In my course on embedded systems, it is explained that memory inputs can be separated from I/O inputs using a "mode bit" for the address decoder. The most obvious advantage of this is that the amount of address lines need to be decoded can be reduced if you have a separate addressing system for your memory and your I/O. Are there any additional benefits? Does having these logically separated have other benefits such as being able to use different clock speeds or transfer logic?
Electronic – Benefit of (logically/virtually) separate I/O and memory bus
computer-architectureembeddedmicrocontroller
Related Topic
- Virtual Memory, Cache, and TLB’s
- ISA efficiency code compaction and memory traffic
- Memory addressing in eMMCs
- The difference between a virtually tagged and a physically tagged memory system
- Banked and non banked memory
- Electronic – Memory architecture and instruction storage – Naimi: AVR Microcontroller
- Electronic – Why are logically related bit fields in MCU registers often in separate locations
Best Answer
My interpretation of what you are describing is to simply sub-divide the address space into two halves - one half for IO and one half for memory. You're not creating two separate busses there, but only one that is used for two logically different tasks.
While that can make decoding of the address bus simpler, it doesn't improve performance at all.
Many embedded microcontrollers use an architecture known as Modified Harvard Architecture. In this you do have a number of different physical busses, each with its own purpose - one for RAM, one for ROM, maybe one for IO. These have the big advantage that it is possible to access different things at the same time which you can't with a single sub-divided bus. It does make programming more complex though.