Intel vs PowerPC

[BWF89]

I was at Apple's website trying to compare their 1.42GHx MacMini to a regular PC.

My PC had a 1.5GHz processor and my parents bought it mabye 4ish years ago. Now their selling MacMinis for $699 with only a 1.42GHz processor all these years later. Now I know Apple overprices their stuff because they have a monopoly on the hardware but how much better would for the sake or arguement would a 1.5GHz PowerPC processor be than a 1.5GHz PC processor? Or are they the no differences in the speed but a difference on how it operates?

[somuchfortheafter]

mac users correct me if i am wrong but i believe the way in which they process information is different therefore a slower processor still gives the power of today's machines, not exactly sure how that works but meh

[xequence]

Ghz isnt the only thing that determines speed... You know how all AMD processors have less Ghz, but are much better then their intel counterparts?

Though I dont know how good PPC is.

[aysiu]

I've generally found RAM amount to be a better indicator of noticeable performance difference than processor speed.

I also found Mac OS X to be a bit bloated. It runs extremely slow (both Panther and Tiger) on my wife's Powerbook (which has 1 GB of RAM), but when I ran the Ubuntu PPC live CD on it, it flew.

[BWF89]

I've generally found RAM amount to be a better indicator of noticeable performance difference than processor speed.

I also found Mac OS X to be a bit bloated. It runs extremely slow (both Panther and Tiger) on my wife's Powerbook (which has 1 GB of RAM), but when I ran the Ubuntu PPC live CD on it, it flew.
My current PC has around 512MB or ram and both the MacMini and the iMac only has "512MB DDR333 SDRAM".

So would switching from a 1.5GHz 512MB of ram PC to the same with a Macintsoh really be that much better?

[aysiu]

So would switching from a 1.5GHz 512MB of ram PC to the same with a Macintsoh really be that much better? Actually, it'd be about the same.

[prizrak]

Well PowerPC is a RISK CPU, what that means is that the instructions it uses are small and simple. They are combined to form more complex instructions but there are more being executed each clock cycle. So the lower clock speed still yields the same results.
I think that the Mini will run faster than the 1.5P4 or Athlon (whichever you got) but you would most likely be better off getting a faster PC (I built an AthlonXP 2100+ for about $300 like a year ago).

[spdl]

I read two seperate articles awhile back that Apple will be doing away with Mac hardware and just creating new OS's that will work on intel-based systems. I will try to find the artticle(s) in question later.

Which get's me thinking...OSX is Linux/Unix-based. So perhaps new Mac OS's will be a unique distribution on Linux.

So, since that is the case, I don't see why anyone would buy a Mac anymore. It's because most people/prospective customers do not know about this major company move.

[xequence]

I read two seperate articles awhile back that Apple will be doing away with Mac hardware and just creating new OS's that will work on intel-based systems. I will try to find the artticle(s) in question later.

Which get's me thinking...OSX is Linux/Unix-based. So perhaps new Mac OS's will be a unique distribution on Linux.

So, since that is the case, I don't see why anyone would buy a Mac anymore. It's because most people/prospective customers do not know about this major company move.

OSX is based on darwin, which is based on BSD, which is based on some unix version.

Why would something that has nothing to do with linux become a linux distro? It wouldent :P And if it was, apple wouldent make any money.

Apple is switching from PPC to x86 though.

[rfruth]

Officially the mid range Mac mini is 1.42GHz PowerPC G4 with 32MB video memory and combo drive (DVD/CD-RW) however new shipping mini's seem to be 1.5GHz with 64 video memory and a DL drive, some genius in Apple's marketing dept must think this undocumented upgrade is a good idea, personally I think it stinks - anyway the PowerPC G4 is a RISC chip where the x86 world is CISC so comparing a 1.5 GHz mini to a 1.5 GHz x86 box is like comparing apple's to oranges (ugh)

[5-HT]

Forgive me and my ignorance if this is completely wrong, but from my limited knowledge of the PowerPC and x86 processors, I believe that if you want a really rough estimate that doesn't account for a lot of factors...

For a rough comparison, I've read that PowerPC's have comparably twice the performance of x86's in terms of clock speed.

I suppose this has to do with the different RISC and SISC architectures and the number of operations that can be performed during each cycle.

Though I believe that modern SISC's like the pentium 4 utilize instruction sets more akin to RISC chips than their predecessors.

[poptones]

Well PowerPC is a RISK CPU, what that means is that the instructions it uses are small and simple. They are combined to form more complex instructions but there are more being executed each clock cycle. So the lower clock speed still yields the same results.

That's totally obsolete thinking. The AMD and Intel CPUs are capable of performing the same basic ops like shift, add, mul in a single clock. Using SSE or even basic MMX extensions you can do multiple shift/add/mul/saturate operations in a single clock. The whole "risc/cisc" debate became obsolete about the time of the slot 1 CPU. Modern cpus combine optimized data paths and microcode for basic ops with more complex, lnger microcode sequences for complex ops. Risc was only really of value back when it was impossible to cram a half billion transistors on a die.

[xequence]

PPC is definitally not double the power of x86 for clock speed...

OSX runs faster on x86 then PPC.

[BWF89]

but you would most likely be better off getting a faster PC (I built an AthlonXP 2100+ for about $300 like a year ago).
Well I was asking because my parents may be getting a new computer sometime in 2006. (I get the old one which I'll install Ubuntu on) and we upgraded from OS9 to OS10 in school and I really liked the operating system. So I figured my parents would be better off useing Mac because of it's stability and ease of use rather than upgrading to Windows Vista.

But if that's true might as well upgrade to a PC Vista than spend more and get less (hardware wise) with a Mac.

[5-HT]

Well PowerPC is a RISK CPU, what that means is that the instructions it uses are small and simple. They are combined to form more complex instructions but there are more being executed each clock cycle. So the lower clock speed still yields the same results.

That's totally obsolete thinking. The AMD and Intel CPUs are capable of performing the same basic ops like shift, add, mul in a single clock. Using SSE or even basic MMX extensions you can do multiple shift/add/mul/saturate operations in a single clock. The whole "risc/cisc" debate became obsolete about the time of the slot 1 CPU. Modern cpus combine optimized data paths and microcode for basic ops with more complex, lnger microcode sequences for complex ops. Risc was only really of value back when it was impossible to cram a half billion transistors on a die.

Thanks for clearing that up.



OSX runs faster on x86 then PPC

I Didn't realize OSX could run on x86 currently beacause of security precautions Apple has taken, well except for the few demos Apple has shown off in anticipation of their switch away from PowerPC.

[xequence]

I Didn't realize OSX could run on x86 currently beacause of security precautions Apple has taken, well except for the few demos Apple has shown off in anticipation of their switch away from PowerPC.

Not normal OSX, but, as with everything, there is a special cracked version :P

[BWF89]

Not normal OSX, but, as with everything, there is a special cracked version :P
I think running OS10 on a PC involves installing Ubuntu and then installing a bunch of Macintosh stuff ontop of it.

[JimmyJazz]

Apple is a hardware company I would be really suprised if they ever stopped making hardware.

[ssam]

ars technica has some very in depth articles about different cpu achitectures http://arstechnica.com/articles/paedia/cpu.ars

which is faster depends partly on what you want to do with the computer and whether you can optimise the code.

powerpc tend to be very good at floating point maths. this is good for multimedia and some scientific stuff.

x68 is usually faster for interger maths, good for more day to day stuff.

quite a lot of effort (it seems) in the history of computing has been converting floating point operations into interger operations, because most processors are better at interger maths. in the olden days some computers had a floating point co-processor.

there are also issues around pipelines. the cpu clock can only click at the time it takes for the slowest processing step. in the ghz world that is a problem. one way to make the clock tick faster is to breakup the task into chunks that take less time to complete. now each instuction goes through the cpu in a number of steps. this number is the depth of the pipeline.

one analogy is if you have to move a load of stuff, instead of picking it up from point A and carrying it to point B, you form a line of people from A to B, and pass from person to person. each pass get counted as a clock cycle.

this is more efficient as long as you make sure that as soon as the person passes on object another one is passed to them. you dont want the people hanging around doing nothing.

in a computer program sometimes you dont know what instruction to do until you have completed the previous one. in this case you get bubbles in the pipeline, and you waste time.

intel used deeper and deeper pipelines (and lots of other smart stuff) to keep the clock speeds rising. because thats how customers judged computer speed.

there are also things like vectorising code, doing several instructions at once. this can make a cpu faster if the code is properly optimised. GCC will soon be able to do this and the powerpc G4 and G5 should see some gains.

a mac mini probably wont be much faster then your current machine for most things. but if you think energy usage and size are important its a very good machine.

[almahtar]

OSX is based on darwin, which is based on BSD, which is based on some unix version.

Why would something that has nothing to do with linux become a linux distro? It wouldent :P And if it was, apple wouldent make any money.

Apple is switching from PPC to x86 though.

OSX's kernel is a hybrid of the BSD unix kernel and the Mach microkernel.

[almahtar]

I think running OS10 on a PC involves installing Ubuntu and then installing a bunch of Macintosh stuff ontop of it.

The PC version of OSX requires an entire physical disk: there's no way for it to coexist on a hard drive with another operating system unless you fool it with a virtual machine or the like. For that reason I will likely never even try it out: I like my Ubuntu.

The version circulating various torrent sites and the like is not a cracked version but a "leaked" beta, so it isn't supported because it isn't an official release, but it isn't cracked or modified either (to the best of my knowlege). I've seen it in action on a friend's PC, and it looks pretty stable.

[erikpiper]

But what if I have a spare HD?

[kairu0]

Then, why not? :)

[xequence]

I think running OS10 on a PC involves installing Ubuntu and then installing a bunch of Macintosh stuff ontop of it.

There are some different ways. One involves that, another is just a normal old installation DVD.

[xequence]

The version circulating various torrent sites and the like is not a cracked version but a "leaked" beta, so it isn't supported because it isn't an official release, but it isn't cracked or modified either (to the best of my knowlege). I've seen it in action on a friend's PC, and it looks pretty stable.

It is a version for developers to take their OSX software from PPC to X86. It IS cracked. Without the cracking, it could only be installed on x86 macs. With the crack it can be installed on any x86 computer.

[erikpiper]

Any1 want to PM me a link for the DVD image?

I am curious, and I would delete it for windows. I just want to see it.

[autocrosser]

Well-This isn't the first time that this has come up--Apple is moving to the X86 chip style because IBM can't keep the speed promises that Jobs made 6~12 months ago--I for one will mourn the "near" death of PPC--the pipeline was short & fast (that's how PPC was slightly faster:smile:) & not as much bad code was written for it (really, almost none--like Linux:smile:)--I was one of the betas for the early OSX & we knew that they were keeping a branch for X86 even in late '99/early '00--in other words--Steve was hedging his bets--You can bet that Apple will still make hardware--that is the only was they have stayed in business this long (after the dark days of the mid-'90's)--OSX is a good operating system, but it still won't fly as the sole money-maker for Apple--I'm betting that they create a code/hardware set so the only ones that run X86 OSX on non-Apple hardware will be Geeks (kind'a like us:rolleyes:)

I've been a Apple person for over 15 years now (I do remember 68K hardware) & the main reason I buy Apple products is the quality--My current unit (Dual 1.25 Mirror-door with 2G of ram) runs 24/7 & has done so from the day I bought it--No down-time other than when I un-dust the ducts & insides--So it has ran solid from October 2002 without a glitch one.
I would think that I could build a X86 unit that would do the same---but it would end up costing about the same I paid for my Apple.........

[BoyOfDestiny]

I've generally found RAM amount to be a better indicator of noticeable performance difference than processor speed.

I also found Mac OS X to be a bit bloated. It runs extremely slow (both Panther and Tiger) on my wife's Powerbook (which has 1 GB of RAM), but when I ran the Ubuntu PPC live CD on it, it flew.

Hmm... As someone who had 2 gig of ram with a athlon, vs 1 gig of ram with an opteron... I'd say cpu makes a huge difference for computational stuff... (Although I confess to ordering 2x512 sticks to upgrade my opteron box...). Indeed though, if you have less than 1/2 a gig, expect swapping...which kills performance...

Anyway, I had a professor, with his mac lappy, answer the question. He acknowledges that Intel is somewhat faster, but the design is more sloppy. PPC is risc based, and the idea is to have things cleaner and simple, so it would require, let's say, more simple instructions to run something that intel could do with complex messy stuff in one instruction (for example).

So what it comes down to is the ratio of cycle to instructions. If you have a ppc that can do 1 instruction per cpu cycle, and an intel box that can do the same (this is of course hypothetical, if you have anything like this, bury it...) However, there are so many factors (pipelining, caching, physical distances for certain parts on the board, bandwith, multi cycle blah blah blah), and I was sleepy in that course (so maybe I missed something)...

Anyway, apple is moving to intel for now, I think if they don't skimp out, people won't be horrified with performance. I'm hoping it will "seem" equal or better than what he/she expects. It may in fact be better, and once applications are optimized for it as well, better than ever.

[prizrak]

Yeah the Apple to Intel move has been confirmed, the "Intapple" systems are due to ship fairly soon (blanking out on the date). As far as performance goes the x86 OS X does run faster than the PPC one (according to Apple's own benchmarks AFAIK). The PPC architecture isn't really dead, but it's being replaced by cell anyway. Also the reason Apple is moving is not that IBM can't make the PPC fast enough it's that IBM is now shifting its focus to become a consulting company so they are more or less discontinuing their hardware operations (hence sale of the PC division to Lenovo).
I think Apple might be able to get away from the hardware stuff now that they have the iPod and finally enter as a competitor on the OS market. I personally hope that they can do it and put some pressure on MS anything that creates competition is good.

[mstlyevil]

Yeah the Apple to Intel move has been confirmed, the "Intapple" systems are due to ship fairly soon (blanking out on the date). As far as performance goes the x86 OS X does run faster than the PPC one (according to Apple's own benchmarks AFAIK). The PPC architecture isn't really dead, but it's being replaced by cell anyway. Also the reason Apple is moving is not that IBM can't make the PPC fast enough it's that IBM is now shifting its focus to become a consulting company so they are more or less discontinuing their hardware operations (hence sale of the PC division to Lenovo).
I think Apple might be able to get away from the hardware stuff now that they have the iPod and finally enter as a competitor on the OS market. I personally hope that they can do it and put some pressure on MS anything that creates competition is good.

IBM is not getting out of the PPC bussiness. Both XBox 360 and Sony playstation 3 are using IBM PPC chips to power their gaming consoles. It is my understanding that there is more money in making processors for specialized units like gaming consoles that there is for pc's. Also Apple moved to Intel largely because IBM refused to develop a mobile G5 chip for laptops and laptops are where the big money is at right now in the pc industry. Intel has several mobile processor platforms that Apple can choose from that perform equal to or better than a G5 because mobile processors have shorter pipelines than desktop processors.

[poptones]

Apple is moving is not that IBM can't make the PPC fast enough it's that IBM is now shifting its focus to become a consulting company so they are more or less discontinuing their hardware operations (hence sale of the PC division to Lenovo).

Is that the line Apple is spouting now?

Don't kid yourself: Apple moved to intel for the drm platform.

[prizrak]

Apple is moving is not that IBM can't make the PPC fast enough it's that IBM is now shifting its focus to become a consulting company so they are more or less discontinuing their hardware operations (hence sale of the PC division to Lenovo).

Is that the line Apple is spouting now?

Don't kid yourself: Apple moved to intel for the drm platform.
It's not their line, I just know that IBM is becoming mostly a consulting company and they weren't making enough money of the Apple sales to care about them.

IBM is not getting out of the PPC bussiness. Both XBox 360 and Sony playstation 3 are using IBM PPC chips to power their gaming consoles.
Yes they make good money of off those but Apple doesn't sell enough for IBM to really care about them.

[LordHunter317]

The short story is that today, unless you want to support Apple, really prefer their styling, or really prefer OS X. You get a better bang/buck buying an Intel-based machine.

This will eventually change when the Intel Machintoshes appear, but that's still some time coming. The price/performance point will disappear then, at least.


My current PC has around 512MB or ram and both the MacMini and the iMac only has "512MB DDR333 SDRAM".

So would switching from a 1.5GHz 512MB of ram PC to the same with a Macintsoh really be that much better?It would be noticably slower for most things, IME. OS X is a memory ***** and 512 MiB isn't really comfortable to run Tiger unless you're very patient (I'm not). And the CPU would be slower, so if you do any CPU intensive tasks, especially integer intensive ones, you'll definitely feel the burn.


Well PowerPC is a RISK[sic] CPU, what that means is that the instructions it uses are small and simple. They are combined to form more complex instructions but there are more being executed each clock cycle. So the lower clock speed still yields the same results.ITYM, RISC.

Modern IA-32 processors are RISC internally too. The Intel assembly instructions are decoded to several simpler micro-ops per instruction, which are then executed.

Also, Intel's Netburst architecture (Pentium 4) has a far deeper pipeline than the PPC 970. It can execute far more instructions simulatenously. This isn't a good thing though, per se. In the right situations, it's great. However, the penalty for when it's bad is also pretty significant.


For a rough comparison, I've read that PowerPC's have comparably twice the performance of x86's in terms of clock speed.Perhaps traditionally. Certainly not now.


powerpc tend to be very good at floating point maths. this is good for multimedia and some scientific stuff.No, the PPCs shipped by Apple are good at single-precision FP only due to Altivec. Otherwise, PPC is hardly a stellar FP performer. Itanium is far better otherwise, thoughtfar more expensive.


in the olden days some computers had a floating point co-processor.They still do now, it's simply on-die as opposed to being external.


there are also issues around pipelines. the cpu clock can only click at the time it takes for the slowest processing step.No, that's not true. The CPU ticks at whatever speed it ticks at. If a single instruction takes longer than a single clock cycle, it takes longer than a clock cycle.

Speed the processor runs at is limited based on the design of physical CPU. The time it takes to execute an instruction has nothing to do with it: we've had multiclock instructions since the beginning.


one way to make the clock tick faster is to breakup the task into chunks that take less time to complete. now each instuction goes through the cpu in a number of steps. this number is the depth of the pipeline.No, pipeline depth referes to the number of different instructions that can be executed simulatenously. On IA-32, that may be one instruction broken up into smaller u-ops, but it doesn't have to be. And frequently, it won't be: it will be several x86 assembly instructions executing in parallel.


Don't kid yourself: Apple moved to intel for the drm platform.:rolleyes: No, that has absolutely nothing to do with it. If they wanted DRM on their platform now, they could have it.

IBM has no interest in keeping their promises made to Apple about PPC 970, nor any desire to compete with Intel in the desktop/laptop space. Apple can't afford to keep buying from a vendor who won't keep them competitive.

At no point has there ever been even the slightest inclination that DRM has anything to do with this: if anything, they're moving to Intel simply so they can leverage the Centrino platform and have competitive laptops once again. Now that Intel-laptops are on par/superior to PPC in battery time, Apple's long-held advantage has simply disappeared.


It's not their line, I just know that IBM is becoming mostly a consulting company and they weren't making enough money of the Apple sales to care about them.IBM Global/Consulting services has nothing to do with it. IBM will still be producing processors for some time. However, they have no interest in competing with Intel in the desktop space, since it would be absolutely foolhardy for them to do so. They lack the fab capacity and they'd have to just produce another IA-32 clone anyway.

[prizrak]

ITYM, RISC.

Modern IA-32 processors are RISC internally too. The Intel assembly instructions are decoded to several simpler micro-ops per instruction, which are then executed.

Also, Intel's Netburst architecture (Pentium 4) has a far deeper pipeline than the PPC 970. It can execute far more instructions simulatenously. This isn't a good thing though, per se. In the right situations, it's great. However, the penalty for when it's bad is also pretty significant.

I have heard at some point in my life that modern x86 CPU's combine the CISC and RISC wasn't sure of the specifics.

IBM Global/Consulting services has nothing to do with it. IBM will still be producing processors for some time. However, they have no interest in competing with Intel in the desktop space, since it would be absolutely foolhardy for them to do so. They lack the fab capacity and they'd have to just produce another IA-32 clone anyway.

Reply With Quote
Umm yeah that's pretty much what I said, there is no point for them to support Apple there is not enough of a market for them. Xbox 360/PS3 is a much larger market and they win no matter which of them is more popular ^_^ well unless Revolution manages to capture some market share (HAHA).

[LordHunter317]

Umm yeah that's pretty much what I said,No, you said it was because IBM is becoming a consulting company. They're not completely and that has nothing do with. Very simply stated: it makes no business sense for IBM to compete with Intel, regardless of purchaser, so they won't.



well unless Revolution manages to capture some market share (HAHA).Revolution is PPC based and fabbed by IBM so it doesn't matter. All next-generation consoles are being built by IBM.

[mstlyevil]

I personally think Apple realised they would now be last on the list for PPC chips because IBM was going to concentrate on supplying gaming console companies. Also as stated earlier by me and Lord Hunter, Apple struggled to get next generation laptop CPU's and technology through IBM because IBM in recent history has shown that they have lost interest in the pc market and did not want to keep up any more. In the long run it only makes sense that Apple align themselves with Intel because if they are to remain competitive, they need to offer everything you can get in a x86 and more.

Apple may deny that they will offer a Macrosoft computer in the future, but their admitance that you can run Windows on Mac tells me that if the demand is there they will change their minds to increase sales and improve the bottom line.

[Ride Jib]

Clock Speed is a MARKETING SCAM!!



ars technica has some very in depth articles about different cpu achitectures http://arstechnica.com/articles/paedia/cpu.ars

which is faster depends partly on what you want to do with the computer and whether you can optimise the code.

powerpc tend to be very good at floating point maths. this is good for multimedia and some scientific stuff.

x68 is usually faster for interger maths, good for more day to day stuff.

quite a lot of effort (it seems) in the history of computing has been converting floating point operations into interger operations, because most processors are better at interger maths. in the olden days some computers had a floating point co-processor.

there are also issues around pipelines. the cpu clock can only click at the time it takes for the slowest processing step. in the ghz world that is a problem. one way to make the clock tick faster is to breakup the task into chunks that take less time to complete. now each instuction goes through the cpu in a number of steps. this number is the depth of the pipeline.

one analogy is if you have to move a load of stuff, instead of picking it up from point A and carrying it to point B, you form a line of people from A to B, and pass from person to person. each pass get counted as a clock cycle.

this is more efficient as long as you make sure that as soon as the person passes on object another one is passed to them. you dont want the people hanging around doing nothing.

in a computer program sometimes you dont know what instruction to do until you have completed the previous one. in this case you get bubbles in the pipeline, and you waste time.

intel used deeper and deeper pipelines (and lots of other smart stuff) to keep the clock speeds rising. because thats how customers judged computer speed.

there are also things like vectorising code, doing several instructions at once. this can make a cpu faster if the code is properly optimised. GCC will soon be able to do this and the powerpc G4 and G5 should see some gains.

a mac mini probably wont be much faster then your current machine for most things. but if you think energy usage and size are important its a very good machine.

@ssam: Thank you for saving me a lot of typing.
@everyone else: this guy has it right on the money. make sure you read the quoted

[LordHunter317]

Clock Speed is a MARKETING SCAM!!It's hardly a marketing scam, as long as you know how to apply it. Which means, you can't compare it across architectures. But everyone who knows anything about benchmarking knows that.


@ssam: Thank you for saving me a lot of typing.
@everyone else: this guy has it right on the money. make sure you read the quotedNo, his post has several basic technical errors. Did you even read the entire thread? He's confused about pipelining, why processors are clocked at the speeds they are, and the comparative FP performance differences.

[prizrak]

No, you said it was because IBM is becoming a consulting company. They're not completely and that has nothing do with. Very simply stated: it makes no business sense for IBM to compete with Intel, regardless of purchaser, so they won't.
There was another post by me above stating that Apple is too low a market share for IBM. Them doing consulting is actually part of not supplying Apple anymore, they are concentrating the majority of resources on creating the consulting infrastrcture so they cut the least profitable branches.

It's awesome that IBM is doing CPU's for all consoles, they really can't lose :)

On a separate note, clockspeed isn't really a marketing scam. It would be pretty dumb to say that a P4 1.5GHz is slower or is equal to a p4 2.0GHz. It's just that lately the clockspeed means alot less when it comes to benchmarking since the CPU's are reaching their highest potential clock speeds.

[mstlyevil]

There was another post by me above stating that Apple is too low a market share for IBM. Them doing consulting is actually part of not supplying Apple anymore, they are concentrating the majority of resources on creating the consulting infrastrcture so they cut the least profitable branches.

It's awesome that IBM is doing CPU's for all consoles, they really can't lose :)

On a separate note, clockspeed isn't really a marketing scam. It would be pretty dumb to say that a P4 1.5GHz is slower or is equal to a p4 2.0GHz. It's just that lately the clockspeed means alot less when it comes to benchmarking since the CPU's are reaching their highest potential clock speeds.

I was just reading a review the other day on the new 65nm P4. Intel has not upped performance but has gotten the power leakage and heat issues under control on that processor which means that they will be able to raise clock speeds up further in the near future. I don't think processors have reached their highest potential clock speeds yet.

[Ride Jib]

No, his post has several basic technical errors. Did you even read the entire thread? He's confused about pipelining, why processors are clocked at the speeds they are, and the comparative FP performance differences.

Yes, yes. I see that. I was caught up in everyone being mislead in the begining of this thread, I got tired of reading and therefore didn't see your post. I started skimming for someone that was on the right path. I figure if anyone is actually interested, they will further research the topic outside of a forum, and his concepts (although containing basic errors) are enough to get people the information they need. But also, thank you for your post of corrections (and for the patience you have that I lack).

[LordHunter317]

There was another post by me above stating that Apple is too low a market share for IBM.No, I don't think that's it. IBM cares little about that. It's the entire market they're asking to compete in: the desktop/laptop space. They simply have zero interest in being there.


Them doing consulting is actually part of not supplying Apple anymore, they are concentrating the majority of resources on creating the consulting infrastrcture so they cut the least profitable branches. Not really. If that were true, then they would be abandoning PPC in all but their mainframes, and they're not. Consulting services has nothing to do with it. Just pretend it's completely seperate company (which is it is, for all intents and purposes).

It's simply as I put it: they don't want to compete with Intel. It makes no businsess to do so. The fact IBM is heavily a consulting company has nothing to do with it. That single fact I'm repeating is the only reason they're not interested in doing business with Apple.

[blueturtl]

I once had the honour of comparing a brand spanking new Dell with a 1.5 GHz Pentium4 chip against my own custom built PC with an Athlon CPU clocked at 800 MHz, and can you believe my Athlon outperformed the off-the-shelf Dell? It outperformed the Pentium4 even in 3D benchmarks and games although the Dell had a faster graphics chip (I had a Ti4200 and the Dell owner had a Ti4600). That was the day I stopped believing in clockspeed as a definite performance maker. The amount of RAM was equal in both machines. I believe the Pentium4 was bottlenecked by the used chipset or RAM or something, but then again it doesn't really matter to the owner of the Dell now does it? He was sure he'd be getting a faster machine because it had a higher clockspeed than mine.

When it comes to CPUs, the chip design has very much to do with how the chip will perform. Differences in architecture will cause differences in speed for different kinds of processing. The clock speed is only a factor in comparing similar CPUs against each other. You might compare a Pentium3 against a Pentium3, but you cannot compare Pentium3s and PowerPCs. This is because the chips are designed differently and thus an 800 MHz Pentium3 will not perform the same as an 800 MHz PowerPC. The IBM CPUs are an efficient design and it is because of the design that a PowerPC of smaller clock can outperform Intel CPUs at the same tasks. Not always, but sometimes.


Now their selling MacMinis for $699 with only a 1.42GHz processor all these years later. Now I know Apple overprices their stuff because they have a monopoly on the hardware but how much better would for the sake or arguement would a 1.5GHz PowerPC processor be than a 1.5GHz PC processor?

What Apple focuses on instead of bragging with their tech specs is what you can use the device for. An acquaintance of mine used to own a G3 (if I recall it was clocked somewhere around 300 MHz) and he was doing video editing on it! I remember thinking how it must have been a really turbo computer but it was really much more modest than top PCs of the time. How many of you did easy and fast video editing on a Pentium the same speed? The Mac mini is a real miracle. It's a computer the size of a cd travel case and you can 1) browse the web 2) do office work 3) view and burn CDs/DVDs, 4) listen to music/watch video files 5) etc... The only thing the Mini can't really do well is play heavy 3D-games, but then again, with the cooling new graphics chips require how would they ever have achieved such small size! Is it faster than a turbo PC? Propably not. It however is fast enough for all those tasks it was intended for and you can almost forget the fact you even have a computer when considering placing things in your room because it's so small. Typically a new PC may be faster than an old PC but it is also bigger, noisier and still butt-uggly. I'm not in any way affiliated with Apple, but I think they're heading the right way.

Why aren't we all on Windows? Because with Windows our PC hardware is much more limited than with Linux. With Linux we get more freedom, we're able to do things we can't do with Windows, we're able to do them better in many cases. If it weren't for Linux I would have sold my newer box (after being severly disappointed at how WinXP was a step backward in ease of use) and bought a Mini. Now though I'm a happy ubuntite and won't be switching to anything before my hardware fails me.

[yesplease]

To answer the thread starters question: when you want to compare computers it is best to compare their real-life performance. So, go to a few hardware sites like tomshardware or anandtech and see if you can find benchmarks.

Because the Macs also have their own software, you should read some reviews on that too. My understanding is that Mac software is stable and secure.

[LordHunter317]

The IBM CPUs are an efficient designThey are a power efficient design, meaning they can perform more operations with the same amount of power input. This is why PPC was generally (And still is) used in embedded designs.


and it is because of the design that a PowerPC of smaller clock can outperform Intel CPUs at the same tasks. Not always, but sometimes.The bottom doesn't translate to that.

And PPC isn't more "clock efficient" whatever that would mean. It doesn't manage to perform more operations per single clock cycle or anything of the sort.


What Apple focuses on instead of bragging with their tech specs is what you can use the device for.Utter nonsense. Altivec is useless for most applications but that didn't stop Apple heralding it throughout the entire Gx era. The next thing will be power/performance, and that's only relevant for laptops...

[blueturtl]

And PPC isn't more "clock efficient" whatever that would mean. It doesn't manage to perform more operations per single clock cycle or anything of the sort.

I could be mistaken, but doesn't the fact the PPC uses less and shorter pipelines transfer directly to that? I mean that it could be faster in certain situations than a system running at higher speed but that has longer/more pipelines?

[prizrak]

blueturtl:
If you want a powerful and small machine that looks nice get a Shuttle barebones system. It'll do everything the mini will and more, the newer ones even come with a PCI Express for video (AFAIK). So you can play 3D games ;) The entire Macs look better than PCs debate has been beaten to death, and the Mac lost. There are so many different cases for PCs that it is impossible not to find one you like. There are also liquid cooling systems for PCs to make them quieter (sp?). The era of Mac hardware has passed sorry but you will need to get over it :)

[LordHunter317]

I could be mistaken, but doesn't the fact the PPC uses less and shorter pipelines transfer directly to that?It can, in certain situations. It can go the opposite way too.

Efficency defined in terms of "operations" per clock cycle is a tough thing to measure, because it isn't constant with respect to runtime. It's also not interesting in any real-world design when determining what processor you use. Either the processor can handle your load at your given power/heat/whatever requirements or it can't. THe number of cycles required to do it are rather uninteresting.

[prizrak]

Which should just look at FLOPS for the performance :)

[ssam]

have a read of http://arstechnica.com/articles/paedia/cpu/pipelining-1.ars/1 that explains pipelining quite well.

i think pipeline terminology can get a bit muddle but i think i was correct in saying that the high clock speed of the intel cpus is achieved by having a very deep pipeline. each instructions takes many little steps through the processor. each step is on a clock tick.

this becomes ineffiecient if you get bubbles in the pipeline, which happen if the code has branches.

http://arstechnica.com/articles/paedia/cpu/p4andg4e.ars/1 is a direct comparison between the P4 and the G4.

i beleive that the cpu discussions at arstechnica are good and accurate.