The truth about Jaguar, your current Mac, and Apple...
- Thread starter MacLuv
- Start date
symphonix
Scratch & Sniff Committee
*Pulls out flamethrower*
The x86 processor is the last legacy processor in the computing world. It doesn't use RISC, and it is a 32 bit extension to a 16 bit extension to an 8 bit processor. I'm tired of hearing this. If you want to use an x86 processor, go out and get a PC.
*Puts flamethrower down, now that MacLuv is nicely toasted*
That said, there are undoubtedly good things in store for Apple in the not-too-distant future. I dont want to add fuel to the rumours, but we can be reasonably confident that the new 64-bit RISC chipset IBM is now producing is destined for Apple, since the chips are apparently equipped with the Altivec (G4) extensions.
I am satisied that we now have a 'real' operating system. It runs great for me, even on a 128mb G3 500 system, with only a few hiccups with the Java machine to detract from what would be perfection (almost sorted out). Though I wouldn't recommend trying it on anything less than that.
So I can't just rip the box open and put in a new processor, so what? If that's what I wanted, I'd have bought a PC. Instead of spending $500 a year on upgrades, I'm spending $1500 every third year on a brand new system - I think it works out better for me.
The x86 processor is the last legacy processor in the computing world. It doesn't use RISC, and it is a 32 bit extension to a 16 bit extension to an 8 bit processor. I'm tired of hearing this. If you want to use an x86 processor, go out and get a PC.
*Puts flamethrower down, now that MacLuv is nicely toasted*
That said, there are undoubtedly good things in store for Apple in the not-too-distant future. I dont want to add fuel to the rumours, but we can be reasonably confident that the new 64-bit RISC chipset IBM is now producing is destined for Apple, since the chips are apparently equipped with the Altivec (G4) extensions.
I am satisied that we now have a 'real' operating system. It runs great for me, even on a 128mb G3 500 system, with only a few hiccups with the Java machine to detract from what would be perfection (almost sorted out). Though I wouldn't recommend trying it on anything less than that.
So I can't just rip the box open and put in a new processor, so what? If that's what I wanted, I'd have bought a PC. Instead of spending $500 a year on upgrades, I'm spending $1500 every third year on a brand new system - I think it works out better for me.
Rhino_G3
Registered
Although I do think that apple definately does need to do something about processor speed I don't think that x86 is the answer. There would be too many conversion issues. Just having gone through the transition to OS X that would royaly tick many developers off to the point of them not supporting the Mac anymore. If a change does occur it HAS to be backward compatable with what we're currently running. Not even so much as a recompile of an app to get it running properly. I do agree with you though. Something will have to be done.
I've been following all the arguments flying around recently regarding PC processors (AMD, Intel) vs Mac processors (be it IBM or Motorola) and the whole RISC vs CISC thing and still find my self unclear on a few things.
As the message states above, CISC has basically won the speed war despite it's less elegant (inferior?) design as there has been so much brute force effort and money thrown into it the defects and weaknesses have been pretty much circumnavigated.
How long can this go on for though? I remember reading quite a while ago that they were reaching the limits of what could be done with CISC, yet still the speeds go up. Will ever decreasing returns mean there is a point where they have to stop (as the cost of getting a few more mhz finally outweighs the benefits of doing so) and completetly start again with a whole new methodology - as Apple effectively did with the move from '040 to PowerPC.
If this point isn't far off then perhaps Apple/IBM/Motorola/Whoever have half a chance as they are already down this road.
Is it perhaps also the case that the whole speed issue is a bit of a red herring and not the be all and end all some people seem to think? Even if Apple's were twice as quick as the average PC would it make that much difference to their market share - a lot of people use PC's and will continue to use PC's no matter what. Repurchasing your softaware library isn't always an option when making the famous 'switch' - among with numerous over issues.
As the message states above, CISC has basically won the speed war despite it's less elegant (inferior?) design as there has been so much brute force effort and money thrown into it the defects and weaknesses have been pretty much circumnavigated.
How long can this go on for though? I remember reading quite a while ago that they were reaching the limits of what could be done with CISC, yet still the speeds go up. Will ever decreasing returns mean there is a point where they have to stop (as the cost of getting a few more mhz finally outweighs the benefits of doing so) and completetly start again with a whole new methodology - as Apple effectively did with the move from '040 to PowerPC.
If this point isn't far off then perhaps Apple/IBM/Motorola/Whoever have half a chance as they are already down this road.
Is it perhaps also the case that the whole speed issue is a bit of a red herring and not the be all and end all some people seem to think? Even if Apple's were twice as quick as the average PC would it make that much difference to their market share - a lot of people use PC's and will continue to use PC's no matter what. Repurchasing your softaware library isn't always an option when making the famous 'switch' - among with numerous over issues.
Krevinek
Evil PPC Tweaker
MacLuv, even if they were willing to release a whole OS X for x86, developers would literally dump the Mac platform.
Apple pushed them through the 9->X transition, and many didn't make it. Quark is showing itself to be a good example alongside the companies Apple ate up to bring software to X. Developers would not be too happy if they had to go and re-optimize their G3/Altivec/VMX code for an x86 chip within two years of having to port and re-optimize for a new OS. Plus you have the fact that the x86 SIMD units are not compatible with each other, making SIMD optimizations take longer.
Apple won't budge from the PowerPC RISC platform until it has to. It is business, so why would Apple shoot the other foot when it has an option for catching up somewhat? You may not trust IBM, but Apple will probably have to rely upon IBM for now.
Apple pushed them through the 9->X transition, and many didn't make it. Quark is showing itself to be a good example alongside the companies Apple ate up to bring software to X. Developers would not be too happy if they had to go and re-optimize their G3/Altivec/VMX code for an x86 chip within two years of having to port and re-optimize for a new OS. Plus you have the fact that the x86 SIMD units are not compatible with each other, making SIMD optimizations take longer.
Apple won't budge from the PowerPC RISC platform until it has to. It is business, so why would Apple shoot the other foot when it has an option for catching up somewhat? You may not trust IBM, but Apple will probably have to rely upon IBM for now.
RiSC core architectures beat CISC cores in the dust, on speed, efficiency and power requirements. But the difference between two cores is not so much related to the core anymore, it is in the embedded peripherals (memory drivers, cache) and above all in the technology used.
Don't think that Intel is the only player, IBM is far from dead, and they are others too: TI, Motorola... and new architectures that become affordable as technologie becomes smaller. Wait for VLIW, cooperative multitasking, massive parallel processors...
Don't think that Intel is the only player, IBM is far from dead, and they are others too: TI, Motorola... and new architectures that become affordable as technologie becomes smaller. Wait for VLIW, cooperative multitasking, massive parallel processors...
Krevinek
Evil PPC Tweaker
As a programmer myself, I can say that recompiling for MMX would be a pain for many using Altivec. Apple would have to spend money and time porting their vector libraries to the different code set, re-optimizing for a usually thinner pipe, and then hope the companies aren't already spending time and money doing the port manually. Writing ASM code just to access a vector unit is tedious, and waiting for Apple to port vecLib to the different vector unit is just as tedious.
Please, don't make this programming stuff look easier than it is. Writing a GUI for a CLI app is easy, porting from one arch to another (even if the rest is done) may seem easy, but it is actually boring and tedious as hell, not to mention expensive depending on which end you are working on.
Please, don't make this programming stuff look easier than it is. Writing a GUI for a CLI app is easy, porting from one arch to another (even if the rest is done) may seem easy, but it is actually boring and tedious as hell, not to mention expensive depending on which end you are working on.
eric halfabee
You talking to me!
Same here, iClient, with maybe a little bit of iJunkie thrown in for seasoning.
