Bits of what?

computers use binary. so what ever you are doing will be converted to a set of 0's and 1's. so a bit is a 0 or 1 and a 32bit CPU can load a number with 32 0's or 1's into a register in the CPU. likewise a 64bit CPU can load a number with 64 0's or 1's into a register.

so a 32bit number in binary can be as big as 4294967295 in decimal. 4G essentially and hence the limit of 4GB RAM on 32bit systems.

i'm not sure about the 64bit password, i think that is referring to encryption, which i don't know anything about.

someone else will probably explain it better, or post another question.

She maybe asking about encryption for passwords or keys. Like 40 bit or 128 bit.
http://compnetworking.about.com/od/networksecurityprivacy/l/aa011303a.htm

It also determines the amound of data that can be transferred. A 64-bit datapath is double that of a 32-bit one, obviously. It's like a computer system that uses dual-channel RAM. It utilizies a 128-bit data path compared to a PC that is single-channel with a 64-bit path.

In earlier times, the bit value was what made one video game system better than another. As technology increased, however, that number came to mean nothing.

So it's essentially about data throughput. It all sounds to me to be something akin to bandwidth. Cheers Crazijoe for the link, it makes sense now on a basic level.

ebackhus said:

It also determines the amound of data that can be transferred. A 64-bit datapath is double that of a 32-bit one, obviously. It's like a computer system that uses dual-channel RAM. It utilizies a 128-bit data path compared to a PC that is single-channel with a 64-bit path.

Click to expand...

Ok, so I have dual channel RAM which is 2 x 64bit = 128bit data path sending info to my 32bit processor with hyperthreading (does that even count?). How does the processor cope, surely that's a bottleneck there or does the difference in frequencies sort this out?

You just said it, the CPU is operating at a much higher frequency, and the RAM is still slower than the theoretical given speeds because of the latencies.

BTW, the hyperthreading just handles multithreaded instructions, the data going to other components still has to go through the FSB and the Northbridge.