Oh, I love a good frenzy!
There is also this thread on the subject, but my own analysis follows.
The recent Macintel debate has gone perfectly mental, and I have to say I'm not surprised. If there's one thing Steve loves to do, it's scare the willies out of people and get them talking... advertising is expensive these days, even if apple's white background saves on ink.
Some perspective I think is in order... there are a number of mindless zealots scampering around venting their objections and fears in the disguise of valid opinion, and there needs to be a bit of a slap about to calm everything down.
Mind you, most people I think are happy with this and can see where this is going, but if you can't see past your nose, I thought I might spell it out.
1. The BIOS thing.
NOBODY KNOWS EXACTLY WHAT THEY'RE GOING TO DO!!! But being Apple, you can kind of add 2 and 2.
For me, even though I know that under the hood, all computers are bit pushers, the thing that sets the tone for a session sitting at a mac is the experience. The experience is programmable, and it wouldn't matter what Apple choses to sit at 0x00000, it will pretty much be configured to set the tone for a mac experience. This means that it will chime, it will show a logo, it will detect the monitor and it's available scan settings, it will REMEMBER what resolution the monitor was last set at, and it will shower you with the love, affection and self affirmation that open firmware has faithfully serviced us with.
The first commercial macintel will probably not have either open firmware, or a standard plastic fantastic golden dragon bios... but something new from intel.
Cmd-Option-P-R will still "zap the p-ram".
2. It will just be a standard Intel PC
They will have a lot in common with standard PCs, but different types of Macs are designed for different types of applications, and each application has certain requirements which dictate what the market considers to be an edge. Apple kind of has the crown for the best video editing platform, but once upon a time SGI held that crown.... in fact in the high end they still pretty much do. The reason for this is that SGI have always had a superior architecture when it comes to moving data in huge wads to various subsystems, particularly texture mapping and other bitmap buffers.
There are a number of architectures present in current PCs AND Macs which hold them back in this respect. PCI is one of them. Blah, blah, blah, PCI Express, blah, PCI-X blah.... but in the long run it's just not going to cut it, because all these technologies are intended to do is hold out until the next generation interfaces arrive.
These interfaces will not be shared bus technologies, but based on a switched interconnect fabric... and will make a whole range of tasks damn near affordable, probably quite simple, and very, very quick.
The new generation Macs will over time incorporate these technologies in such a way as to make sense to the Apple user-base. Apple users (particularly in the high end) need more than gamers. Video is a big, huge, thing, (literally) and the next generation machines will need to accomodate a whole new subset of accelerated and distributed APIs for media production.
Remember, SGI have been making a linux machine called the Altix for some time now. It's based on Intel, but it's anything but a PC. NASA has a few... they're pretty.
3. RISC is better.
Not when it's more expensive/slower/hotter/non-existant etc.
There are ideals, and there is reality. I can totally relate to the orgasmic performance of a dual 970 machine (and the religious zealotism that goes with that - you have to justify the purchase somehow), but if it were a choice between that and a quad IA64, I know where I'd rather be. The first machines we see might even be PowerBooks, and they will probably be Pentium-Ms.
Anyway... when was the last time you fired up your Athlon, IA64 or P4 and thought, y'know, this just doesn't feel as smooth as my RISC based machines? For me it was 1995.
4. Transition worries.
I really don't think that Apple would get this wrong a second time. In Steve's keynote, he was pretty acidic about the transition to PPC, and I have to totally agree with that sentiment. I still have my Quadra - just in case.
This transition has been pretty much planned from day one as an option, and the writing would have been on the wall the day Steve came back from his 40 days in the desert. One could argue that the writing was on the wall way before then. Anyone remember the Rhapsody x86 version? This is not new.
I think if you look at what's been going on behind the scenes, Apple have probably wanted to do this for a while, but their hardware engineers have probably a lot more invested in PPC than warranted the change, and to change BOTH the OS and the hardware at the same time would have had him impaled.
Rosetta is the key, and probably the last thing they needed in order to make the transition possible. People need to know that they can have at least some kind of smooth transition.... there will probably be some Altivec problems, but hey, if you're relying that heavily on Altivec based optimisations, chances are you have a pretty current app, and will have a universal binary ready before you can sing "a spoon full of sugar helps the medicine go down".
Benchmarks aren't really going to give an indication at the moment of what's going on with these machines. Pretty much all the software you're going to run is eventually going to be universal binaries, and even if Rosetta has to be called for PPC binaries, most of the native API stuff won't actually be emulated anyway.
There is also this thread on the subject, but my own analysis follows.
The recent Macintel debate has gone perfectly mental, and I have to say I'm not surprised. If there's one thing Steve loves to do, it's scare the willies out of people and get them talking... advertising is expensive these days, even if apple's white background saves on ink.
Some perspective I think is in order... there are a number of mindless zealots scampering around venting their objections and fears in the disguise of valid opinion, and there needs to be a bit of a slap about to calm everything down.
Mind you, most people I think are happy with this and can see where this is going, but if you can't see past your nose, I thought I might spell it out.
1. The BIOS thing.
NOBODY KNOWS EXACTLY WHAT THEY'RE GOING TO DO!!! But being Apple, you can kind of add 2 and 2.
For me, even though I know that under the hood, all computers are bit pushers, the thing that sets the tone for a session sitting at a mac is the experience. The experience is programmable, and it wouldn't matter what Apple choses to sit at 0x00000, it will pretty much be configured to set the tone for a mac experience. This means that it will chime, it will show a logo, it will detect the monitor and it's available scan settings, it will REMEMBER what resolution the monitor was last set at, and it will shower you with the love, affection and self affirmation that open firmware has faithfully serviced us with.
The first commercial macintel will probably not have either open firmware, or a standard plastic fantastic golden dragon bios... but something new from intel.
Cmd-Option-P-R will still "zap the p-ram".
2. It will just be a standard Intel PC
They will have a lot in common with standard PCs, but different types of Macs are designed for different types of applications, and each application has certain requirements which dictate what the market considers to be an edge. Apple kind of has the crown for the best video editing platform, but once upon a time SGI held that crown.... in fact in the high end they still pretty much do. The reason for this is that SGI have always had a superior architecture when it comes to moving data in huge wads to various subsystems, particularly texture mapping and other bitmap buffers.
There are a number of architectures present in current PCs AND Macs which hold them back in this respect. PCI is one of them. Blah, blah, blah, PCI Express, blah, PCI-X blah.... but in the long run it's just not going to cut it, because all these technologies are intended to do is hold out until the next generation interfaces arrive.
These interfaces will not be shared bus technologies, but based on a switched interconnect fabric... and will make a whole range of tasks damn near affordable, probably quite simple, and very, very quick.
The new generation Macs will over time incorporate these technologies in such a way as to make sense to the Apple user-base. Apple users (particularly in the high end) need more than gamers. Video is a big, huge, thing, (literally) and the next generation machines will need to accomodate a whole new subset of accelerated and distributed APIs for media production.
Remember, SGI have been making a linux machine called the Altix for some time now. It's based on Intel, but it's anything but a PC. NASA has a few... they're pretty.
3. RISC is better.
Not when it's more expensive/slower/hotter/non-existant etc.
There are ideals, and there is reality. I can totally relate to the orgasmic performance of a dual 970 machine (and the religious zealotism that goes with that - you have to justify the purchase somehow), but if it were a choice between that and a quad IA64, I know where I'd rather be. The first machines we see might even be PowerBooks, and they will probably be Pentium-Ms.
Anyway... when was the last time you fired up your Athlon, IA64 or P4 and thought, y'know, this just doesn't feel as smooth as my RISC based machines? For me it was 1995.
4. Transition worries.
I really don't think that Apple would get this wrong a second time. In Steve's keynote, he was pretty acidic about the transition to PPC, and I have to totally agree with that sentiment. I still have my Quadra - just in case.
This transition has been pretty much planned from day one as an option, and the writing would have been on the wall the day Steve came back from his 40 days in the desert. One could argue that the writing was on the wall way before then. Anyone remember the Rhapsody x86 version? This is not new.
I think if you look at what's been going on behind the scenes, Apple have probably wanted to do this for a while, but their hardware engineers have probably a lot more invested in PPC than warranted the change, and to change BOTH the OS and the hardware at the same time would have had him impaled.
Rosetta is the key, and probably the last thing they needed in order to make the transition possible. People need to know that they can have at least some kind of smooth transition.... there will probably be some Altivec problems, but hey, if you're relying that heavily on Altivec based optimisations, chances are you have a pretty current app, and will have a universal binary ready before you can sing "a spoon full of sugar helps the medicine go down".
Benchmarks aren't really going to give an indication at the moment of what's going on with these machines. Pretty much all the software you're going to run is eventually going to be universal binaries, and even if Rosetta has to be called for PPC binaries, most of the native API stuff won't actually be emulated anyway.