PowerBook To Get Dual-Core G4?

[DakotaGuy]

too bad they don't use ibm's 750 chip anymore.

Why? The 7447A is a MUCH better processor. I have had G3 based machines and they were fine, but in NO WAY do they compare to the newest G4's. Everyone complains about the 167Mhz FSB speed...um the 750 ran at 100Mhz. It did not have Alti-Vec which makes a huge difference especially on the newest applications and OS from Apple. Last, it never came close to clocking as high as the G4 has gone, 1.5Ghz. Look at benchmarks anywhere (ie. iBook G3 vs. iBook G4) and you will see the current G4's kill the old IBM G3's. The G3 was a good processor, but WAS is the key word.

 

[neutrino23]

There may be more than just heat issues affecting the G5. The report from IBM engineers posted around a month or so ago referred to wide variations in power demand per chip. It may be that they can make a few good ones that could go in a PB but they can't make them consistently. There may be other issues related to PB use that haven't been made public. It could be that there is not just one killer problem but that overall the G5 puts too much demand on the battery and the heat dissipation limits.

The G4 has been designed for low power use from the start. A dual core CPU seems great for a laptop. You get bursts of computing power when you need it and when idling you just shut down the second core to conserve battery power.

According to the Bare Feats report cited earlier a dual processor 1.42MHz G4 did not trail too far behind a 2.0GHz dual g5. A dual core G4 running at 1.6MHz or better would be a wicked fast laptop.

 

[TomSmithMacEd]

Really all of you G5 powerbook! now people are built into thinking the g5 is amazing. I do think a dual core g4 would be quite faster than a single g5.

 

[afields]

I think apple could easily call this a g5 by name. G5 mobile, or "mini" or something. This wouldn't just be a minor improvement to the g4, they mine as well call it a g5.

 

[titaniumducky]

In short, processors typically have one core for each processor. The core is in charge of doing the computing. This core makes up the processor. Having dual core processors is like having two processors in one-except that they are within a processor, so they can communicate faster and usually use the same resources. So, on the clock, they are faster, but in real life it all depends on what you are doing. High end apps like FCP and Motion will run faster, but most games will not just because the processor is dual core. However, adding faster RAM, a better video card, etc. will speed up your games. If I'm wrong about anything, please correct me. This is just from my own knowledge.
–Chase

So Intel's "HT-technology" (HyperThreading) is just their name for dual-core P4s, right?

 

[Maxx Power]

that's the most number of buzzwords i've ever seen in a post with coherent sentences. as for the ppc 74xx having a weaker fpu/mhz vs. the p4, i would say no. perhaps you mean athlon or pentium m. practically any modern processor today has a better fpu/mhz vs. the p4.

Thanks for commenting on my logical reasoning unit and sentence creation shareware. I read that the fpu on the G4 is per unit weaker than that of the P4 from www.arstechnica.com the link is:
http://arstechnica.com/cpu/01q4/p4andg4e2/p4andg4e2-3.html

and i quote "The P4's FPU hardware is a little beefier than that of the G4e, but a little less beefy than that of the PIII or the Athlon. The P4's designers slimmed down the FPU and weakened it a bit, banking on increasing clock rates and the widespread replacement of x87 code for SSE2-optimized code to take up the resulting performance slack. A closer look at the P4's FPU will show just how it has been weakened."

 

[Analog Kid]

If this chip happens I don't see any need for a G5PB until you can reasonably put more than 4GB of RAM into a Powerbook...

 

[eric_n_dfw]

So Intel's "HT-technology" (HyperThreading) is just their name for dual-core P4s, right?

No, I believe that HT is having a single core with the ability to do some instructions in parallel. Dual core chips would actually have the equivilent of two full CPU's on one chip. (very over simplified there, but I think that's the difference)

An Ars' link on HT: http://arstechnica.com/paedia/h/hyperthreading/hyperthreading-1.html

 

[CmdrLaForge]

I think that it would be really great to get powerbooks with dual core G4s. I would buy one in a second. Of course the fsb is important, but the article says it has a fast one.

At 2GHz I would definitly prefer a dual core G4 over a single core G5.

 

[Matrix9180]

How much you wanna bet those were for XBox game development for the XBox 2 that will have G5's?

I seem to recall MS saying the XBox2 SDK was going to be for OS X... something about a custom NT kernel it would run on

 

[eric_n_dfw]

I seem to recall MS saying the XBox2 SDK was going to be for OS X... something about a custom NT kernel it would run on

What does the NT kernel have to do with OS X?

AFAIK, MS has shipped the beta XBox2 dev stations as G5's with that custom NT kernel on them. I'm sure OS X is nowhere to be found on those machines. (In fact, I wouldn't be surprised if they put big XBox logo stickers covering the Apple ones.)

 

[jettredmont]

No, I believe that HT is having a single core with the ability to do some instructions in parallel. Dual core chips would actually have the equivilent of two full CPU's on one chip. (very over simplified there, but I think that's the difference)

An Ars' link on HT: http://arstechnica.com/paedia/h/hyperthreading/hyperthreading-1.html

Hyperthreading (aka SMT to the less eXtreme Intel folks out there) does not do anything in parallel, except in some very special cases (which aren't generally described as "in parallel" but which are close enough).

HT allows two thread's instructions to exist in the CPU's instruction flow at the same time. They will not get done in parallel, but one will get done then the next. Where this is a benifit is that a side effect of classical threading design on CPUs is that going from one thread's instruction to another thread's instruction causes latency because they can't both be in the flow at the same time.

Since a modern OS will be constantly switching the CPU from one thread to another, hyperthreading allows for a significant performance boost, and it allows one thread to be using the FPU of the chip while another is using the Integer units (kinda sorta in parallel, but from the CPU perspective these are serial tasks).

 

[jettredmont]

If its being announced in October we could definetly be seeing this in early Q1 2005.

I don't think we'll be seeing it before the New Year though.

Remember Dual Cores are only good for single applications if they are written with multiple processors in mind (30% increase). They are also good for running two CPU intensive applications at the same time. However if they aren't clocked significantly higher (I'm presuming they are going to be released in 1.33Ghz to 1.8Ghz models) you won't see a massive speed up in individual applications.

That FSB hike will be great for media encoding applications though 🙂

Given that you will always have quite a few processes running in OS X, you will always get a performance boost from dual CPUs or dual cores. At the minimum, all OS X overhead (window management, disk management, UI control and responsiveness, etc) can be shuttled off to one of the two CPUs while your main app has the other CPU all to itself.

One can, in fact, argue that the UI responsiveness bought by having the "other" CU doing that is worth the price of admission all on its own!

On the other hand, you may be surprised at how many of your apps are actually multi-threaded in non-trivial ways. It's not just FCP and ProTools!

 

[shawnce]

So Intel's "HT-technology" (HyperThreading) is just their name for dual-core P4s, right?

No. HyperThreading is Intel's name for making a single core CPU appear to be a dual core CPU from the POV of an HT aware OS.

Basically P4 (and G5) have the ability to have hundreds of instructions in flight on the CPU core and it can be difficult for a single code stream (aka thread) to fully leverage this ability because of data dependencies between instructions in the stream and/or the instruction make up of the stream. So HyperThreading (known generally as SMT or simultaneous multithreading) allows instructions from multiple threads (currently 2 in the case of HT IIRC) to be intermixed on the CPU core with the goal being to better utilize the CPU execution units/pipelines.

This is done by the CPU and OS cooperatively maintaining state/context per thread for the instructions in flight.

IBM has PPC970 based cores with SMT capabilities in the works as well as dual core CPUs (rumored to be a better implementation of what Intel has). Note you can use SMT per core on multi-core CPUs to better leverage the capabilities of the individual cores (dual core CPU appearing to be a quad core CPU, etc.)

For more detail... Arstechnica has an article on SMT / HT written a little while back.

 

[mac512pbg4]

rdf naming allowed

just call the dual core g4 a "g6 mobile", a "leap-frog" improvement

fits easily into SJ's reality distortion field

 

[Sun Baked]

]IBM has PPC970 based cores with SMT capabilities in the works[/B] as well as dual core CPUs (rumored to be a better implementation of what Intel has). Note you can use SMT per core on multi-core CPUs to better leverage the capabilities of the individual cores (dual core CPU appearing to be a quad core CPU, etc.)

For more detail... Arstechnica has an article on SMT / HT written a little while back.

The new CPUs (the GR-UL) with SMT are NOT based on the PPC970.

The PPC970 (GP-UL) is a Power4-based chip, the next major chip -- the Power5-UL/GR-UL -- with SMT is a Power5-based CPU.

 

[gekko513]

I do agree to your conclusion, though, a minor note, 2xDDR400 makes 800 MHz, and since it is double data rate, that makes 1600 MHz.

Nah ... DDR400 runs at 200MHz and the double data rate gives it the name DDR400.

The first DDR ram was DDR266 which was double data rate PC133 RAM.

 

[shawnce]

(kinda sorta in parallel, but from the CPU perspective these are serial tasks).

Actually from the POV of the CPU execution units they are done in parallel (all modern CPU can execute instructions out of order, etc.). They are only serial from the POV of a given execution stream. In the case of SMT capable CPUs multiple stream can have multiple instructions in flight on the same core, executing in parallel relative to each other but still serially relative to the execution stream (thread) they came from.

Of course a given execution unit can only be executing a single instruction for a given stage in the units pipeline (the P4 / G5 have several stages in their various execution units), so at that level of granularity you can say things are done serially.

 

[Mord]

there will be dualcore-dualboards manufactured by gigadesign and others for the mdd/quicksilver too? (2x 2core 1,6ghz g4s?)

i'd like one of those in my cube muhahahaha (i'd bet money that if it is possible powerlogix would do it first but half of them wont work right.

 

[Frobozz]

MOSR

MacOSRumors.com has already reported on this stuff. No new news here. But it's nice to see more than one publication say it. My only gripe is the terrible track record of the Register. Then again, MOSR isn't particuarly accurate, either.... 🙂

I am willing to bet that there is a huge hunk of truth to this rumor either way. I don't see any problem with a dual core G4 in a powerbook as opposed to a G5. We all know the 970mp will probably be the next G5 chip, which means it won't be ready for portables soon. The dispersing of heat concentration will help the 970mp, and the FreeScale chip, though.

 

[Ibjr]

their CPU's as of late look backwards in the extreme compared to IBM, AMD, and Intel's Pentium M. (Lets be honest everything looks sweet compared to the P4.)

Drop the buzz words. The G4 beats the P3 at the same clock speed. The P3 is the PentiumM.

Advice: Instead of buzzwords lets stick to the facts.

 

[Frobozz]

I think that it would be really great to get powerbooks with dual core G4s. I would buy one in a second. Of course the fsb is important, but the article says it has a fast one.

At 2GHz I would definitly prefer a dual core G4 over a single core G5.

I totally agree. More speed is always a good thing, and aside from the marketing spin I think a dual core g4 would be better than a single core, single g5 in the current crop. Also, it helps Apple diversify their supply chain and use those G5 chips for desktops with different power and heat dissipation limits.

 

[gekko513]

Drop the buzz words. The G4 beats the P3 at the same clock speed. The P3 is the PentiumM.

Advice: Instead of buzzwords lets stick to the facts.

The Pentium M has more in common with the P3 than the P4, but it more than just a P3.

And the current G4 has more pipeline stages than the G4 had in the days of the P3, so we don't really know if the G4 still beats the P3 at the same clock speed.

 

[Ibjr]

The Pentium M has more in common with the P3 than the P4, but it more than just a P3.

And the current G4 has more pipeline stages than the G4 had in the days of the P3, so we don't really know if the G4 still beats the P3 at the same clock speed.

Given that the P3 beat the P4 at similar clockspeeds for most uses, i thought the Pentium M's changes were purely for power to insure stablity at the higher clock speeds.

The addition to stages doesn't really matter because Intel would have to do similar tasks.

 

[JoePike]

Now we're talkin' turkey. G5 or no G5, the day they put a 2+ ghz anything in a powerbook is the day I abandon my beloved and trusty prismo.

-Joe