IBM PowerPC 970 vs Intel Pentium 4

bluecell · Dec 18, 2002

I was intrigued by this latest article on Geek.com, which talks about the performance of IBM's 970 processor vs Intel's P4. It basically says what everyone pretty much already knew. The 970, when it's released (and regardless of Apple's adoption), will be behind in comparison to the P4 at its highest clock speed.

When released, the 970's highest clock speed will be 1.8GHz. The Intel P4 is already at 3GHz and that number is likely climb before the 970's debut. However, it should be recognized that the 970 at 1.8GHz delivers 937 SPECint2000 Base and 1051 SPECfp2000 Base. By comparison, the P4 at 1.8GHz only delivers 596 SPECint2000 Base and 618 SPECfp2000 Base. The P4 at 3GHz does manage to beat the fastest 970 with 1085 SPECint2000 Base and 1092 SPECfp2000 Base.

It's interesting, I've been reading more and more about IBM's future PPC plans and I'm very surprised. They say that they hope to make the 970 reach 6GHz within the next couple of years. Hopefully, IBM will actually make the effort and Apple's PPC dilemma will be a thing of the past.

So, if Apple does decide to adopt the 970, how fast do you think it will be by the end of 2004?

Mr. Anderson · Dec 18, 2002

Well, it is a better processor than the G4 - and if the performance is lacking a bit, well, they'll most likely continue the dual processor options. I'd love to see a quad too.

Who knows how fast it will be by the end of 2004. If they manage to get an extra GHz by then I'd be surprised. 6 GHz is a long way, but is it feasible to think that they can implement large speed upgades?

I personally can't wait to own one.

D

oldMac · Dec 18, 2002

Chip Complexity

Well, one of the theories of RISC used to be that you could go faster (MHz) by keeping the complexity of the chip down. Seems that hasn't happened in reality.

In fact, our chips keep getting more complex *and* faster. The truth is that there isn't that much of a difference between Intel's "CISC" designs and PPC's "RISC" these days except that Intel has another stage or two in the process. So, complexity doesn't seem to be slowing Intel down.

Some have speculated that it's actually the complexity of the Altivec unit that prevents Motorola's G4s from running faster. IBM was not a big proponent of the vector unit when it was originally introduced into the PPC plans. Maybe IBM knew what they were talking about and Altivec is preventing the G4 from reaching its true speed potential?

Assuming that it's Altivec that has tied the hands of the G4, perhaps IBM knows how to make a better (more scalable) vector unit?

We can only hope.

Falleron · Dec 18, 2002

This is not good. The fact that a P4 3.06Ghz chip will outperform the IBM chip is worrying. I know the IBM chip is better than the G4, but still, this is a big problem. We will probably have to compare a 1.8Ghz 970 to a 3.6Ghz P4/5! Its bad (actually, there are things that the SPEC dont/cant measure which makes this not such a bad chip). However, still behind the equivalent Intel chip.

edesignuk · Dec 18, 2002

Indeed, this is worrying for the future of the PPC and in turn Apple. Let's hope they can get those 970 clock speeds up ASAP and Apple can start kickin' some wintel a$$ once more.

Mr. Anderson · Dec 18, 2002

I noticed at the message board on that site that someone brought up the good point that the 970 is a 64 bit chip - so actual milage might vary. He also mentioned that its all speculation at this point.

I'm not worried, a dual 1.8 GHz 970 will be blindingly fast compaired to the current machines - get a new faster bus and memory and we'll be fine. If they make it to the 6GHz mark, well, all the better.

D

Chad4Mac · Dec 18, 2002

What about the chip's efficiency.

When I heard about the Megahertz Myth, the most valuable attribute I learned about was the way the chip processed information. I want to say, how smoothly and quickly the chip brought info through the pipeline, and out the other side.

Would it be safe to say that this 970 chip is more efficient, that is, over the P4? Could this be another selling point for Apple? A 64-bit over 32-bit type thing...

On the other side, could this (the fact that there will be a difference in gigahertz between the 970 and the P4) hurt Apple? Sounds like the G4 and P4 Megahertz Myth type thing all over again? No?

And, just a point. The people that will truley utilize these next genreation chips (you people on these forms) are the people that will benefit from these chips. You guys doing video and photo rendering on a daily basis, and those "render farms."

IN your case, I would like to look at power consumption between these two chips. Which one would be more efficient in energy conservation? ****, it will save you money in the long run.

"...go with Apple's new 64-bit technology and save five hundred dollars every year in energy savings."

Chad4Mac

Funkatation · Dec 18, 2002

Hmm. A single P4 3Ghz+ may be faster than a SINGLE 1.8GHZ 970. But I would guess that apple would continue the all dual processor strategy, so there will be 2 babies in this thing. That should more than make up for it I think.

ExoticFish · Dec 18, 2002

worrying about performance?

if this new 970 is really that much faster that what we've got now, even if it doesn't quite reach the P4's speed when released, that would still be absolutely amazing speed for OS X. I'll be the first to say that my 1GHz TiBook isn't as fast as my friends 1.7 GHz Dell laptop... but Apple has proven that speed doesn't make a good machine. but my point is that right now we have a HUGE performance gap between Mac's and PC's so this chip sounds like it'll close that gap by a very significant number. I'm all for it and ready with open arms!

MacBandit · Dec 18, 2002

Re: worrying about performance?

Originally posted by ExoticFish
if this new 970 is really that much faster that what we've got now, even if it doesn't quite reach the P4's speed when released, that would still be absolutely amazing speed for OS X. I'll be the first to say that my 1GHz TiBook isn't as fast as my friends 1.7 GHz Dell laptop... but Apple has proven that speed doesn't make a good machine. but my point is that right now we have a HUGE performance gap between Mac's and PC's so this chip sounds like it'll close that gap by a very significant number. I'm all for it and ready with open arms!

I would venture to guess that your friends laptop doesn't have the battery life of your Dell. If this is true which one is faster when the Dell is dead and you Ti is still ticking along?

Catfish_Man · Dec 18, 2002

True...

Originally posted by Falleron
This is not good. The fact that a P4 3.06Ghz chip will outperform the IBM chip is worrying. I know the IBM chip is better than the G4, but still, this is a big problem. We will probably have to compare a 1.8Ghz 970 to a 3.6Ghz P4/5! Its bad (actually, there are things that the SPEC dont/cant measure which makes this not such a bad chip). However, still behind the equivalent Intel chip.

...SPEC does NOT test vector performance, and is heavily influenced by compilers. I would guess that the 970 will be miles ahead of the P4 in vector and double precision integer performance, and a little bit slower on single precision scalar code. Here's my guess at the stats:

Code:

                     PPC970               Prescott
frequency           1800                     3600
IPC                  3-4                       ~2
cache               512k L2                1MB L2
power                40W                     80+W
die size             ~120                     ?
bus                  800MHz                800MHz
SPECFP guess         1051                  ~1350
SPECINT guess         937                   ~1350
max ram               2^64              2^32 (4GB)
process              .13u SOI               .09u

When the 970 switches to .09u manufacturing is when it really gets fun. Then it'll be able to compete against the P4/5 on an equal footing (they'll have room to do things like add a second core, or a bunch of cache, or an on chip memory controller. Plus, the clock frequency will go up).

ExoticFish · Dec 18, 2002

wait... you are talking about HIS Dell right?

but yes I agree, that's what I was trying to say anyway! My TiBook technically is slower, but that doesn't mean that my TiBook doesn't kick the crap out of his Dell in every way, it does! He constantly has that thing plugged in! The Dell is HUGE and the fan is so much louder! He tries to multitask using Winblows and I just laugh in his face and show him how many apps I always have open and running and how great it still runs. The only comment he ever makes is "hey you wanna play some UT2K3? oh wait, you're using a gay Mac" Just wait till the port is released, then he'll have nothing to say as I'm fraggin his ass to death!!!

bluecell · Dec 18, 2002

Hmmm... Looking at the information that IBM has given thus far, the 970 at a little over 2GHz will most likely outperform the 3GHz P4. The 'MHz myth' is actually a myth to some extent, it's when the P4's clock speed is more than twice as fast as a G4. Just the fact that at 1.8GHz, the 970 totally outperforms the 1.8GHz P4. It even outperforms the P4 at 2GHz in a big way (640 SPECint2000 Base/704 SPECfp2000 Base).

People should know by now that there's more to a processor than just the clock speed, although it helps. This article from Ars Technica compares the G4 to the P4 in great detail. Basically, it says that the G4 is more efficient and other things besides clock speeds do have to be taken into account. Take the pipeline, for example. The article's diagram (pictured below) shows how efficient the G4's pipeline really is compared to the P4. What Intel lacks in efficiency, it makes up for in speed. If the G4 was set at higher rates, it wouldn't be as much of an issue. Another thing is that making everything dual processor isn't the answer.

Top: Intel P4/Bottom: Motorola G4

bluecell · Dec 18, 2002

Re: True...

Code:
Originally posted by Catfish_Man
...SPEC does NOT test vector performance, and is heavily influenced by compilers. I would guess that the 970 will be miles ahead of the P4 in vector and double precision integer performance, and a little bit slower on single precision scalar code. Here's my guess at the stats:

Code:

PPC970 Prescott frequency 1800 3600 IPC 3-4 ~2 cache 512k L2 1MB L2 power 40W 80+W die size ~120 ? bus 800MHz 800MHz SPECFP guess 1051 ~1350 SPECINT guess 937 ~1350 max ram 2^64 2^32 (4GB) process .13u SOI .09u

When the 970 switches to .09u manufacturing is when it really gets fun. Then it'll be able to compete against the P4/5 on an equal footing (they'll have room to do things like add a second core, or a bunch of cache, or an on chip memory controller. Plus, the clock frequency will go up).

I agree, but isn't the 970's bus at 900MHz?

pgwalsh · Dec 18, 2002

Chard4Mac
IN your case, I would like to look at power consumption between these two chips. Which one would be more efficient in energy conservation?

I read in the paper the most home computers only increase your electric bill between $3 and $6 dollars a month, if left on all the time.

Originally posted by bluecell
Another thing is that making everything dual processor isn't the answer.

Cool animated gif showing the pipeline.

Two processors helps with FCP, AP, and Cubase. These apps can take advantage of dual processors and they all tax processors big time. When you start loading effects and transitions you start seeing the processors bog down.

bluecell · Dec 18, 2002

Originally posted by pgwalsh
I read in the paper the most home computers only increase your electric bill between $3 and $6 dollars a month, if left on all the time.

Cool animated gif showing the pipeline.

Two processors helps with FCP, AP, and Cubase. These apps can take advantage of dual processors and they all tax processors big time. When you start loading effects and transitions you start seeing the processors bog down.

Yes, dual processors help apps like FCP, Logic, Maya, etc, but making everything dual processor to compensate for lack of speed is probably something Apple isn't too happy about. I've seen some pretty upsetting benchmarks that show that even the top end dual processor PowerMacs fall behind in performance compared to a single top end P4 or Xeon machine.

Flickta · Dec 18, 2002

Well, the only thing that confuses me is that the upcoming 970 (top of the line!) will be as fast as a P4 available now! And Intel won't just sit and wait.

And imagine if PC's become dual processor? Seems like Apple's big day is far away... too far away.

Mr. Anderson · Dec 18, 2002

Originally posted by Flickta

And imagine if PC's become dual processor? Seems like Apple's big day is far away... too far away.

There already are some - but it requires that the software can recognize the multiple processors.

D

bluecell · Dec 18, 2002

Originally posted by Flickta
Well, the only thing that confuses me is that the upcoming 970 (top of the line!) will be as fast as a P4 available now! And Intel won't just sit and wait.

And imagine if PC's become dual processor? Seems like Apple's big day is far away... too far away.

Actually, the 970 at its top end of 1.8GHz isn't as fast as the P4's current top end at 3GHz. It is pretty close. At 3GHz, the 970 would destroy all of the P4s and Xeons out there today. If IBM stays on target, they could very well help Apple come back to the front. From reading everything IBM has put out there, the chip will get smaller (.09u) and they expect to reach 6GHz within the next couple of years. Hopefully, they'll keep their promise and maintain development of PowerPC (which includes more processor options) and not cause a problem for Apple like Motorola did.

pgwalsh · Dec 18, 2002

Originally posted by bluecell
Yes, dual processors help apps like FCP, Logic, Maya, etc, but making everything dual processor to compensate for lack of speed is probably something Apple isn't too happy about. I've seen some pretty upsetting benchmarks that show that even the top end dual processor PowerMacs fall behind in performance compared to a single top end P4 or Xeon machine.

I agree that that making everthing dual just to compensate for lack of speed isn't a good thing. I was coming from the point of view as when they do match the speed, I hope they don't drop the second processor.

I don't know anyone that has a computer with the 3 Ghz Intel inside. Most PC users I know are happy with what they have. I think the speed appeals more to gamers and hard core professionals. Most of my friends don't play games and aren't hardcore professionals. Many people I know in Silicon Valley don't want to spend the money right now on beefy machines. The majority of the working professionals want laptops. They all have Apple envy.

Flickta · Dec 18, 2002

Originally posted by dukestreet

There already are some - but it requires that the software can recognize the multiple processors.

D

Right you are! And some software, especially games, do not recognize them. And Games are one of the major impluses for lots of people to buy a certain comp.

Flickta · Dec 18, 2002

Originally posted by bluecell
Actually, the 970 at its top end of 1.8GHz isn't as fast as the P4's current top end at 3GHz. It is pretty close. At 3GHz, the 970 would destroy all of the P4s and Xeons out there today. If IBM stays on target, they could very well help Apple come back to the front. From reading everything IBM has put out there, the chip will get smaller (.09u) and they expect to reach 6GHz within the next couple of years. Hopefully, they'll keep their promise and maintain development of PowerPC (which includes more processor options) and not cause a problem for Apple like Motorola did.

I've been hearing it for several years. "this processor is slow, yes, but if they can manage it to become faster, it will just crash any competitors!" - Again, the competitors won't wait.

bluecell · Dec 18, 2002

Originally posted by Flickta
I've been hearing it for several years. "this processor is slow, yes, but if they can manage it to become faster, it will just crash any competitors!" - Again, the competitors won't wait.

I totally agree.

frogstomp · Dec 18, 2002

Step up to the Plate Apple!

I love my Apple, but I have lost all dellusions that it compares to mainstream systems. Yes the OS is superior, but we need speed to compete.

If IBM can not keep up even with its new 970, it is time for Apple to make some serious choices.

I have no qualms with any chip makers, if they are stable and fast let's use them.

jayscheuerle · Dec 18, 2002

You Technolusters!!

For 95% of the Mac users out there, performance worries disappeared a couple hundered mHz ago. A 1.8 gHz PPC 970 won't be discernably different from an 800 mHz G4 (or even G3 for that matter). Most of the apps that people run are relatively simple and the biggest change may be that the Aqua overhead is no longer noticed, unless Apple makes it even more burdensom to force us into hardware upgrades (does that sound jaded?

)

That said, I'd be happy to not have the burden of defending an obviously slower processor to people who think that's the most important attribute in a computing experience.

Let's at least get Apple back in the race!

IBM PowerPC 970 vs Intel Pentium 4

macrumors member

Moderator emeritus

macrumors 6502a

macrumors 68000

Moderator emeritus

Moderator emeritus

macrumors 6502

macrumors regular

macrumors 6502a

macrumors 604

macrumors 68030

macrumors 6502a

macrumors member

macrumors member

macrumors 68000

macrumors member

macrumors 6502

Moderator emeritus

macrumors member

macrumors 68000

macrumors 6502

macrumors 6502

macrumors member

macrumors newbie

macrumors 68020

Our Staff