Z80: Difference between revisions

Computers using the Z80 tended to run at a relatively high clock speed by for the time, compared to certain other processors such as the MOS 6502 and Motorola 6809.  Doing so was required to achieve a decent level of performance, as on average, the Z80 required a larger number of clock cycles per instruction executed.  This high clock speed would typically eliminate cycle stealing as a means of sharing video memory between the display generation hardware and the CPU.  If a display is to be generated whilst video memory is being accessed, this leaves a choice of comparatively expensive VRAM with a dedicated read port for display generation, or a memory contention scheme where by the CPU will is halted when the display hardware requires access to video memory.  The ZX Spectrum uses a memory contention scheme, but with a slight quirk — the /MREQ line that signifies whether the Z80 is performing memory access is not decoded by the original Sinclair models, meaning that the CPU can be slowed down as a result of values placed onto its address bus when not performing a memory access.  This issue was later fixed by Amstrad for the Spectrum +3 and +2A/+2B, but by this time, a certain amount of software had grown to depend on the behaviour of the original Sinclair models.