PowerBook To Get Dual-Core G4?

[macsrus]

Good testing. Just one question, though.

Did you test both 32-bit and 64-bit SuSe? (Or are there both for the PPC - I couldn't find much PPC info on the SuSe website)


If your threads don't need more than 31-bits, in theory it shouldn't make much difference to G5 optimized code - both 32-bit and 64-bit should be about the same. (If you app uses lots of pointers, the 64-bit could be somewhat slower than the 32-bit.)

Thanks in advance...

Enterprise Server 8 and 9 both have PPC versions....
to get 32 bit OS I modified the kernel and did a custom compile...

I got my copy of SUSE from the NOVELL developer program....
The SUSE Enterprise Server 9 is a release candidate 5


My code does need more than 2 GB's per thread but only during decomp and recomposition.....

Currently my problems are decomposed and recomposed on an the following systems.... IBM SP, Intel Itanium, AMD Opteron, SGI Origin......

After decomp we run the code on Either the IBM SP or on any of the clusters we have... the move the answers back to one of the Listed 64 bit systems....
Currently we need 13 GB ram on a single thread to do the decomp/recomp

 

[VIREBEL661]

Good news (if it really happens - guess we'll see in October). I still think the G4 is a really phenomenal processor that never developed into what it should've been. I still think the tech is sound and that they can push this processor forward to be a cutting edge portable processor (even though Apple's laptops rock as they are). Let's hope they're interested in doing business and not drop the ball this time.

 

[gskiser]

I couldn't agree more SiliconAddict

Do people pull this stuff out of thin air?!?!

Because then two years down the road when everything is optimized for the G5 (Note I said G5 not 64-bitness.) our laptops are going to look pathetic. Thanks but if I'm spending close to $4,000 on a laptop it isn't going to contain a CPU that in all likelihood is going to be EOLed by next summer.


Apple will be screwed if they don't get a G5 in a PowerBook by summer at the latest. Realisticly Moto won't be delivering a dual core CPU anytime soon. If we are talking another year with a craptastic G4 in its current state the PC world is going to make chopped dog food out of the current G4's, and that's even if Moto can squeeze a few more speedbumps out of the current gen G4. Sorry folks but as I have harped on before the Pentium M and new AMD mobile CPUs are making Apple's mobile hardware look... 🙁

SiliconAddict hit the nail on the head. I'm so tired of people saying "the G4 has more than enough power for anyone" and "we don't really 'need' a G5 PB yet". Maybe some of us actually look further than six months down the road when we make a $3,000 purchase. The fact is, I average 5-6 years out of my computer. I'm currently on a 400mhz G3, and have finally outgrown it. I know I don't 'need' a G5 Powerbook right now, and a G4 would suffice for the next few months. However, in <2 years, I don't see that still being the case. Apple has committed to the G5, 64-bit, Tiger, etc... The fact is, the G4 PB is probably going to be EOLed soon. I want a processor that has SOME potential down the road and that can take advantage and fit into the direction Apple is going.

The G4 is still capable of doing everything I may need today. But why in the world would I buy a chip which was introduced in 1999 and is obviously on its way out. I'd rather by a 1 rev. G5 powerbok, and get AppleCare, than I would buy a last revision G4 powerbook. While it may be reliable, its not going to be on Apple's radar in a year. They've made it clear, the G5 is their future, Tiger is on its way in early '05. Why spend $3000 on a G4 powerbook that you already know isn't going to be capable of using an already anounced operating system its fullest potential? Besides, not matter how much faster they keep making the G4, the FSB sucks.

The G4 PB is fine for people who upgrade their computers every year. However, for the average person who needs their computer to last more than a year, its a waste of money. Dual core would be nice, but it still doesn't solve the above problems. At this point and time, with the exception of the PowerMac G5, Apple's entire hardware line looks pretty pathetic when compared to the competition; especially their portables'.

I'll end with on of my favorite quotes from Apple's website where they tout the G5 and blast the 167 FSB that holds back even the fastest processors. Too bad their own G4 1.5Ghz Powerbook falls into this class; but of course they don't mention that...

"Bottlenecks in the system architecture are usually what hold back the fastest processors. PowerPC G5 features an industry-leading 1.25GHz frontside bus for each processor, offering a staggering 20GBps throughput on dual 2.5GHz PowerPC G5 systems. That’s a huge leap over the Power Mac G4, with a bus speed of 167MHz. That means you won’t have a bottleneck getting information to the chip for processing."

 

[inkster218]

Dangitt you guys.

I was about to buy a new G4 PB after Paris and now you got me thinking about the whole "$3k laptop purchase why not wait until the G5" thing.

Hornswaggle reggle flekkin schmekkin.

 

[pigwin32]

Currently we need 13 GB ram on a single thread to do the decomp/recomp

When I was a lad we were happy to have 256KB and in that we had to run a SMP OS, email, word processing, editor, compiler. You young people just don't know how lucky you are, 13 GB on a single thread, mutter mutter.

On a positive note, I'm pleased to see some rumour action on the G5 PB front. I'm keen to replace my TiBook, not because it's not still useful, but because I want some more speed, a superdrive, plus bluetooth, and hopefully a higher resolution LCD. But I'm not buying another G4 and I'm not leaping into Intel, I do that for a day job. I would seriously consider a dual-core G4 but it really sounds like an interim solution for Apple if they are considering it at all. Tiger is going to be tuned for the G5, I want a G5 PB. Of course if IBM were to release a G5 laptop with say Yellow Dog Linux bundled, that could work for me 🙄

 

[MikeBike]

You have no idea what you're talking about.

SiliconAddict hit the nail on the head. I'm so tired of people saying "the G4 has more than enough power for anyone" and "we don't really 'need' a G5 PB yet". Maybe some of us actually look further than six months down the road when we make a $3,000 purchase. The fact is, I average 5-6 years out of my computer. I'm currently on a 400mhz G3, and have finally outgrown it. I know I don't 'need' a G5 Powerbook right now, and a G4 would suffice for the next few months. However, in <2 years, I don't see that still being the case. Apple has committed to the G5, 64-bit, Tiger, etc... The fact is, the G4 PB is probably going to be EOLed soon. I want a processor that has SOME potential down the road and that can take advantage and fit into the direction Apple is going.

The G4 is still capable of doing everything I may need today. But why in the world would I buy a chip which was introduced in 1999 and is obviously on its way out. I'd rather by a 1 rev. G5 powerbok, and get AppleCare, than I would buy a last revision G4 powerbook. While it may be reliable, its not going to be on Apple's radar in a year. They've made it clear, the G5 is their future, Tiger is on its way in early '05. Why spend $3000 on a G4 powerbook that you already know isn't going to be capable of using an already anounced operating system its fullest potential? Besides, not matter how much faster they keep making the G4, the FSB sucks.

The G4 PB is fine for people who upgrade their computers every year. However, for the average person who needs their computer to last more than a year, its a waste of money. Dual core would be nice, but it still doesn't solve the above problems. At this point and time, with the exception of the PowerMac G5, Apple's entire hardware line looks pretty pathetic when compared to the competition; especially their portables'.

I'll end with on of my favorite quotes from Apple's website where they tout the G5 and blast the 167 FSB that holds back even the fastest processors. Too bad their own G4 1.5Ghz Powerbook falls into this class; but of course they don't mention that...

"Bottlenecks in the system architecture are usually what hold back the fastest processors. PowerPC G5 features an industry-leading 1.25GHz frontside bus for each processor, offering a staggering 20GBps throughput on dual 2.5GHz PowerPC G5 systems. That’s a huge leap over the Power Mac G4, with a bus speed of 167MHz. That means you won’t have a bottleneck getting information to the chip for processing."

Aside from the 64bit registers, what is APPLE exactly going to optimize for?
- Apple will improve the GCC 3.5(?) compiler with advanced features, which will benefit both the G4 and G5.
- Apple will move more code to Altivec? G4 and G5
- Apple may write better graphics drivers: Depends upon the graphics chip.

The big advantage of the G5 was that it could scale to 2.5 ghz,
whereas the G4 is still at 1.5. But, clock for clock the G4 is more powerful.
A dual core G4, in a Powerbook, is a hell of an improvement, esp. since we probably won't see a G5 in a powerbook till the 65 nm process, and that would still be only a 25% performance improvement.
Vs. a dualcore G4 100% improvement, at least with the stuff I run.

 

[manu chao]

SiliconAddict hit the nail on the head. I'm so tired of people saying "the G4 has more than enough power for anyone" and "we don't really 'need' a G5 PB yet". Maybe some of us actually look further than six months down the road when we make a $3,000 purchase. The fact is, I average 5-6 years out of my computer. I'm currently on a 400mhz G3, and have finally outgrown it. I know I don't 'need' a G5 Powerbook right now, and a G4 would suffice for the next few months. However, in <2 years, I don't see that still being the case. Apple has committed to the G5, 64-bit, Tiger, etc... The fact is, the G4 PB is probably going to be EOLed soon. I want a processor that has SOME potential down the road and that can take advantage and fit into the direction Apple is going.

The G4 is still capable of doing everything I may need today. But why in the world would I buy a chip which was introduced in 1999 and is obviously on its way out. I'd rather by a 1 rev. G5 powerbok, and get AppleCare, than I would buy a last revision G4 powerbook. While it may be reliable, its not going to be on Apple's radar in a year. They've made it clear, the G5 is their future, Tiger is on its way in early '05. Why spend $3000 on a G4 powerbook that you already know isn't going to be capable of using an already anounced operating system its fullest potential? Besides, not matter how much faster they keep making the G4, the FSB sucks."

You bought your 400 Mhz G3 (if it is a laptop) in 1999 or 2000, the same year or one year after the G4 was introduced and expected it to last five to six years. And now you refuse to buy a G4 laptop one year after the G5 was introduced because you expect it to last less than two years. Interesting logic.

 

[wdlove]

Dangitt you guys.

I was about to buy a new G4 PB after Paris and now you got me thinking about the whole "$3k laptop purchase why not wait until the G5" thing.

Hornswaggle reggle flekkin schmekkin.

If you have a need for a new PowerBook now, then you should go ahead and purchase. These are rumors, there is no way of knowing what will happen at MWSF at this point. At least you will have about six months enjoyment before something else will be available.

 

[idkew]

not to mention- your g4 laptop will not suddenly decrease in speed when a g5 powerbook is released. you will not be missing out on anything. you will actually be gaining over 1ghz in clock speed if you get the g4. that is a huge jump, and i don't see how you wouldn't be happy.

you will never be able to stop progress, so why fight it?

 

[gskiser]

Poor Comparison

You bought your 400 Mhz G3 (if it is a laptop) in 1999 or 2000, the same year or one year after the G4 was introduced and expected it to last five to six years. And now you refuse to buy a G4 laptop one year after the G5 was introduced because you expect it to last less than two years. Interesting logic.

True, and I've learned from my mistake. Don't buy a processor that is on its last leg. Yes, I was able to make it last, though it hasn't been easy. However, you cannot compare 1999 with 2005. The G4 had altivec, was superior to the G3, but it wasn't revolutionarily different. There IS a major difference between the G4 and G5 which is far greater than the difference between the G3 and G4 five years ago. If Apple had announced in 1999 that the G4 was 64-bit and that was the direction they were heading, there would be no way my G3 machine would have lasted 5 years.

I think my logic is pretty sound, at least for my budget and future needs. Like I said though, its up to the individual. If you want a G4 PB, go for it. Its fine for now, but I don't see it being fine in 1 year. I'm just saying that I personally don't plan to get one until Apple proves it has a future. Each individual can decide for themselves though based on their needs.

 

[pigwin32]

The big advantage of the G5 was that it could scale to 2.5 ghz,
whereas the G4 is still at 1.5. But, clock for clock the G4 is more powerful.
A dual core G4, in a Powerbook, is a hell of an improvement, esp. since we probably won't see a G5 in a powerbook till the 65 nm process, and that would still be only a 25% performance improvement.
Vs. a dualcore G4 100% improvement, at least with the stuff I run.

<cough>fsb</cough>
I'm not sure where you get your information re the 65 nm G5, given the issues IBM appear to have had moving to 90 nm I can't see 65 nm being just round the corner. But yes, point taken and for the sake of argument if Apple did choose to use the dual core G4 the accepted PB roadmap goes out the window. Trying to both squeeze a G5 into the PB and improve performance over the dual core G4 PB is likely to be significantly more difficult than trying to get a G5 into the PB right now. Of course none of this addresses the increasingly 64-bitness of the OS, what are the implications for users of G4 machines once Apple starts tweaking for the G5/64 bit?

 

[AidenShaw]

Most apps have no use for 64-bit, and will stay 32-bit

<cough>fsb</cough>Of course none of this addresses the increasingly 64-bitness of the OS, what are the implications for users of G4 machines once Apple starts tweaking for the G5/64 bit?

Apple will have to support 32-bit applications with the 64-bit OS, and will have to support G4 systems with a 32-bit variant (fat images or however) of the OS. It will probably be 5 years before the G4 systems are marginalized.

Most applications have no need whatsoever for 32-bit - these will stay 32-bit so that the same images (with no extra work) will run on all Macs (G3/G4/G5...).

Your 64-bit applications will be the ones that really will run faster if you give each application 6 GiB or 8 GiB or more of RAM. That's not laptop territory for a while....

In other words, a PowerBook G5 with 2 GiB max of RAM wouldn't have any "64-bit advantage". The dual-core G4 would whomp its butt, especially on MP-aware applications.

(If nothing else, there's heat limiting memory size. I have a Dell with 2 GiB of DDR, and the memory compartment is usually the hottest part of the case - those two 1 GiB DDR SO-DIMMs are warm!)
____________________

It would be interesting for Apple to produce a Portable PowerMac G5 - a G5 in a larger laptop (say 1.7" thick, 8-10 lbs) that would have room for the cooling, and even 4 GiB or more of RAM.

Keep producing the thin G4 PowerBooks for people who want small size, lighter weight, and reasonable battery life.

But, give an option for a more powerful portable - even if it's larger and has average to mediocre battery life. Lots of media pros would snap them up - often the other equipment has to be plugged in, so who cares if the Portable PowerMac G5 only gets less than 2 hours or so on the battery?

(Note the big market for the 10 lb. 3 GHz Pentium 4 laptops....)

 

[Mac Dummy]

I think this might be a good stopgap until Apple & IBM work out the cooling issue with a portable G5 chip. One thought I had the other day was...find a way to use liquid cooling like is being used on the dual 2.5 PowerMac, to solve the cooling problem. 😎 Just a thought.

 

[spud]

Aside from the 64bit registers, what is APPLE exactly going to optimize for?
- Apple will improve the GCC 3.5(?) compiler with advanced features, which will benefit both the G4 and G5.
- Apple will move more code to Altivec? G4 and G5
- Apple may write better graphics drivers: Depends upon the graphics chip.

The big advantage of the G5 was that it could scale to 2.5 ghz,
whereas the G4 is still at 1.5. But, clock for clock the G4 is more powerful.
A dual core G4, in a Powerbook, is a hell of an improvement, esp. since we probably won't see a G5 in a powerbook till the 65 nm process, and that would still be only a 25% performance improvement.
Vs. a dualcore G4 100% improvement, at least with the stuff I run.

just like the late model G3s were "good enough", but now you they'll hardly run 10.3 (or in the case of my powerbook, won't run it at all).

if i'm going to invest in a new computer, it will need to be a g5.

 

[MikeBike]

Here's a Guy with a Brain!

In other words, a PowerBook G5 with 2 GiB max of RAM wouldn't have any "64-bit advantage". The dual-core G4 would whomp its butt, especially on MP-aware applications.

It would be interesting for Apple to produce a Portable PowerMac G5 - a G5 in a larger laptop (say 1.7" thick, 8-10 lbs) that would have room for the cooling, and even 4 GiB or more of RAM.

EXACTLY: The G5 scales up to 2.5 ghz.
But, if you're only going to put a 1.8-2.0ghz G5 in a powerbook,
then, a DualCore G4 is By Far, a Better Solution.

What you guys really want is a TRANSPORTABLE G5 IMAC.
Hopefully, the IMac rumors will be true and you guys mindlessly drooling over a G5 can get your wish, at a lower price, then a PowerBook!

Again, as far as I know, there are No G5 specific optimizations:
PowerPC assembler routines coded to use 64bit registers.
Even then, similar routines could be written for the G4.
What I believe Apple is doing is tuning the GCC3.5 compiler to product Much Better code for the G5, but, again, those improvements will also benefit the G4. A Win\Win situation.

 

[MikeBike]

Again, as far as I know, there are No G5 specific optimizations:
PowerPC assembler routines coded to use 64bit registers.
Even then, similar routines could be written for the G4.
What I believe Apple is doing is tuning the GCC3.5 compiler to product Much Better code for the G5, but, again, those improvements will also benefit the G4. A Win\Win situation.

Please correct me if I'm wrong.

 

[MikeBike]

<cough>fsb</cough>
QUOTE]

If you read the article you'll see talk of an on board memory controller.
i.e. better fsb.

 

[Sun Baked]

<cough>fsb</cough>
QUOTE]

If you read the article you'll see talk of an on board memory controller.
i.e. better fsb.

If it had an AGP port Apple probably could ditch the Northbridge and use a RIO-to-HT bridge chip and use a KeyLargo2 for I/O simplifying the chipset.

 

[AidenShaw]

Again, as far as I know, there are No G5 specific optimizations:
PowerPC assembler routines coded to use 64bit registers.
Even then, similar routines could be written for the G4.
What I believe Apple is doing is tuning the GCC3.5 compiler to product Much Better code for the G5, but, again, those improvements will also benefit the G4. A Win\Win situation.

The mainline compilers of interest (gcc, vc++, xl*) already support 64-bit integers as a native language datatype. When the compiler output is for a 32-bit machine, 64-bit operations generate a short series of 32-bit integer ops. When set for 64-bit architectures, a single 64-bit operation is generated.

Unless your algorithm depends heavily on extended precision integers (some crypto codes come to mind), the 32-bit code is usually good enough. Since the only visible arithmetic difference between the 74xx and the 970 is 64-bit integers, few applications will see much difference., (The 750/74xx already have 64-bit floating point and 128-bit VMX instructions.)

As far as the compiler tuning goes, the engineers look for the optimizations that help the 970 without hurting the 74xx (too much). So, as long as you stay away from 970-only stuff (64-bit) or 74xx-only stuff (little-endian compatibility) you can often come up with code that runs good (or better) on both - even though it might not be the best possible choice for either chipl

 

[beatle888]

You bought your 400 Mhz G3 (if it is a laptop) in 1999 or 2000, the same year or one year after the G4 was introduced and expected it to last five to six years. And now you refuse to buy a G4 laptop one year after the G5 was introduced because you expect it to last less than two years. Interesting logic.

True, and I've learned from my mistake. Don't buy a processor that is on its last leg.

i was going to respond to that post for you but you beat me to it.
by the way i absolutely agree with you. i wont buy a g4 pb. my laptop will do just fine until that beauty hits the shelves. it will be way different than comparing a g3 to a g4. FSB throughput is an important factor, not to mention the amount of ram it will most likely hold.

 

[pigwin32]

If you read the article you'll see talk of an on board memory controller.
i.e. better fsb.

I read the article, your comparison was G4 to G5, not dual core G4.

The big advantage of the G5 was that it could scale to 2.5 ghz

The big advantage of the G5 is that it doesn't have a 167MHz FSB.

Given Motorola's previous inability to deliver chips to spec why would Apple consider using the e600? Apart from memory addressing is there really no other advantage to using the G5, i.e. Apple's tuning of apps for the G5 doesn't give any greater performance than for the G4? A dual-core G4 with onboard memory controller is a tantalising prospect for PB addicts. The big question is where would a dual-core G4 PB fit into the product tree, is it part of the main trunk or a dodgy branch that will be pruned when the PB is G5-ready?

 

[AidenShaw]

Intel will have dual-core chips in its entire line of processors by the end of 2005

The big question is where would a dual-core G4 PB fit into the product tree, is it part of the main trunk or a dodgy branch that will be pruned when the PB is G5-ready?

Main trunk, by necessity.

http://www.eweek.com/article2/0,1759,1587826,00.asp

Intel officials have been saying that a dual-core Itanium chip, code-named Montecito, will be released in mid-2005, followed by dual-core processors for desktops and mobile computers. The company has yet to release the code names for those chips.

Now Intel will have dual-core chips in its entire line of processors by the end of 2005....

The dual-core PB G4 will be pruned by the dual-core PB PPC9xx, competing against Intel laptops with dual-core hyper-threaded 64-bit "Pentium 4" and dual-core "Centrino" chips.

 

[Mord]

just like the late model G3s were "good enough", but now you they'll hardly run 10.3 (or in the case of my powerbook, won't run it at all).

if i'm going to invest in a new computer, it will need to be a g5.

look into xpostfacto it will let you run 10.3 on any g3 mac and i mean any, even upgraded early pci macs

 

[themacman]

How bout this Idea

Dual Core G5 Dual 3.0 Ghz Powermac They could call it.

So fast and so quite it will pounce on anything, and tiger will come standard on it

 

[Surreal]

apple could just add a line of nbnotebooks.

i mean, it would be a boon to their sales if they lowered the normal g4 prices...had a g5 and dual g4 line

the lowly single g4 (very fine...) would be the LOWEST...at 1.5...1.33...not low at all

as it is..an ibook is not a logical option for me...i make music ..use reason..logic..ableton live...and i make 3d animations using maxon cinema 4d (and hopefully later on maya). all this is only relevant because i need memory.

soldering the memory on makes sense in terms of crippling a line.

but since everything up the 12 pb is crippled...my lowest choice is a 15. at 2000.

now, i am a student, and if i am going to buy 2000 worth of computer, i am already out of "viable." and i plan on buying it...but if i do, i want it to have the MOST potential to last, simply because i WILL NOT be making a purchase of another computer within 4-6 years.

unless they prove to people that a dual core g4 is NOT a stop gap measure... sales wwould be bad. i would not buy it...not because i didnt think it was fast, not because i wouldnt think it could last 2-3 years, but because i KNOW it wouldnt last AS long as a g5.