Great addition to the info provided. Unless one actually did some low-level programming with a CPU, this stuff gets hard to understand. I think the basic crux is that the x86-64 has a 'mode' where it allows for a much larger addressing space than 4GB. In this mode, it uses the same 32 bit instruction set of the x86 so that it is fully compatible with all existing x86 apps that existed at that time. The Intel Itanium was a totally new CPU architecture that was 64-bit and accessed lots of memory, but its native instruction set was not compatible with the 32-bit x86 CPU and so people had to re-write/recompile their code, which was more effort than people wanted to take. When Intel realized it was not being adopted, it incorporated various levels of x86 'emulation' to allow people to run older apps. But too little, too late. Thus the x86-64, while called a 64-bit CPU, is really more of a modified x86 core that allows a much larger addressing space.. This was what happened when it first appeared. Since then CPU designs have evolved significantly and I have not tracked their capabilities.

yes.. the size of the address register affects how much memory can be accessed. Not all registers inside a CPU are the same size. (But most are) When we say a CPU is '8 bit', that used to reference the basic size of the OpCodes. However, as CPUs became more and more complex, there was no single way to define what '64-bit' means. It became more of a marketing game.

When switching to 32 bit instructions, all drivers for hardware devices, such as video cards, had to be re-written. This takes much time. I remember when using Linux in these days, the list of hardware peripherals that would work was very, very limited. It was generally a nightmare taking a random PC and installing Linux on it, unless you knew how to recompile drivers. As Microsoft had to support thousands of different hardware combinations, it took them a while to get a stable OS out.

But it is not correct. The "64-bit" Athlon did not have a 64-bit instruction set, nor did its address register have 64 bits. It could more 64 bits of data at once.

I don't understand what you mean by ls\st. Fetching multiple pieces of data to chain into a longer command is beyond the scope of what I wanted to discuss in this video.

if your PC is less than 5 years old, you've like installed the 32-bit version of Windows.. Install the 64-bit version and you will likely have access to 8GB RAM.

There are many groups of registers inside a CPU. Saying the Athlon is a 64-bit CPU comes from their marketing department. It did not have a 64-bit instruction set, nor did it have a 64 bit address register. It did have an address register with more than 32 bits.. about 38 bits, I think; which gave it the ability to access more RAM than the INTEL designs (until Intel copied AMD in this x64 design)

There are many registers inside a CPU. The registers that move data on the data bus are 64 bits so that 64 bits can be moved at once. However, there is an address bus register and this can be any size. In the 8086 it was 20 bits and so 2^20 was 1MB and thus the memory limit. In the 32 bit pentiums, it was 32 and so 2^32 was 4GB and that limit. The "64 bit" AMD enhancement in some Athlons increased the size of the address register to about 36 bits I think (not 64), and so this beat Intel at its own game and allowed more than 4GB of memory to be used. AMD started outselling INTEL in server CPUs until Intel copied AMDs enhancement.

Not quite that simple.. if you have 4 address lines.. then you can 'address' 2^4 or 16 'things'.. the size of the databus determines how many bits you can transfer.. so with 4 address lines and 8 data lines I can move 16 x 8 bit chunks of data or 16 Bytes of data. If I have 20 address lines, I can move 2^20 or 1Mega chunks of data.. If the data moved as on 8 data lines, I'm moving 1 byte of memory, so 20 address lines allow me to move and access (or address) 1MByte of memory.. I'm simplifying it a bit here.. .. basically the number of address lines (which are controlled by the address register) determines how much memory your CPU can use.