why are macs so goddam expensive???

[PFVA]

I'm not a computer expert unlike the rest of you guys but i have acouple of little points:

Not just Macs but electronics in general are more expensive in the UK and Europe than here parcially because the tax is close to 10% higher than ours, and don't forget that here when we see the regular IMAC for 1300.00 the 8.something tax is nt included yet. In addition your currency is quite stronger than the dollar.

If i go to a store here i probably will be able to get a Pc for 750.00/800.00 instead of the 1300 the IMac will cost me but i doubt it will be a PC with a quality equivalent to the IMac, and after i add all the additionals to make it equivalent the price won't be that far apart. It's probably different though if you build your own machine as there is more available parts to put PCs together than Macs.

Did i mention i love Macs? and i don't find the 1300.00 for the regular IMac or the 1500.00 for the other one expensive. It's me who has financial problems requiring that other priorities be followed otherwise i would be typing right now on a llovely newer IMac instead of my old one.

[pogo]

How can you compare x86 and PPC and say stuff about "jumping through hoops"? PPC is RISC, for christs sake. It'll take you at least 5 instructions to just call a subroutine, whereas on x86 it takes just one.

PPC:

1+2 : lets get the subroutine address.
3: oh shit, it doesn't stack return addresses. Better do that myself.
4: w00t! call the subroutine!
5: mmm, better get that return address back...

x86:

1. Call.....

That doesn't make the x86 faster. The PPC can use different function units (the G5 has 12) to handle the different parts of a subroutine call, so can do all that (plus some FP and vector math) at once.

More usefully because the parts of the subroutine call are distinct they need not be done at the same time: e.g. the assembly programmer or compiler can schedule the address fetch earlier to avoid stalls if it's not in the cache.

Incidentally thats also the reason I just laugh at people who are still saying stuff like "PPC is faster than x86 at the same clock speed"

RISC is old hat imo. It was a good idea at the time - simple system means instructions run in 1 cycle and it can be clocked higher. It will take you more instructions to do stuff but all in all it will still be faster.

Except that doesn't apply anymore, 80% of x86 instructions also run in 1 cycle, they clock higher than PPC anyway, and you have all the benefits of a nice, friendly CISC architechture to go with it. Altivec? The P4 has MMX,SSE,SSE2 - Altivec is no advantage.

IMHO the results speak for themselves: if RISC was so much slower it would be reflected in [application and synthetic] benchmarks. But at similar clock speeds PPC machines are faster than x86. The x86 architecture may be easier to program but it isn't faster.

As for Altivec, it's SIMD done right first time, instead of the three attempts by Intel which together aren't as capable. The simplicity and symmetry of the architecture (32 AltiVec registers, 32 FP registers, 32 integer registers) makes it easier for developers to fully exploit and compilers produce the best code.

[Snakey]

That doesn't make the x86 faster. The PPC can use different function units (the G5 has 12) to handle the different parts of a subroutine call, so can do all that (plus some FP and vector math) at once.

Yea, yea, so can the P4.

More usefully because the parts of the subroutine call are distinct they need not be done at the same time: e.g. the assembly programmer or compiler can schedule the address fetch earlier to avoid stalls if it's not in the cache.

See above. Besides, they still need to be done, whereas they DONT on x86, so it can be doing something more useful instead.

IMHO the results speak for themselves: if RISC was so much slower it would be reflected in [application and synthetic] benchmarks.

Bog did a side by side comparison on some lightwave reders and seems to be happy that the PC choice for him is the best one.

My *own* tests with various apps bear out the fact that its just not the case, too.

I am not anti-mac (although I'm not really what you'd call a fan, I do own a mac, but I don't much like OSX, for example) but its all the insistance that "macs are just better" that does my head in. Sure, they look nice and they work well, but the myth they are better in every possible way is just not true. You hear loads of "it just works" comments, but strangely, nobody says anything when "it just doesn't work".

You hear a lot more people moaning about Windows (even though XP is pretty much rock solid.) I wonder why that is... could it be because its got a HUGE userbase, perchance ?

[pogo]

More usefully because the parts of the subroutine call are distinct they need not be done at the same time: e.g. the assembly programmer or compiler can schedule the address fetch earlier to avoid stalls if it's not in the cache.

See above. Besides, they still need to be done, whereas they DONT on x86.

Which bit doesn't it need to do on x86 ? There may not be a seperate function call for it but it will still have to get the subroutine address, save the old address, etc., or it will have a hard time keeping track of where it is. All the PPC does is make it clearer what work's being done and gives the compiler/programmer extra opportunities to optimise code.

[Snakey]

Which bit doesn't it need to do on x86 ? There may not be a seperate function call for it but it will still have to get the subroutine address, save the old address, etc., or it will have a hard time keeping track of where it is. All the PPC does is make it clearer what work's being done and gives the compiler/programmer extra opportunities to optimise code.

Bah.

It doesn't need seperate execution units to read the address (in two parts) for a start... Besides, that was just an obvious example. It takes two instructions to fill a 32bit register. The x86 does that in one, and only takes 1 cycle (or "function unit") to do it.

Anyway, I didn't want to turn this into a CPU argument, its pretty much pointless since most of the people here who own macs have already said macs aren't necessarily as fast - and that doesn't matter.

[Kronos]

ok i admit i dont know everything about the mac but i do know two things (well they might be wrong but hey)

1. generally in speed terms the old 680x0 processors were at least 150% more efficient in mips/processor speed compared to x86 and lets face it windows 3.1 or Amiga WB 3 *evil cackle* BUT dual processor crap is only 25% efficiency increase at best by adding a second CPU (unless its a render farm situation for lightwave etc)

2. things have moved on finally with the athlon64 and specific XP compilation combination soon to become reality.

dont get me wrong OS X looks better than the XP gui and the mac cube is gorgeous but the days of caring who makes my computer ended when commodore and atari hit the wall a long time ago. im not going to pay a premium for a computer thats made by a semi-bespoke company. AFAIC the last computer i would pay a premium for breaking the mould was the Amiga 1000 in the days of the $2000 286 CGA/EGA crap alternative or $3000 macs with no colour!! and the atlernative was something like 5-10 faster, had unobtainable specs for the other machines and WAS CHEAPER!! jay miner, may you rest in peace forevermore


steve jobs had his chance and fucked it up before windows was even shipped by givin his enemy a mac to work on pre-release, apple is just cashing in on the "i hate microsoft" thing. i do hate microsoft but im not going to make apple richer for it. i can run linux or BeOS or something if i really wanted to i can make my machines boot into an emulator and have an Amiga to write my documents on. im talking about the hardware inside the boxes.

if you really hate microsoft you dont have to get a mac per-se.....just get the best pc box in looks and performance to your taste and go down your local car boot sale with a couple of tenners if we all did this maybe MS would crash yayyyyyyyyy!!

[Kemper_Boyd]

IMHO the results speak for themselves: if RISC was so much slower it would be reflected in [application and synthetic] benchmarks. But at similar clock speeds PPC machines are faster than x86. The x86 architecture may be easier to program but it isn't faster.

You should look at some decent benchmarks the 2 GHz 90 nm Dothan is about
40% faster on SPECint 2000 than the 2 GHz 90 nm PPC 970fx and uses about
half the power to boot. With a decent FSB and clocked up to desktop power
levels it would probably outperform even the fastest Opteron.

*edit*

Dell "Mobile Workstation M60" 1528/1087 SPECbase2k
Dell Precision Workstation 360 (3.4E GHz Pentium 4) - 1418/1489

See this is one of those stats that shows how peoples intuition about
computer performance is often wildly wrong. A laptop beating Intel's
latest and greatest desktop processor? Yes. Right now Intel has
faster processors in their laptops than their desktops. Oh and in
this case it's at about a quarter of the power too.

[Anders]

Still, haven't seen that many "fuck me I love my PC" posts

I love my PC. I just don't feel the need to rub it in other peoples faces.

What I like about Macs is the design, the look of it all, the interface, etc. I don't really give a toss about the hardware.

I guess Apple charge what they do, because they can get away with it.

[ji]

First, comparing processors by gigahertz (megahertz, whatever...) does not work. It might work, if you know that the compared processors are similar in every other way, and x86s and PPCs certainly are not.

I don't even see, how one could compare processors of different families with any kind of code. They have different instruction sets, different architecture, different everything. So I don't believe it, if/when Apple says "our processors are as fast as x86 with twice the gigahertz", neither do I believe that 2.8 GHz x86 is faster than 2.8 GHz PPC. It depends on the code waaaay too much.

So, what's left is your feeling of speed. Which ever one does the job for you faster is faster.

Why I'm looking for Macintoshes these days, is that I'm comlitely fed up with Linuces über geekness. I don't have anymore all the time in the day to fix and adjust the OS, I want to do my job. Apple's environment is the most promising, since it has the best parts of Unices, and a nice enough GUI, should I need one. Redmont product family is too restricted for my mind, althought something funny is going on there too, they are giving a compiler for free too.

Summa summarum, which ever is faster for you is the fastest. And it might be even your years old Amiga... (Or am I complitely wrong with my opinion?)

BTW, would Yak care to comment on programming PPC and x86 on assembler level, as he might have quite a lot experience on both of them?
As Snakey said, it does take quite a lot of code to write easy things on PPC, but as Pogo said, it gives a lot of possibilities for optimisation. But there are macro assemblers available for both processors, aren't there.

[Fuzzle]

When CISC meant microcoded and RISC meant pipelined, RISC won easily. That was a long, long time ago. Now they use almost identical archetectures.

Obviously, the decades of cruft on x86 doesn't help, but if you spend enough on engineers, you can make very fast processors. You can even argue that the extra control logic of CISC takes up less room than the cache for the longer programs of RISC. I think we'll see more complicated plastering of new over old with the hardware sorting out.

So, what's left is your feeling of speed. Which ever one does the job for you faster is faster.

I think it's down to that. Or more to the point: what can get the job done for you. If one fit every need, we'd all be using that one.

I don't really do anything needing huge amounts of computation, so I think a Mac would be a good fit for me for my next machine. Though it would be sweet to play Half-Life 2.

[Kemper_Boyd]

Obviously, the decades of cruft on x86 doesn't help, but if you spend enough on engineers, you can make very fast processors. You can even argue that the extra control logic of CISC takes up less room than the cache for the longer programs of RISC. I think we'll see more complicated plastering of new over old with the hardware sorting out.

Which is why the less complicated Dothan beats the more complicated
Prescott and uses about a quarter the power? Complicated means
more transistors, means leakage, means heat, means higher R&D,
means longer development, means more testing, means less cache,
means lower performance. Take a look at the Itanium2 compared
to a contermporary x86 it's a simple beast but delivers killer performance,
particularly relative to clock speed. Quite simply the complicated x86
chickens are at long last coming home to roost. Most RISC families
are also up to their neck in shit too so PPC fans shouldn't start cheering
just yet.

[Fuzzle]

Which is why the less complicated Dothan beats the more complicated
Prescott and uses about a quarter the power? Complicated means
more transistors, means leakage, means heat, means higher R&D,
means longer development, means more testing, means less cache,
means lower performance.

OK, OK... Simpler is better in the hardware. But really I was talking about the instruction set. SSE3 and hyperthreading doesn't explain Dothan vs. Prescott (4Ghz no matter what explains it). Itanium kept (slow) backward compatibility. Instruction sets do make a difference, but they aren't what's driving complexity or speed. If they did, then x86 would have roosted a couple decades back. Graphics accelerators are an example of setting the hardware free from a crufty instruction set, not PPC or Itanium.

[romanista]

Macs do have a lower speed/price ratio than most PC's.

Still, haven't seen that many "fuck me I love my PC" posts

this seems like the old megahertz myth still works... a 2 ghz pc isn't faster than my 1 ghz mac where all compontent work better together and the bus speed isn't slower..

[romanista]

On this page you can watch apple explain it in a little movie http://www.apple.com/nl/g4/myth/

[mightyzim]

I don't think you'll find its just the Ghz making the difference, rather the cost, use and availability of the units.
As Sty pointed out Apple is brand name, there is limited supply for the demand and the unit itself is closed as opposed to the pc which is an open standard(s).
Most people want a pc (the thing they use at work) for the cheapest price possible, they want it to run windows (the thing they use at work) and they don't want to have to bother with anything technical (like the tv, just 4 channels and volume please)

It takes somebody with knowledge to know what the difference between a pc and mac is and why they would want to buy one. In the past it was normally for marketing work, desktop publishing and graphics work. This part of market has been pushed over to the pc side due to the increase of processing power and graphic chipsets.
Apple has gone back to making itself a brand and a premium brand at that. This means their units *must* be expensive compared to their competitor in order for them to keep the premium brand.