Fewer transistors, hence less die space. Same reason DRAM is more dense (hence larger) than SRAM.
On 2016-May-28, at 7:12 PM, drlegendre . wrote: > So what's the reasoning behind using gate capacitance (or inductance) to > store the bit state? It would seem obvious that setting a bi-stable hi or > lo would be a much more reliable method of saving the state. > > Is it a matter of power consumption, or switching speed, or both? > > On Sat, May 28, 2016 at 8:49 PM, Brent Hilpert <hilp...@cs.ubc.ca> wrote: > >> On 2016-May-28, at 6:22 PM, drlegendre . wrote: >>> >>> Could someone also clarify what is meant by "gates" in this sense? Are we >>> talking about the gates (G) of a FET, as in Gate, Drain and Source - or >> are >>> we referring to the composite logic gates (NAND, etc.), built up of >>> multiple bipolar - or MOS - transistors? >> >> Yes, they're talking FET gates, the internal registers would operate under >> the same basic principle as DRAM does. >> >> Other early microprocs used dynamic registers, I forget which, perhaps >> others can list them. >> >> Far from the first time a processor had dynamic registers. >> I've been told that the IBM 709 used inductive (rather than capacitive) >> storage for the main registers.