< So, while a 256 or 512 bit data bus is usefull (and < already in use in main frames - only called data path), < even a 128 Bit address bus is just nonsense. There is some truth to this. DEC went to 64bits as they knew 32bits were not enough for some applications. They were also trying to design for the next 20years to repeat the life of the VAX (32bits). Now, having a large address space isn't completely about filling it. It's about having addressing modes and arithmetic adaquate to deal with large data objects. In the 32bit model 4gb is the limit without thrashing. Those limits was exceeded by disk farms with tens of Gigabytes of databases and solving large mathmatical models. The current example would be doing windows graphics programming (1280x1024x16m) using only 16bit pointers, you do a lot of thrashing to get the the next 64k chunk and there will be a lot of them! That was the problem that spawned the VAX! Brief history lesson: The PDP11 was reachhing the point where a fully filled machine (4mb of ram) could not process in one lump the large problems like weather models or MRI data. It was NOT a speed issue it was 16 bits made for too small a data pointer. Also doing 32/64bit math required more steps (or FP hardware) as the registers were all 16bit. So to add two 24bit numbers on a PDP-11 required multiple steps and that was a speed impact. Going to 32bit registers made that a single instruction and paid back as speed without processors running at faster clocks. If it were speed the 11/74 would have been an ongoing event rather than the 4-6 than did get made. The birth of the VAX (and DGs Eclipse) grew So computer designers bumped the size of things to assure they could comfortably address the likely largest object they can resasonably expect to see for some time to come. Not so they could have 2^64bytes of ram but because they already have 2^40bytes of disk farm! All this was no surprize to computer designers. They foresaw it in the 60s knowing that larger memories and disks were needed to solve the problems that were growing. The datapath issue is strictly about speed as the VAX could have an 8bit data bus (in a trade for speed) like the 68008. While it has a big effect on speed it has no interaction with addressability. It's addressing the system problem of multiple devices competing for the common resource (MEMORY). By making the datapath wide you can grab several bytes at the same time and while the cpu is dealing with that the ram is available to run a IO cycle to a disk system or other IO. How big wide enough is, that is determined by how much silicon you have. Allison