@sketch3d.de said:
...Core i7-3740QM
if he wants to build a notebook... yes. 
@sketch3d.de said:
This is just not true, most apps available are still 32-bit because they simply do not benefit from accessing more than 2GB RAM
interesting... and this is the reason why sketchup is now large adress aware... right?!?
yes, my microsoft word never needs more than 2GB...
sorry, but this is the most stupid statement in this whole thread... yes, there will always be apps that don't benefit from more processing power or more RAM because they don't need it. And even sketchup will not benefit from more speed if you're modeling only a cube! But i can tell you, that working on a 200MB+ model in sketchup is not really funny... (i'm not speaking about high res textures, but geometry! textures are already all the lowest low res dummys) 2-3 min auto saving time is really a joke for 200MB... the only thing you can do is to disable the autosave.
And you will wish for 64bit if you have to split the geometry into parts to be able to export it to the renderer or if you have to restart sketchup after every single export, because sketchup doesn't fully clear the memory after export and you run out of memory on every second or third export... (it's better now than before but still there)
And it IS just true! because this is the way processing power will evolve in the next years... the future is parallel. Single core performance improvements got more and more limited in the last years (sandy to ivy was ~5% and Haswell will get ~10% - great!) and if you want more speed you have to use the other 7, 11, 15, 23 or 31 idling threads of your system!
And if it may be hard to implement multicore support for single modeling operations, i really don't get it why it isn't already used at least for rendering animations, background saving or running ruby scripts in a second process.
If i render an animation i have to split the sequence in 5 or 6 parts and start the rendering on 6 different sketchup instances. Is it really that complicated to do this automatically from one instance?!?