The rumour that the industry is moving towards parallelisation started with the introduction of HyperThreading in the P4. That's almost 10 years ago and yet we're still not at a point where more cores actually make more sense than more clock speed for the majority of applications. One can just hope, that the software industry will finally catch up with the hardware and write their apps with true multithreading support.
Oh I don't disagree. Parallelism isn't easy. That's why I think what Apple is doing with frameworks like Grand Central are so essential. What used to be an incredibly specialized skill set has just been somewhat normalized.
Sure, it's not as good as someone who knows how to design core logic from the beginning with parallelism in mind - but writing code in high level languages is far more inefficient than writing in assembler too and that hasn't held us back (OK,
some might argue
)
As per buy for today or for tomorrow, personally I'd rather buy for today and get the most out of the machine today, rather than waiting for the apps I use to finally get true multithreaded support and waste my time waiting for a machine that is slower than a higher clocked machine with less cores.
I guess it depends on just how CPU bound the applications you use really are. If you are someone who is that dependent on CPU speed, buying fewer high speed cores today and replacing a machine tomorrow isn't going to be an issue. It will probably even be a tax write off
For me, since I'm just a pro-sumer hobbyist I have to plan for the long haul. Multiple cores are the good, cost-effective long term bet. That and having 8 memory slots with the two socket configurations vs. 4 with the single socket configuration.
Personally, over 2/3's of what I see people blame on CPU limitations are either a lack of sufficient RAM, bottlenecked GPU (Apple with their miserly and expensive options doesn't help here!), dog slow hard drive subsystem or a combination of all three!
I'm not really sure that the generation of software we've got today really is the last generation of single threaded applications. Again, the demand for applications that really utilise all available threads of the machine is about a decade old (possibly even more as MP systems have been available long before consumer grade processors with multiple cores/threads).
I think with what you are seeing Apple and nVidia do with frameworks, much of this is going to start to be abstracted out at the framework level. Again, not ideal - but a far sight better than where we are today.
The MHz barrier is no longer theoretical and is now real. Multiple cores are the only remaining frontier for continued ramping of performance.
Final Cut is rumoured to be truly multi threaded with the next release, but who knows?
I'll put it this way: With as long as it's taken for the current upgrade, and with the foundational stuff Apple added starting back in 10.5, if it's not leveraging all of their new frameworks it will be pretty embarrassing for them.
Not that it's stopped them from doing bone-headed things in the past
But here's to hoping. I think all the hand-wringing about Apple being "distracted" or "abandoning" the Pro market due to iOS is dramatically overblown. If you look at everything that has happened with 10.5 and 10.6 "under the covers" it's clear that Apple has some pretty major and ambitious plans for Mac OSX. And they have historically and I see no reason for them to stop showcasing core technologies in their own Apple branded applications to spur other developers along.
If anyone thinks they are moving away from showcasing their tech in their own software, just look at Garage Band on the iPad 2. I hope it would be logical to assume it's taking longer with the Final Cut suite due to it being slightly more complicated than Garage Band on the iPad, but sometimes I have to wonder
BTW - this is where developer-focused practices of using cross platform tools to "write once" really, really hurt us all. It's also something I was glad to see Apple take head on with Adobe and others in iOS. The last three, four generations of Photoshop have SUCKED on the Mac due to the homogenization of the application by Adobe's cross platform tools. Those tools and practices benefited the Adobe developers,
not you and I as the end user. I think that's why you see Apple writing software like Final Cut and Aperture - to try to spur others to take advantage of all the cool stuff they have. I always shake my head when people spout off about Apple "needing" competition to keep them on their toes when in reality it's quite the opposite.
