The problem is that shoveling wide pointers has a cost; 64-bit builds have measurable costs even now.
It's not _always_ wise to solve tomorrow's problems today -- tomorrow often has its own technologies. On Thu, Jun 12, 2014 at 1:50 PM, Don Guinn <[email protected]> wrote: > True, except that is a real memory limit. Virtual memory could be much > greater than that. Like phone numbers. We need 10 digit dialing in the USA > but that is 10 trillion phone numbers. We don't have near that many phones > in the USA, but 10 digits makes management of numbers easier. Same for > virtual memory. > > > On Thu, Jun 12, 2014 at 2:20 PM, John Baker <[email protected]> wrote: > >> By this back of the J'envelope calculation we will probably never address >> more than 2x ^ 266 bytes of memory. I am assuming we can stuff a byte into >> a single baryon. Various estimates of the total number of baryons in the >> observable universe is around 10x ^ 80. >> >> _1 x: (2x ^ 265 + i. 5) % 10x ^ 80 >> 0.592855 1.18571 2.37142 4.74284 9.48569 >> >> 266 yields a ratio close to 1. >> >> If it turns out that dark matter can be used for computer storage this will >> be off. >> >> >> >> >> On Thu, Jun 12, 2014 at 12:48 PM, Don Guinn <[email protected]> wrote: >> >> > 2^48 memory essentially exhaustable? Remember when 16M was all we could >> > ever need. Before that 512K was bunches. >> > ---------------------------------------------------------------------- >> > For information about J forums see http://www.jsoftware.com/forums.htm >> > >> >> >> >> -- >> John D. Baker >> [email protected] >> ---------------------------------------------------------------------- >> For information about J forums see http://www.jsoftware.com/forums.htm >> > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
