How much better are Apple/IBM PowerPCs compared to x86 Athlon64/Pentium4 ??

ShePearl

ShePearl

Limp Gawd
Joined
Apr 16, 2004
Messages
417
I'd like to know how x86 based PCs, (Intel Pentium4, and AMD Athlon64) stack up against Apple/IBM PowerPC CPUs ??

I mean, Microsoft chose to use IBM's PowerPC variant CPU for XBOX2, and even IBM is supplying CPUs for Nintendo Revolution, and Sony PlayStation3.

There must be a good reasons why these companies chose not to use x86 based Intel Pentium4 or AMD Athlon64 processors. Especially Microsoft (XBOX) who knows extremely well about x86 based processors. Their first (current) XBOX uses Intel's Pentium3 733MHz CPU...

They must be better performer.. but how much better are Apple/IBM's PowerPC CPUs compared to PCs (Pentium4, Athlon64) ?
 
From what I have read they are not better and in fact the x86 intel/amd cpus are superior. They are even faster nowdays in graphics and video rendering... I think it all comes down to profit... the ibm/motorolla chips are being sold much cheaper to them.
 
Shane said:
From what I have read they are not better and in fact the x86 intel/amd cpus are superior. They are even faster nowdays in graphics and video rendering... I think it all comes down to profit... the ibm/motorolla chips are being sold much cheaper to them.
Click to expand...
Hmm... I think you're mis-informed (no disrespect). I'm quite sure Apple/IBM's PowerPC CPU is the performance king when it comes to "desktop" CPUs. Just not sure how much better they are compared to aged x86 based CPUs such as Athlon64, and Pentium4.
Also, bear in mind that I'm comparing CPU performance aspect only. :)
 
I want some of what ShePearl is smoking. It is amazing how long Apple has been able to get people to belive such nonsense.
 
Watch this turn into another PC vs. Mac thread...

Google around a little bit, this isn't going to be answered by a bunch of users giving opinions. There are many resources online that will show performance comparisons. All you'll get in these posts are things fighting back and forth...
 
ShePearl said:
Hmm... I think you're mis-informed (no disrespect). I'm quite sure Apple/IBM's PowerPC CPU is the performance king when it comes to "desktop" CPUs. Just not sure how much better they are compared to aged x86 based CPUs such as Athlon64, and Pentium4.
Also, bear in mind that I'm comparing CPU performance aspect only. :)
Click to expand...

no.... p4's and a64's are faster. I can dig up some benchmarks if you want
 
It dedicated applications like consoles, routers, and DSP the risc architecture used by PowerPC, mips and others tends to be more efficent. Now that clockspeed has risen quite a bit the CISC architecture begins to shine in dealing with huge varities of task seen by modern cpus.

ars technica has a some good articles on processor architecture including a comparsion of powerPC with a P4 below

http://arstechnica.com/articles/paedia/cpu/p4andg4e.ars
 
draksia, having been around as long as you have, you should know that x86 != CISC.

no current desktop x86 chip is a true CISC design. in fact, many have taken on pieces of RISC architecture.
 
sabrewolf732 said:
http://www.pcworld.com/news/article/0,aid,112749,pg,8,00.asp

there, the athlon 64 pwns the hell out of the G5
Click to expand...

I've read quite a few articles on the subject and from all of them I've concluded that the x86 CPUs perform better BUT this article is not a good example of it at all. A big problem in comparing the two is that the software is totally different on the two platforms. Different optimizations work for different CPUs and it's not really fair to use an app like Microsoft Word on both of them because the application is compiled differently on either side. The same goes for Quake, it's been optimized for Intel and AMD's stuff.
 
nylint said:
draksia, having been around as long as you have, you should know that x86 != CISC.

no current desktop x86 chip is a true CISC design. in fact, many have taken on pieces of RISC architecture.
Click to expand...

I fail to see how long I have been has anything to do wether I know what x86 is.

You are partically correct though. x86 itself is inf act a CISC. The true defenition for a RISC that one instruction does only one operation like add, shift, store or load. Support for things like add from memory are infact two operations one add and one load.

Current processors especially the P4 will actually take a complex instruction and translate it into several smaller micro ops. This is what the trace cache on the P4 actually stores. These micro ops are very similar to traditional RISC instructions. In reality current processors are still consider CISC's because of the ability to handle multiple operation instructions.

Most modern cpus are in fact a some combination of the RISC and CISC styles and will continue to evolve into systems that can handle large instruction and simple instruction equally well.
 
Deadlierchair said:
I've read quite a few articles on the subject and from all of them I've concluded that the x86 CPUs perform better BUT this article is not a good example of it at all. A big problem in comparing the two is that the software is totally different on the two platforms. Different optimizations work for different CPUs and it's not really fair to use an app like Microsoft Word on both of them because the application is compiled differently on either side. The same goes for Quake, it's been optimized for Intel and AMD's stuff.
Click to expand...

I dont see how one can get results with software that is exactly the same on both platforms
:confused: They have to be coded differantly to be able to run on the differant platforms. Maybe I didnt fully get what you meant?? elaborate? :confused:
 
ShePearl said:
I'd like to know how x86 based PCs, (Intel Pentium4, and AMD Athlon64) stack up against Apple/IBM PowerPC CPUs ??

I mean, Microsoft chose to use IBM's PowerPC variant CPU for XBOX2, and even IBM is supplying CPUs for Nintendo Revolution, and Sony PlayStation3.

There must be a good reasons why these companies chose not to use x86 based Intel Pentium4 or AMD Athlon64 processors. Especially Microsoft (XBOX) who knows extremely well about x86 based processors. Their first (current) XBOX uses Intel's Pentium3 733MHz CPU...

They must be better performer.. but how much better are Apple/IBM's PowerPC CPUs compared to PCs (Pentium4, Athlon64) ?
Click to expand...

As you can see the Apple's are slower by alot and get their ass handed to them!
http://www.digitalvideoediting.com/2002/07_jul/features/cw_macvspc2.htm
http://www.robgalbraith.com/bins/multi_page.asp?cid=7-6451-6423
 
nylint said:
draksia, having been around as long as you have, you should know that x86 != CISC.

no current desktop x86 chip is a true CISC design. in fact, many have taken on pieces of RISC architecture.
Click to expand...

From my readings, even current x86 chips are externally CISC. Just as you say, however, they appear to be mostly post-RISC internally. They decode CISC instructions into micro-ops, for example.
 
sabrewolf732 said:
I dont see how one can get results with software that is exactly the same on both platforms
:confused: They have to be coded differantly to be able to run on the differant platforms. Maybe I didnt fully get what you meant?? elaborate? :confused:
Click to expand...

They are "exactly the same" in their names and what they do but how they work internally to get that accomplished depends on the operating environment. You can't just take a copy of Microsoft Office 2003 for PC and then run it on a Mac. The apps have to run differently because of the different operating environments, thus making exact apples to apples testing very difficult.
 
draksia said:
It dedicated applications like consoles, routers, and DSP the risc architecture used by PowerPC, mips and others tends to be more efficent. Now that clockspeed has risen quite a bit the CISC architecture begins to shine in dealing with huge varities of task seen by modern cpus.
Click to expand...

CISC really doesn't have much meanning in todays computers. (pure RISC is very rare in something as complex as a full blown CPU as well, it's fairly common in more simplistic tasks like microControllers).

Assigns and jumps (loop, goto, if, call) take up about 99% of code size and 95+% execution time. Coding in HLL or ASM doesn't really change that. The idea that we should have these complex instructions to mate HLLs with the underlying hardware is well, wrong.


nylint said:
draksia, having been around as long as you have, you should know that x86 != CISC.

no current desktop x86 chip is a true CISC design. in fact, many have taken on pieces of RISC architecture.
Click to expand...

Whle the x86 ISA is deffinately a modern equivelent of a CISC architecture (it ain't PDP-11 or VAX CASE, but those have deffinatley fallen by the wayside).
You're right in that an x86 instruction, even the simple ones, don't get within 10 feet of an execution unit.
The actual execution core is decidely more RISC than the ISA. Large regsiter files, fixed lenght, single cycle (for the most part, FP still tends to be multicycle excution) micro ops, and hard coded micro-op routines replace complex instructions.
Of course you can't wish away things like the multitude of addressing schemes....

ShePearl said:
I'd like to know how x86 based PCs, (Intel Pentium4, and AMD Athlon64) stack up against Apple/IBM PowerPC CPUs ??

I mean, Microsoft chose to use IBM's PowerPC variant CPU for XBOX2, and even IBM is supplying CPUs for Nintendo Revolution, and Sony PlayStation3.

There must be a good reasons why these companies chose not to use x86 based Intel Pentium4 or AMD Athlon64 processors. Especially Microsoft (XBOX) who knows extremely well about x86 based processors. Their first (current) XBOX uses Intel's Pentium3 733MHz CPU...
Click to expand...

$$.
IBM is largest (second largest behind TSMC?) semiconductor fab in the world, volume discount baby.
You don't neccessarily need the most power CPU in the world to run a simple console, XBox does pretty well on it's P3 733, if a G5 1.8 gives a bit of ground to a P4 3.0 or A64 2.0, so be it.
G5 is the coolest, lowest power of any of the current major chips. (expect possible with respect to mobiles) Lower power means smaller PSU, less cooling = cheaper.
Consoles don't have a code base, XBox 2 doesn't need compatibility with XBox 1, no brand loyalty anymore...

My bet is, when all is considered, G5 was more cost effective.
 
It should also be noted that Apples version of the powerPC is quite a bit stripped down from the one found in IBM workstations and Servers.


It features on die memory controller, IBM version of hyper threading, and in the near future or possibly even now a dual core version.
 
Isn't there an easy way around the operating environment problem? Couldn't one just do all the benchmarks in Linux on both sides?
 
ShePearl said:
Hmm... I think you're mis-informed (no disrespect). I'm quite sure Apple/IBM's PowerPC CPU is the performance king when it comes to "desktop" CPUs. Just not sure how much better they are compared to aged x86 based CPUs such as Athlon64, and Pentium4.
Also, bear in mind that I'm comparing CPU performance aspect only. :)
Click to expand...

I love how someone asks for some information and then denies or doesn't believe the information they recieve because it doesn't support their beliefs.
 
having worked in embedded computing some years ago, I have worked with powerpcs 603e and g3.
Performance wise (using a C benchmark of my own compiled with gcc for both cpus) :
at same clock (we were in the 200-300Mhz days) G3s were a bit better than Pentium IIs, but not that much. Sun UltraSPARCs were much better (250MHz outperforming a 400Mhz PII).

But there was one real good reason to use powerpc : power consumption and heat, we had cpu with passive heatsink (and not huge ones).

I have not worked with g4 and g5 so I can't comment on those. But with the g5 they seem to have catch up (at least amd) on frequencies.
 
The Superlative said:
Isn't there an easy way around the operating environment problem? Couldn't one just do all the benchmarks in Linux on both sides?
Click to expand...


yea.
You you go with a linux distro for both x86 and PPC,
you use open source projects to control compiler optimization.
And you might appease the 'it's not the same, you can't compare it camp.'
But I doubt it. By the time you're down to 'its the oprating system that's makeing it slower, not the computer...' it's little more than an excuse, certainly not a reason.
If that's an argument, then you can't compare AXP, A64 and P4 either, they are very different processors that have very different needs from software.

Adobe makes good products, they have a good reputation for being pretty well optimized for any hardware they run on, the incorperate propriatary SIMD extensions, and in general have a reputation for producing high performance software. As far as I, and probably most people, are concerned, that's plenty close for a direct comparison.
You didn't fiddle to much with the OS?
aren't running anything stupid in the background?
you've got good well optimized code for the same app on the both computers?
you're running the same test on the same data in both cases?
Sounds reasonable to declare a winner to me.
 
Pc's are much Faster then MAC's and at a cheaper price. The only reason many peopel seem to think they are better for graphic design is becuase MAC's were the first with just about everything. First with firewire etc. They also had better monitors with highe resoltions. So the only reason many think MAC's are better for designing is because of its rep. Today PC's are much faster. Even Adobe says PC's are faster which is prtty bad since Apple uses Adobe products to compare performance.
 
I have a question for the OP........Why didn't you post this in the mac forum where you would get your disingenuous question answered to your satisfaction?
 
Deadlierchair said:
The apps have to run differently because of the different operating environments, thus making exact apples to apples testing very difficult.
Click to expand...

they are run differantly, they are coded differantly to be able to run.
:confused:
 
sabrewolf732 said:
they are run differantly, they are coded differantly to be able to run.
:confused:
Click to expand...

Yes? What I'm saying is that you can't just plop PC software on a Mac and get it to work. They are different pieces of software.
 
Deadlierchair said:
Yes? What I'm saying is that you can't just plop PC software on a Mac and get it to work. They are different pieces of software.
Click to expand...

No one 'ploped' PC software on the G5.
Premier and PS both have long histories of good, well optimized code running on the PPC. The results were, at best, mixed for G5 here, loosing easily in Premier accross the board, and aquitting itself nicely in PS, though the dual CPU Macs were still bested by the Dual CPU PC.

Office for Mac has been around since Office 1,

Q3 was on the Mac, what a few months after it was released on the PC?

we're not talking about emulations, or compatibility layers, or cheap hackery.
These are proffessional suites, compiled and run natively on the PPC by major develpment houses who have an interest in making their software run as well as it can.

If you don't like the results, find ones you do like.
If proffessional software is so biased towards the PC, there is a plethora of open source software, OSes included, that provide you the oppertunity to compare the hardware on your terms.

If the results posted so far present an agenda and not a comprison, don't whine, prove it.
 
stavroguine said:
having worked in embedded computing some years ago, I have worked with powerpcs 603e and g3.
Performance wise (using a C benchmark of my own compiled with gcc for both cpus) :
at same clock (we were in the 200-300Mhz days) G3s were a bit better than Pentium IIs, but not that much. Sun UltraSPARCs were much better (250MHz outperforming a 400Mhz PII).

But there was one real good reason to use powerpc : power consumption and heat, we had cpu with passive heatsink (and not huge ones).

I have not worked with g4 and g5 so I can't comment on those. But with the g5 they seem to have catch up (at least amd) on frequencies.
Click to expand...

This is the kind of comment we need. Factual arguments :)

What kind of device were you designing?
 
Deadlierchair said:
Yes? What I'm saying is that you can't just plop PC software on a Mac and get it to work. They are different pieces of software.
Click to expand...
i never said you had to do that, the peformance comparison I linked to didnt do that either. they had software which was made with a separate set of code for both platforms benchmarked
 
M3at said:
I love how someone asks for some information and then denies or doesn't believe the information they recieve because it doesn't support their beliefs.
Click to expand...
Because information/facts, and opinions are two very different things. Learn to distinguish between them.
 
ShePearl said:
Another article.

Analysis: x86 Vs PowerPC Winner seems to be PowerPC though.
Click to expand...


What did it win at? I saw no comparision... no screenshots with benchmarks showing anything. :rolleyes:


http://www.digitalvideoediting.com/2002/07_jul/features/cw_macvspc2.htm
http://www.robgalbraith.com/bins/multi_page.asp?cid=7-6451-6423

Now here is actually proof and it shows that Apple's Mac's are slower. :p
They cost to much and there are not alot games out for Mac's.

Until you can show me better review with benchmarks and comparisions then I must concur that you live in denial.
 
Deadlierchair said:
Yes? What I'm saying is that you can't just plop PC software on a Mac and get it to work. They are different pieces of software.
Click to expand...

Wasn't the xbox developer kit released on a specialized version of win2K running on a Dualie G5?
 
Just an add...instead of barking about faster/slower/blahblahblah, which is old and tired...why don't you look at the capabilities like gigaflops per second, math co-processors and junk like that? I think that's what was originally intended by the thread... :cool:
 
Here is your basic rundown

The P4 is fast
The A64 is fast
The G5 is fast

Any given month someone can ramp the clock speed of a chip and turn the tables on everyone else.

At this point the P4 had it's 4.0ghz chip cancled, the G5 needed water cooling to do 2.5ghz, but AMD seems to be on a roll and I believe the FX-55 is doing 2.6ghz.

I'd have to say the FX-55 is more than likely the fastest thing on the desktop right now.

They are all good designs for the most part. I'd argue that the P4 was a design mistake made by the marketing department, made a sucess by the process engineers for the most part.

So you'll need to be more specific if you want a better answer. General questions can only be answered with generalizations.

If you want to know whether a 2.0ghz G5 vs a 2.0Ghz A6(1mb L2 or 512K L2, 64bit or 128 bit memory) vs a 3ghz P4 (533bus or 800bus, HT?, rambus v ddr v ddr2, 512Kb L2 or 1MB L2 or 2mb L3).

So many different version of all these chips, Heck even apple makes changes to it's systems to speed up or slow down the systems for marketing. Their new dual 2.0ghz system is faster than their old dual 2.0ghz system thanks to a faster bus. And the new single 1.8ghz system is alot slower than their old 1.8ghz system because the made it a 600mhz bus to cripple perfomance to justify the lower price.
 
"Better" is a relative term. As previous posters have stated, modern x86 CPU's don't resemble a CISC processor at all. Think of modern processors as CISC->RISC->CISC converters. There are many reasons to do this, but most of it has to do with speed enhancements that CISC processors cannot do without a severe performance hit. Things like out-of-order executions, branch prediction, and pipelining are MUCH easier to do if all of your instructions take the same amount of clock cycles to perform. Pipelining is a great example of this.

Imagine you have an assembly line, and six workers operating in tandem to assemble a package. Now, to speed things up, you have broken up each step that your operators perform to take 10 seconds each. Now, while worker number one is finished doing his job, he passes it off to worker number two, and grabs the next package heading his way. This is very much the way a RISC pipeline works. Each instruction takes a uniform amount of time, and as such, you can build and "assembly line" to make sure that more stages in your line are kept busy at a given time. One package goes in, and one comes out per cycle. (Please note that this is a perfect-world analogy and does not take mispredictions or stalling into account). Very efficent.

Now, let's say that you can't break up these different steps, and instead, you have one worker peforming a series of operations to build a package. Each package that comes in takes a completely different amount of time to assemble. One might take 10 seconds, and one might take up to a minute. As a result, the amount of work you can do varies by the kind of instruction you are given, and you can't do more than one thing at once. This is how CISC operates.

The problem with RISC is that x86 programs were never written with this kind of simplicity in mind. As a result, all x86 programs would have to be re-written in order to take advantage of the new features a processor would have been built for. So, in order to eek more performance out of an aging platform, chip designers figured out a way to get the best of both worlds: RISC performance and backward compatibility. The CPU now can take a CISC instruction, and break it into a series of uniform RISC instructions. The amount of RISC instructions you will get out of one CISC will vary, but now you can use the whole assembly line routine to execute it.

When you couple this theory with several new tricks that processors can execute, you see that you do in fact get the best of both worlds. You can now use programs that were written for a 80286 on a modern chip, but have the ability to run new programs at blistering speed.

The architecture of the modern PowerPC is still a RISC at heart, and as such, there are no CISC to RISC conversions. You will also note that there are some striking similarities between AMD's 64-bit chips and the PowerPC. This is not an accident. IBM and AMD have a license sharing agreement, and IBM has been helping AMD with their fabrication over the last few years. The two companies have been co-developing technology for some time. Silicon On Insulator, the very technology that lets AMD's chips run the way they do, was co-developed by IBM.

So, in closing, the PPC and X86 CPU's really aren't all that different when you get right down to it. The thing that differentiates them from each other is the software that runs on them, and how it was written.

Oh, and FYI, someone mentioned a 'hybrid' version of Win 2K that would run on the PowerPC. This is actually quite viable. Windows NT was initially based of VAX/VMS (Dirty little rumor), and as such, was also built to run on DEC Alpha chips. There are still versions of WinNT available today for DEC Alpha. Since 2K is based of off NT to a large extent, it's not far fetched at all to think that Mickeysoft has a version for other RISC chips, such as the PowerPC.

Questions/Comments/Corrections are welcome.

Matt.
 
Wow, 2 pages and no one starting in about how OSX is better than XP (to obviously take the thread into a completely BAD direction).

I would think that if the game console is designed properly then you would be in a GPU bottlenecked situation most of the time anyhow, right? I'm looking past the current console generation. I sure hope that things like antialiasing and HD resolutions become common enough with the upcoming console GPUs.

I mean the CPU has to be "fast enough" but things like heat, power consumption, and low cost will be the drivers. Then the real ingenuity comes into play with the new GPUs.

B
 
brentsg said:
Wow, 2 pages and no one starting in about how OSX is better than XP (to obviously take the thread into a completely BAD direction).
Click to expand...
i appreciate osx, however i despise how next to nothing runs on it.
the one thing i really like mac for is because of the design (one button mouse aside)
all their cases, monitors and keyboards are designed very simply, yet effectively. it is very... well good looking.
 
Okay, this thread is about the IBM POWER series of processors, not the Apple that uses them...
 
The original point of this thread was about the XBox 2 and why it was using a PPC chip. For the most part it's piracy. The XBox was the only deviation of consoles using mainstream chips. They basicly made a small very powerful computer into a console, and look what it got them. My XBox basicly is my main computer, I have a keyboard, mouse, linux, games, ect... even media ripping. I can sit back on my insanely nice couch chair and use it in ease. They want to limit piracy. Not many Linux distro's are coded/compiled for ppc, infact I would venture to guess that ppc has less then 10% of all the software created. They also want to limit the knowledge of the chip to a little as possible to stop hackers.

As to the PPC vs x86, it's a toss up. Depends on the application you need. For consoles you don't need a 2.5ghz chip, I mean look at the XBox now... it's visuals are as stunning as most people's computers (most users of this forum are enthuasists, so please keep you comments of running at 1600xwhatever resolutions to yourself, you aren't a normal user. :) ) at any rate I forsee around a 1ghz chip for the next Xbox maybe 1.2 or so... more then enough to do anything for the next 3 years.

Personally, just looking at the way things are progressing AMD has the upper hand IMHO, they haven't hit the ghz wall yet, they make very cheap high preformance processors, and they don't need water cooling to make their computers run. I, and most likely many people don't care about how the instructions are processed, we care more about ease of use, price, ect....
 
You must log in or register to reply here.
Back
Top