Allow me to put it in layman terms.
Think of the memory bus as a street. The 64 bit is a 4 Lane highway while the 128 bit is an 8 lane highway. Logically, the 8 lane highway can handle twice the number of vehicle compared to the 4 lane highway. Translate this to memory bus will means that theorically, a 128bit memory bus can handle twice the amount of data compared to a 64bit memory bus.
As for how much memory is enough, let's take it this way;
For example, I have a carpark that have 256 parking slot and a 128 lane road that can only handle about 256 vehicle traveling in and out of the carpark at any one time. So if I were to increase the number of parking slot available to 512, will it increase the rate at which the the car will go in and out of the carpark via the 128 lane of road? Logically , theorically and practically, the answer is no.
Now let's put it in into your question: "So with the 256MB vs 512MB in subject, any proof to support your claim?"
Your graphic card have a video memory of 256MB and a 128bit memory bus that can only handle about 256MB of data transferring from the GPU to the VRAM at any one time. So if I were to increase the amount of video memory available to 512MB, will it increase the rate at which the data will be transfered in and out of the GPU and the VRAM via the 128bit memory bus? Logically , theorically and practically, the answer is once again no.
So to conclude, there's actually not much difference in terms of performance when comparing a 512MB VRAM and a 256MB VRAM variant of 8600M GT.
Hope my explaination have persuaded you in letting you know the rough idea what are all this about. If you still need concrete proof, feel free to post here. I'll try my best to answer your questions.
OT: Recently just realized my Acer's 8600M GT is manufactured by MSI.