Geano Posted June 25, 2020 Share Posted June 25, 2020 7 minutes ago, efaardvark said: We have not gone to 128 bits because there is a cost for doing so and we don't really need it. There's a whole bunch of reasons for the cost. At the hardware level we use parallel buses. That means that every bit gets it's own data line. If two wires touch then the computer stops working. If two wires even just get too close then there's crosstalk and your computer becomes unreliable. Routing all those wires across the motherboard to connect the CPU, memory, PCI buses, etc. becomes a problem the wider the bus is. An 8-bit bus is easy to design. 16 bits is also pretty easy. 32 bits starts to get troublesome. 64 bits is downright tricky. This is also a problem for chip design internally, for similar reasons. Ok, so why even go to 64 bits then? Well, with 32 bits you can only count to 4 billion. If you have a file that you want to reference a particular byte of data in then your file can only be 4 gigabytes in size. If you have memory addresses that you want to reference then you can only have 4GBytes of memory and still be able to reference each byte individually. Lots of people want to use files more than 4GB in size or have more than 4GB of memory in their computers. Yes, there's tricks like using two 32-bit registers to hold a single 64-bit number, but now things like your math libraries and other code at the software level get complicated and slow. Worst case they have to do twice as much work and run half as fast. It is worth it to go to 64 bit buses and let the hardware do most of the work, even though it makes the hardware a bit harder to design and more expensive to build. Going to 128 bits would make the hardware extremely hard to design and build, as well as expensive. At the same time going from 64 bits to 128 bits on address buses and integers doesn't buy you nearly the gains that going from 32 to 64 bits did. With 64 bits you can reference data in files that are up to 18,446,744,073,709,551,616 bytes in size, or individually address over 18 thousand terabytes of memory. Very few people have files that big or computers with that much memory. (yet.) Maybe we'll get there some day, but for now it isn't worth the costs. I see that makes sense. This is stuff I have been studying pretty heavily as I am studying computer science. So any info certainly helps me. Makes sense as well that there is no true tangible gains for the expense even in the enterprise data center type environments where clustered computing works just as well if not better for the price based on what your saying. So asuming the cost wasn't a gactor what possible gains could wee see from 128-bit hardware if say the spec was worked out? Link to comment Share on other sites More sharing options...
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!Register a new account
Already have an account? Sign in here.Sign In Now