With audio, it is seriously child's play. Compression is about as CPU intensive as you can get with audio outside of voice recognition, and it just isn't there.

Video is mostly offloaded on to GPU's these days. The MBP GPU is definitely a big step down from what you can get on a desktop, but you can always get a dedicated bit of hardware that you attach to your MBP to get that job done. The real killer on video has traditionally been I/O. With SSD's and Thunderbolt, you have some alternatives.

> I still do not think that Intel can cram in the same level of performance into a 35W chip that use to go into a 90W platform just 2-3 years ago.

You might want to check out Moore's Law.

Not true. Try running a really complicated Logic set with a lot of virtual instruments on a low-end machine and watch it grind to a halt. In the last year we've seen new virtual instruments that bring analog emulation to a new level but they can eat as much as half of my Macbook Air's CPU playing a single note. Extrapolate this to a professional mix with as many as a hundred tracks and you can see why some people still need a real computer.

Assuming that we can keep cranking the moore's law handle in one way or another people will always find a way to use that extra power even if the rate of return is somewhat logarithmic there is always a wow factor in something that is better even if only by a marginal amount.

The question I guess becomes how much of this can we move to the cloud? If the math for the virtual instruments in this example could be run inside a data centre and then streamed to the computer , perhaps some of the need for high end local hardware is alleviated.

Of course this will be cyclical , we move to a world where software is something that is paid for by the month and after a while of paying lots of different monthly fees for different apps somebody will start offering people a "local cloud" that will look like a great deal because you only pay once.

Music is a particularly difficult case because latency is an absolute killer. If your response time is over about 10-20 milliseconds people will start to complain. You could farm out offline rendering I suppose but so far it's a lot easier to build a big beefy workstation and freeze tracks to audio if necessary on the local disk.

By a long shot.

But I'll stand corrected. I'm obviously just playing with inferior tools.

Barking up the wrong tree, mate. Obviously he won't try it, because he doesn't know what the hell he's talking about - and certainly never heard of Logic.

Highly polyphonic physical modeling synthesis is more CPU intensive than compression... then you start adding on many instances of complicated reverb effects, some effects that require fft/ifft passes, autotune, &on and on, all with realtime constraints (not to mention all of the copying of buffers and context-swapping that goes on with the way these systems are designed)... and you get a high-cpu monster pretty quickly.

The memory bandwidth, # of cores, fixed point and floating-point computational characteristics of MBP's are not that different from their desktop brethren (smaller, but they tend to be proportional). Bottom line is chewing half the power still puts you ahead after a couple of iterations of Moore's Law.

If your problem is bigger than that, or requires specialized hardware, it tends to make sense to support external hardware and/or distributed computation, in which case the MBP's limitations tend to matter far less.

No offense, but that's not even wrong.

Don't do much audio work, do you? Between softsynths and DSP effects on 32+ tracks, professional-level audio workstations will eat all of the processing power available on pretty much any modern CPU and ask for more.