the PPC970, I am salivating... Well, it's better than an overclocked 1.25 GHz G4 anyway.

[suhail]

Off topic�
Wouldn't it be great if they got Bill Gates to do a switch ad

[passmaster16]

Originally posted by suhail:
Off topic�
Wouldn't it be great if they got Bill Gates to do a switch ad

Wow you wern't kidding when you said off topic

[D'Espice]

There actually is one huge bottleneck in the current motherboard and microprocessor architecture. And it has to do with DDR and the memory controller.

As we all know, the MDD-PowerMacs switched from SDR-SDRAM to DDR-SDRAM, which effectively doubles the available bandwith. In order to utilize that bandwith though, the data has to be shuffled with DDR-FSB between the microprocessor and the memory.
This access (microprocessor -> memory) is done by the so-called memory-controller which is part of the chipset. The chipset is accessed with the full FSB-bandwidth, which currently is only 133 MHz (MDD867) / 166 MHz (MDD1000, MDD1250). That's half the memory bandwidth, which is why the MDD-PowerMacs are not much faster than the QuickSilver PowerMacs.

FYI: Eug is right, the Motorola microprocessor has to be redesigned in order to be even capable of utilizing a DDR-FSB.

passmaster16: Yeah, suhail does that all the time, best would be to ignore him... you know: Don't feed the trolls

[eddiecatflap]

...would a larger cache get round that problem ?

[D'Espice]

Originally posted by eddiecatflap:
...would a larger cache get round that problem ?

No because the memory would still be accessed with 133/166 MHz instead of 266/333 MHz and that's not very intelligent from an architecture's point of view.

[Un-Inferior]

if only the multitude of great pc motherboard makers like abit, asus, tyan, epox, etc... were competing over designs for apple computers.

[Metzen]

Originally posted by Un-Inferior:
if only the multitude of great pc motherboard makers like abit, asus, tyan, epox, etc... were competing over designs for apple computers.

We'd find ourselves with a bunch of cheap, low quality boards, with their highend boards being of the same price and quality as Apple's

Not even the "great pc motherboard makers" can overcome the great G4 FSB conundrum.

[JNG]

Originally posted by Metzen:


We'd find ourselves with a bunch of cheap, low quality boards, with their highend boards being of the same price and quality as Apple's

Not even the "great pc motherboard makers" can overcome the great G4 FSB conundrum.

Though we'll never know for sure, I think I agree with this. There are considerable engineering challenges involved.

[Eug]

I think most of the Asus boards are pretty good. The Abit ones on the other hand...

Anyways, there is no reason Apple cannot consult 3rd parties (eg. IBM or someone else) for their mobo designs if the needs arise.

[jaisun]

What makes any of you think that the 970 could be used with the current architecture anyway? I'm sure (at the very least) the two parts are pin-incompatible. If that is indeed the case, a mobo overhaul must take place before the processor can even be used. Why would they stop at merely remapping the pins to work with the current arhitecture?

I think Apple knows they're lagging behind in hardware performance...right now they don't have many options since Moto is unwilling/unable to dedicate the proper resources to develop certain technologies (i.e. a DDR-able G4). My guess is that, until Apple can safely ditch Moto, or until they (Moto) pull a miracle fabbing process out of their collective arses, Macs will continue to ship with SDR busses and bandwidth-starved G4s. However, once something akin to the 970 steps up, we should see a whole new generation of Macs...the only thing carried forth from the old (current) machines will be the ability to boot into a Mac OS!! Oh and maybe FW 1 since Apple seems to be slacking on that a bit...

Anyway, that's just my $0.02.

[gumby5647]

I don't see why everyone's looking forward to the 970 so much

I don't think it's the processor that they are looking forward to per se, but more the 900Mhz system bus.

[Superchicken]

is a system bus anywhere near as complicated as a proccessor? And why wouldn't apple be able to make faster ones right now?

[gumby5647]

Originally posted by Superchic[k]en:
is a system bus anywhere near as complicated as a proccessor? And why wouldn't apple be able to make faster ones right now?

Meh, that's the 65 million dollar question right now. Sorry Apple, but there is abosolutely NO reason that the iMac should not be on a 133Mhz bus by now...

[beb]

Who gives a sh-t about Apple. I just want to see a version OS X running on a 970, a Power 4, 5, 6 an AMD or Itanium or a Xeon or a bioprocessor or something that is clearly better and faster than a G4.

I want it to be so unbelieveably fast that Apple has to ship a trust fund with every powermac. "Warning this mac will tear your balls off..."

I want to see the funky going to warp speed colors or the rest of the planet doing the slow-motion jiggy. I want for nobody to say ever again in my entire life that OS X seems slow -including me.

[Scoo]

Originally posted by Metzen:


We'd find ourselves with a bunch of cheap, low quality boards, with their highend boards being of the same price and quality as Apple's

Not even the "great pc motherboard makers" can overcome the great G4 FSB conundrum.

Asus and Tyan make high quality products. Now go drink your Kool-Aid.

[zazou]

You see, it's like this:

The MPX/LogicBoard bus that the current G4s run on is also a bi-product of Motorola engineering. They desigend the bus, et al, more or less and it used across the entire G4 line all the way down the embedded market in one way or another.

The 970 will basically require a whole new LogicBoard to handle all the aspects of the chip, and God willing Faster (than 333) DDR, 8x AGP and every other goodie like FireWire 2 we all crave so badly.

But, in the end, as I said. The CPU, the Bus... The Bus, the CPU.....

...Um, can we get a commment from you Mr. Motorola?

[Simon]

OK, this may be a purely academic question because none if this is going to happen, but what do the pros here think about:

(a) If my dual 867 had a 266MHz bus (with DDR RAM on a FSB @ 533MHz) how much would this influence overall performance? would the G4s be able to show more power, i.e. is the bus speed the bottleneck at the moment?

(b) If my dual 867 had a 900MHz FSB a la PPC970 how much would overall preformance increase? What would be the bottleneck now? G4s or bus?

(c) Is the 7457 the G4 that should support DDR at full speed without the Apple system controller hack? How much benefit will we get from this w/o CPU clock increase?

(d) Will the native DDR support on a G4 increase heat? Cooling issues?

(e) If Apple can't get Moto to ship >1.4GHz, the PPC970 is still a year away and they see slipping sales because of an alleged performance gap to the x86 world what is preventing a top of the line PowerMac and Xserves with quad G4 boards?

(f) If Apple sees itself stuck with the current G4s at not much higher clock for most of 2003 wouldn't it be wise to just do to desktops what they have done to portables - give minor bumps but lower the price substancially. What are the odds of this happening? Would this make it harder for Apple to introduce a PPC970 because suddenly prices would have to leap upward?


And finally, am I the only guy here slowly but surely getting the impression that Steve's babbling about 2003 being the greatest R&D year and new DLDs, etc. could be Apple's way of trying to distract customers from the fact that the Mac is going to be somewhat stuck for the next ten to twelve months? I mean the iPod is great and I'd like something like an Apple branded Treo or even some kind of home didgital set top box to put my MP3s on my stereo and my movie files on my TV, but hell, those are just goodies. Nobody gets serious computer work done with a DLD. I need a Mac to run my code, I can't do 3D simulation code on an iPod - even if Apple comes out with nice new DLDs, what about their main product? Or do you think we'll get steady significant bumps across the line throughout 2003 and Steve will drop in some DLD goodies just for fun? Honestly, I'm asking and not trying to troll around here. Any thoughts on this?

[nvaughan3]

eug, you the same eug I see at anandtech?

[Eug]

Originally posted by nvaughan3:
eug, you the same eug I see at anandtech?

Yep. Hey there...

And finally, am I the only guy here slowly but surely getting the impression that Steve's babbling about 2003 being the greatest R&D year and new DLDs, etc. could be Apple's way of trying to distract customers from the fact that the Mac is going to be somewhat stuck for the next ten to twelve months? I mean the iPod is great and I'd like something like an Apple branded Treo or even some kind of home didgital set top box to put my MP3s on my stereo and my movie files on my TV, but hell, those are just goodies. Nobody gets serious computer work done with a DLD. I need a Mac to run my code, I can't do 3D simulation code on an iPod - even if Apple comes out with nice new DLDs, what about their main product? Or do you think we'll get steady significant bumps across the line throughout 2003 and Steve will drop in some DLD goodies just for fun? Honestly, I'm asking and not trying to troll around here. Any thoughts on this?

I had thought that the 7455 wasn't really designed to go much 1 GHz. ie. Isn't the 1.25 GHz G4 already pushing the limits of the design? If so, then 2003 is a make or break year. (Would a die shrink even help much?) I have confidence that Apple will pull thru with flying colours though, with IBM's help.

My other question is what is BEYOND the PPC970. My impression is that is designed to make it to 1.8 GHz. Sure, the 1.8 GHz PPC970 should just scream on a per-GHz basis, but then what?

As for the Treo, despite being a Handspring fan myself, I don't like the Treo. As for Apple, I don't think it'd go this route since it looks too cluttered. If they did do a PDA (and that's a big IF), I'd expect a much sleeker design.

[D'Espice]

I think that debate is preposterous anyway, the PPC970 is to be shipped somewhen late 2003 which clearly is way too late for Apple. They need something faster in mid 2003 that beats the hell out of the current Motorola microprocessors, clearly not a 100 MHz faster G4.

[Simon]

Originally posted by D'Espice:
I think that debate is preposterous anyway, the PPC970 is to be shipped somewhen late 2003 which clearly is way too late for Apple. They need something faster in mid 2003 that beats the hell out of the current Motorola microprocessors, clearly not a 100 MHz faster G4.

Honestly, I think we'll have to be glad to see a 7457 with true DDR support and maybe a tad higher clock speeds. I really don't expect Moto to give us more until Fall 2003 which is already an optimistic guess for the 970 intro! Do you guys think that as soon as IBM can pump these chips out in volume Apple will sell PowerMacs using them? If IBM goes to volume production of the 970 by Fall 2003 why should we see any Apples using them before january 2004?

OT:

Originally posted by Eug:
As for the Treo, despite being a Handspring fan myself, I don't like the Treo. As for Apple, I don't think it'd go this route since it looks too cluttered. If they did do a PDA (and that's a big IF), I'd expect a much sleeker design.

And that is exactly the reason I'd like to see Apple do it. The Treo is by no means perfect (why the keyboard? why color which kills the batteries?) and actually I'd prefer this or this but I'm absoluely confident in Apple's design team.

[Hornet]

Originally posted by Simon:
(a) If my dual 867 had a 266MHz bus (with DDR RAM on a FSB @ 533MHz) how much would this influence overall performance? would the G4s be able to show more power, i.e. is the bus speed the bottleneck at the moment?

A large influence on overall performance. The G4's would "show more power". The bus speed is currently like the hand brake is still on as the G4 car attempts to let rip at 100mph.

How much of a difference? Well let me give you an example. BadAndy, a poster at the arstechnica forums (yeah I lurk a lot over there, some great info from him comes out), made a comment about this not too long ago. He is an altivec coder, and seems to be quite good at his job, or so the impression is. Thus, he knows the G4 and its problems well. The example...

He owns a G4 533 dual processor. That machine has two (2) altivec units; one per processor. The machine has a bus speed of 133mhz. How much of a bottleneck is that bus? Well, he is able to make A SINGLE ALTIVEC UNIT use the ENTIRE bus, all on its own. What was that? A single altivec, using the entire bus. Never mind the rest of the processor. Never mind the second processor. Never mind the second altivec unit (ie the one on the other processor).

Big problem hey. Well, partly altivec chews through bandwidth, as it does some things quite quickly - but the problem still stands, the G4 bus is a single lane road in the place of an 8 lane highway in LA. Data just cant move through it fast enough. 166mhz barely helps. What is needed is a jump of several times the speed. Do you hear 900mhz ringing in your ear now?

(b) If my dual 867 had a 900MHz FSB a la PPC970 how much would overall preformance increase? What would be the bottleneck now? G4s or bus?

A 900mhz bus is... 6.7x as fast as BadAndy's G4 533DP. A little guessing on my part, would say that a 900mhz bus would not only allow the "rest" of the processor to be active on the bus at the same time as the altivec, but also the second processor and its altivec. The 900mhz bus pushes 7.2GB/sec (or is it 7.6...) of bandwidth - 6.4GB/sec usable on the 970 (the rest is used for other stuff, I cant remember). The G4 has something like 1GB/sec. Altivec at its highest peak, can hit nearly 4GB/sec

Anyone can now see, a 900mhz 6.4GB bus will truely let rip. Big time.

(c) Is the 7457 the G4 that should support DDR at full speed without the Apple system controller hack? How much benefit will we get from this w/o CPU clock increase?

No idea sorry. Problably a bit, as the bus would be 167 -> 333.

(d) Will the native DDR support on a G4 increase heat? Cooling issues?

No

(e) If Apple can't get Moto to ship >1.4GHz, the PPC970 is still a year away and they see slipping sales because of an alleged performance gap to the x86 world what is preventing a top of the line PowerMac and Xserves with quad G4 boards?

Our previous example. 4 processors on a bus so slow that part of 1 can fill it No, thats why you haven't seen a quad processor G4. Thats another great thing about the 900mhz bus - suddenly a quad mac is quite possible, even very practical. More than 4 processor? who knows, the 970 was designed for up to 16 processors.

(f) If Apple sees itself stuck with the current G4s at not much higher clock for most of 2003 wouldn't it be wise to just do to desktops what they have done to portables - give minor bumps but lower the price substancially. What are the odds of this happening? Would this make it harder for Apple to introduce a PPC970 because suddenly prices would have to leap upward?

Its about time apple lower prices by a decent amount. They make >40% on some powermac sales. I think a price cut is more than possible.



Hope that has helped

[Simon]

Originally posted by Hornet:
The 900mhz bus pushes 7.2GB/sec (or is it 7.6...) of bandwidth - 6.4GB/sec usable on the 970 (the rest is used for other stuff, I cant remember). The G4 has something like 1GB/sec. Altivec at its highest peak, can hit nearly 4GB/sec

Anyone can now see, a 900mhz 6.4GB bus will truely let rip. Big time.

Hornet, thanks for the info! While reading your comments on bus bandwidth and processor requirements I became quite puzzled. If I understand correctly Altivec can deliver 4GB/s and we have the G4 MaxBus which can handle about 1GB/s. What the hell were Moto's engineers thinking when they designed that bus? I have two G4s in my Mac which can pump more than 8 times the amount of data that my bus can transport? That's ridiculous. I can't believe Moto's engineers didn't see this problem comming when they designed MaxBus...

[Hornet]

You bet Simon. Truth be told, 4GB/sec is the "peak" altivec hits (like red-lining your car in effect, minus the problem of blowing up the engine). While it is rarely done, it shows off the pure problem of a tiny, 1.1GB/sec 133mhz bus. In truth, a single G4 533 would do fine on a 133mhz bus.

In BadAndy's situation, the second processor was useless. So really, Apple would need... 333mhz bus at least now, to cater not only for the second processor, but overall processor clock increase (thus moving more data around - although I believe altivec remains as fast as it was on G4/350's, rather the "rest" of the processor gets faster - moves more data, more bus usage etc). 400mhz would be more "comfortable". Now flash back, a quad mac on a 133mhz bus. Simply put, the 4th processor would never be used, and the 3rd would hardly ever be used especially when any altivec work was going on (thats just "inventing" a processor priority thing for the sake of this post, saying that "Processor 1 gets top priority for data" etc - I don't know if thats how it works in reality).

You bring up an interesting point. Indeed moto engineers designed Altivec. They knew full well its potential to suck MaxBus dry - let alone a second altivec. My - pure guess - would be that MaxBus was designed in the G3 age (at 66mhz and 100mhz), moved to 133mhz, but never really "leaped" like it needed to with altivec (ie it only incrementally increased over its time - 66mhz in the beige machines, 100mhz in the B&W's, 133 in the digital audio G4, 166 in some MDD models etc). What it really needed was a leap.... 133 -> 900 style

The killer problem however is this. Moto has little need to make a DDR bus G4 (ie moving data at 266 and 333mhz - like the AMD world at least). Its "other" G4 buyers (yes, apple is not the sole buyer, in fact they apparently buy less than half the G4's moto makes - take that with some salt, I cant back that one up at all) are embedded buyers. Case example, one of them is a networking comapany making routers (this I do know, however I forget the name). Lets say they make a new gigabit router. Lets say, a 500mhz G4 can move data around the entire number of ports, with full speed. Thus, 500mhz is whats needed, whats bought, and whats used in several million products. Fast forward to a year later. Gigabit is still the router speed being made. Same number of ports. Funnily enough, 500mhz still does the job 100% fine. The point is, there is little push from Moto's other buyers (who collectively represent more sales than apple, apparently) to push forward. Their 500mhz does just fine. The same is true for DDR - the embedded market that moto supplied sees little need for a faster bus for their needs. In the end, moto sees little point allocating the funds to making a DDR G4 for a single customers use. That unfortunatly is part of the reality apple face when dealing with moto.

Kinda makes it hard for apple.... doesn't it.

This is why the 970 is so important. As far as I know, Apple and IBM will be the only users (well even that is totally based on rumors lol). IBM have the need to push forward. Apple have the need to push forward. The 970 will be able to develop quickly - its clients require it - they do not want/would not mind if it remained stagnant. Thus, the 970 is not only.... a faster chip (big plus)... not only a faster bus (huge huge plus)... but a market, an environment change, which may be the single biggest change that will help them remain competitive throughout the life of the 970 - as there is the need to continue on

OK ramble mode off. That last bit could all be wrong of course if the 970 ends up going to the "slow poke" embedded market. If not, its probably a great, great thing for apple.

[Metzen]

Originally posted by Hornet:
A 900mhz bus is... 6.7x as fast as BadAndy's G4 533DP. A little guessing on my part, would say that a 900mhz bus would not only allow the "rest" of the processor to be active on the bus at the same time as the altivec, but also the second processor and its altivec. The 900mhz bus pushes 7.2GB/sec (or is it 7.6...) of bandwidth -

It's 7.2GB/sec

Originally posted by Hornet:
6.4GB/sec usable on the 970 (the rest is used for other stuff, I cant remember).

Most busses don't use a Point to Point system because it incurs overhead (its packet based), although it is typically regarded as faster. Being packet based incurs a overhead and some latency that is roughly 800MB at peak (400MB up and 400MB down). The effective bandwidth is 3.2GB/sec for AltiVec (streaming).

Originally posted by Hornet:
The G4 has something like 1GB/sec. Altivec at its highest peak, can hit nearly 4GB/sec

Anyone can now see, a 900mhz 6.4GB bus will truely let rip. Big time.

...
Not sure about the peak streaming speed of Altivec, but a Dual G4 1.25GHz (22MKeys) scales linearly with Altivec tests, as well as scaling theoretically with the PPC970 (18MKeys). Because of this, it is evident that Altivec isn't being choked on the bus, if it's being choked at all.

Of course, that is debateable as to whether or not RC5 is keeping data in close proximity using L2 or L3 Cache and not streaming (which is most likely).

Originally posted by Hornet:
Our previous example. 4 processors on a bus so slow that part of 1 can fill it No, thats why you haven't seen a quad processor G4. Thats another great thing about the 900mhz bus - suddenly a quad mac is quite possible, even very practical. More than 4 processor? who knows, the 970 was designed for up to 16 processors.

It's not the 900MHz (450MHz effective oneway) bus that makes this possible as much as the Point to Point bus. Of course this is all dependant on whether or not you'll need a seperate "Northbridge" for each processor or not (AMD-Style MP) or if the 970 will share the bus(G4 MP).

Originally posted by Hornet:
Its about time apple lower prices by a decent amount. They make >40% on some powermac sales. I think a price cut is more than possible.

Apple makes about 30% on the sales of there machines (this is total computers, not individual). Undoubtedbly, that markup goes up if you buy RAM through the Apple store

For Apple though, I'm quite pleased with their prices in this tough economy.

[cowerd]

The 900mhz bus pushes 7.2GB/sec (or is it 7.6...) of bandwidth - 6.4GB/sec usable on the 970

Thats 3.2GB/sec each way. The figures quoted are total bandwidth and the 970 interconnect is 2 uni-directional busses with 3.2GB/sec bandwidth each.

[Eug]

Apple makes about 30% on the sales of there machines (this is total computers, not individual). Undoubtedbly, that markup goes up if you buy RAM through the Apple store

For Apple though, I'm quite pleased with their prices in this tough economy

Well, you're pleased with the new iBooks' prices partially because Apple has slashed prices. The same holds true for other Macs, esp. the previously overpriced PowerBook. Thus, there's no way that the 30% number is still true.

Back to my previous question: Is it true that the 7455 was only really designed to hit about 1 GHz-ish?

[iChristopher]

And that is exactly the reason I'd like to see Apple do it. The Treo is by no means perfect (why the keyboard? why color which kills the batteries?) and actually I'd prefer

As a Treo owner I had to jump in here... I LOVE my Treo. I have been an OmniSky, Samsung I300, and Treo user. The Treo is the best gadget I've purhcased in a long time, I rate it up there with my Tivo and iPod as devices that make me wonder how I ever did without them.

The keyboard on the Treo is very handy to me, it makes mobile email a reality for me. I would never do email with Graffiti or a phone keyboard, WAY too slow and clunky.

The color screen makes surfing with Balzer a pleasure. The world is in color, why not your PDA? The color is key for the Internet apps in my opinion, and it's just much more pleasent to work with. Color can also make the UI easier to navigate in small ways, but it's the small things that make a difference.

The speed this things surfs at isn't mind blowing, but it did exceed my expectations and is fast enough that I'm actually willing to pull it out of my pocket and say look up a phone number instead of calling Directory Assistance. It's fast enough that it's practical to use, and that's more than any other mobile device has ever done for me.

I was in a coffee house last week with a prospect and was able to surf the Internet on my TiBook by connecting the Treo via its' USB cable. All the Sprint plans have unlimited data now, so I was able to show him a bunch of demo sites and the speed again was fast enough that it was usable.

Email and SMS are also very practical on this device. With Snapper Mail I was able to send my Word format resume to a potential employer from my car yesterday. I use SMS all the time now to send and recieve urgent brief messages to colleagues.

The Treo is a best in class device hands down. I know it because I actually USE it. If you are on the fence regarding Treo, just go get one. It comes with an ear bud, car charger, and has vibrate and speaker phone modes. It all works great!

You might think I work for Sprint or Handspring, I don't! I am just very pleased with my new toy, a practical toy that I have zero regret spending $500 on.

[iChristopher]

Apple is selling ancient technology in the iBook, it does not surprise me they can sell them at those prices.

The iBook should be G4, and I suspect it will be sooner rather than later.

Originally posted by Eug:
Well, you're pleased with the new iBooks' prices partially because Apple has slashed prices. The same holds true for other Macs, esp. the previously overpriced PowerBook. Thus, there's no way that the 30% number is still true.

Back to my previous question: Is it true that the 7455 was only really designed to hit about 1 GHz-ish?

[raferx]

Jaguar is still a dog on my buddy's 700MHz iBook... a G3 just doesn't cut it.

[Metzen]

Originally posted by Eug:
Well, you're pleased with the new iBooks' prices partially because Apple has slashed prices. The same holds true for other Macs, esp. the previously overpriced PowerBook. Thus, there's no way that the 30% number is still true.

I vaguely remember reading it had dropped to around 25%, which is a pretty big slash, but that 30% figure was all I could find that was fairly recent and relatively quickly.

Originally posted by Eug:
Back to my previous question: Is it true that the 7455 was only really designed to hit about 1 GHz-ish?

Motorola's original roadmap said 1GHz+

Originally posted by
iChristopher:
Apple is selling ancient technology in the iBook, it does not surprise me they can sell them at those prices.

Radeon 7500, 800MHz raw processing power, up to 6 hours battery life...

Seems to hold it's own fairly well, and at a reasonable price. I'd take a $1600 (CAN) 800MHz G3 iBook vs. a $2350 800MHz G4 iBook.

Originally posted by
iChristopher:
The iBook should be G4, and I suspect it will be sooner rather than later.

And I predict speed bumps coming up for Apple's whole product line for the next year.

Originally posted by raferx:
Jaguar is still a dog on my buddy's 700MHz iBook... a G3 just doesn't cut it.

I'm on a 400MHz iMac DVSE here, and while it is slower than OS9, it is still pretty speedy and responsive.

I have to use 800MHz Athlon (w/256MB of RAM) and 450MHz PIII's (w/256MB of RAM) at school. Both are running Windows 2000 Professional/Server...

You have no right to complain about a sluggish GUI. On OSX

[Eug]

Seems to hold it's own fairly well, and at a reasonable price. I'd take a $1600 (CAN) 800MHz G3 iBook vs. a $2350 800MHz G4 iBook.

There is no $1600 (CAN) 800 MHz G3 iBook. But I'll let that one slide.

Anyways, that's precisely the issue. Why the huge price difference? Certainly not because of chip costs. You present that scenario because you know that that is what Apple is doing with the G4 in PowerBooks.

While I might not want to spend $750 more for a G4 over a G3 and no other changes, I'd spend $300 more. ie. I'd take an 800 MHz G4 iBook at CAD$2400 over a G3 800 MHz iBook at CAD$2100. Indeed, I'd take that over a CAD$4800 G4 PowerBook too. (And that's what Apple is afraid of.)

[Metzen]

Originally posted by Eug:
There is no $1600 (CAN) 800 MHz G3 iBook. But I'll let that one slide.

Whoopsie! I meant 700MHz iBook... But it seems you caught that

Originally posted by Eug:
Anyways, that's precisely the issue. Why the huge price difference? Certainly not because of chip costs. You present that scenario because you know that that is what Apple is doing with the G4 in PowerBooks.

While I might not want to spend $750 more for a G4 over a G3 and no other changes, I'd spend $300 more. ie. I'd take an 800 MHz G4 iBook at CAD$2400 over a G3 800 MHz iBook at CAD$2100. Indeed, I'd take that over a CAD$4800 G4 PowerBook too. (And that's what Apple is afraid of.)

Either way, the iBook is a good value computer. The G3 is on it's 3rd generation itself, 4th if you don't include the change to copper (750 -> 750CXe -> 750FX). To say that it is running on dated technology...

The 750FX only has a 5 stage pipeline. It's up to 1000MHz on a 5 stage pipeline. That is nothing short of miraculous. The only thing it doesn't have that the G4 has is AltiVec. The 750FX should be better in virtually every other way. The 750FX has support for a 200MHz bus, software overclock/underclocking on the fly, it's fast, it's extremely effiecient, but it lacks Altivec.

It's amazing how much Altivec means to people now.

Sadly, Altivec truly commands that much of a difference.

[Simon]

Originally posted by Metzen:
The 750FX only has a 5 stage pipeline. It's up to 1000MHz on a 5 stage pipeline. That is nothing short of miraculous. The only thing it doesn't have that the G4 has is AltiVec. The 750FX should be better in virtually every other way. The 750FX has support for a 200MHz bus, software overclock/underclocking on the fly, it's fast, it's extremely effiecient, but it lacks Altivec.

This is entirely speculative but what are the chances Apple was able to convince IBM to build a higher clocked 750FX with Altivec? Think of this as filling the gap up to the time we get the 970. There was some tlak about licensing Altivec from Moto and I'm also sure that Apple saw the G4 problem coming since Moto told them that they were stuck on the 7500. What if IBM would build a 1.5+GHz 750FX with Altivec running on a 200MHz (400MHz FSB w/ DDR) bus? Wouldn't that be a sweet CPU to keep us buying PowerMacs till the 970 arrives in about +1 year from now?
Or is this bull because the 750FX can't do SMP like the original 750? Could this be added to a hacked up 750FX?

[Metzen]

Originally posted by Simon:
This is entirely speculative but what are the chances Apple was able to convince IBM to build a higher clocked 750FX with Altivec?

IBM doesn't care about Apple with the 750FX.

Originally posted by Simon:
Think of this as filling the gap up to the time we get the 970.

Maybe Apple thinks of it like that, but I'm sure IBM doesn't.

Originally posted by Simon:
There was some tlak about licensing Altivec from Moto and I'm also sure that Apple saw the G4 problem coming since Moto told them that they were stuck on the 7500.

Rumor, speculation. Odds are strongly against any of that happening. IBM doesn't need to license anything from Motorola, Apple, IBM AND Motorola all sat in the Austin Texas fab and designed Altivec, VMX, Velocity Engine jointly.

Originally posted by Simon:
What if IBM would build a 1.5+GHz 750FX with Altivec running on a 200MHz (400MHz FSB w/ DDR) bus?

Won't happen. The 750FX is one of IBM's babies. It's not meant to be a highend chip. It's meant to be a low-power, low-cost, high-performance chip. And that's exactly what it is. It's target is the embedded market, the G4 is targetted for the highend embedded market. The lowend has more volume (as is typically the case). If you think Apple buying roughly half of Motorola's G4's means Motorola doesn't give a crap about Apple, Apple probably buys a much smaller fraction of 750FX's from IBM to make it worthwhile.

Originally posted by Simon:
Wouldn't that be a sweet CPU to keep us buying PowerMacs till the 970 arrives in about +1 year from now?

I speculate the 970 will be here in less than a year. Till then, the G4 is doing quite fine.

Originally posted by Simon:
Or is this bull because the 750FX can't do SMP like the original 750? Could this be added to a hacked up 750FX?

You ain't hack'in no chip!