Ok, I'll get this out of the way quickly before it gets swamped by other comments:
Snow Leopard will *not* magically make your programs run faster just because of all the new multithreaded performance tweaks and CUDA integration etc. I know to some of you this might sound like a stupid and obvious comment but the number of people that seem to think that it'll wave a magic wand is surprisingly high. 10.6 should make writing parallel code a darned sight easier in theory, true. It should be a fair bit snappier since everything under the hood is meant to be rewritten in parallel (significantly a parallelised Finder). However, existing code will not run significantly faster, if at all. Single threaded programs will run just as fast. There's a chance that existing parallel codes will be able to use better system libraries to run faster but, honestly, the gains will be minimal. This isn't to say that down the line 10.6 will prove to be far faster than 10.5 due to multithreading, but its going to take time for the newer technologies to be adopted.
Apple is hyping up the multithreading of 10.6 which can lead to such assumptions as I've mentioned above. Why are they doing this? Well, 10.6 is largely an 'under the hood' improvement. As others have said, there's not a lot of obvious stuff that users will see in which they can exact improvements. Snow Leopard comes with all kinds of fantastic technical improvements and performance benefits but most of those will not be seen by the user. So they're caching in on the current chitter-chatter of multiprocessing.
I certainly can't wait for 10.6 to come out and I'm very excited in regards to programming with CUDA etc. The multithreaded Finder would be enough on its own for me to upgrade. I just hope people can be realistic with their expectations of the new OS.
Facebook
Twitter
Instagram