Comparing the Z80 and 6502 to Their Relatives

A few thoughts:

> the complexity of some instructions (e.g. DJNZ)

Well, of course the idea of DJNZ was to implement a very common pattern (decrement a register and jump (normally backwards) if the result was not zero) - this tended to simplify code rather than make it more complex.

> However I couldn't keep all instructions and timings in my head.

I was never really interested in the timings, but I did get to the stage (not by conscious memorisation) of being able to assemble and disassemble Z80 code in my head, with some accuracy.

> I preferred, however, the fact that I/O was memory mapped on the 6502.

Many (most?) Z80 systems used memory mapped I/O. It's down to the hardware designer.

> Speedwise the 1MHz 6502 and 4MHz Z80 were on par.

This is a bit of an exaggeration, the 6502 was efficient but not that efficient. While generally understood that the Z80 took 2x-4x ticks to execute instructions as the 6502, in the real world its larger register set meant properly-written Z80 code could avoid expensive, slow round trips to memory.

Outside of artificial benchmarks real world performance shows that the 6502 is roughly 2x as efficient per clock cycle as the Z80[0], i.e. a 1 MHz 6502 is approximately equivalent to a 2 Mhz Z80.

This is reflected in the computers of the day, i.e. TRS-80s were not being blown out of the water by Commodore PETs.

[0] https://github.com/soegaard/minipascal/blob/master/minipasca...

> I preferred, however, the fact that I/O was memory mapped on the 6502.

The Z80 could do memory mapped IO as well of course (used at least in some arcade machines), but why waste valuable address space when there's an entire 64 KB of extra address space reserved for IO ;)

I'm a 6502 guy myself, but the Z80 instruction set is also easily holdable in one's memory.

As others said, a 4MHz Z80 is clearly capable of outperforming a 1MHz 6502 as is evidenced by the many ZX Spectrum demos that show off 3D/plotting effects.

But 6502 has 256 registers! Full ZP of them.

In practice it just meant the “zero page” could be anywhere, not the 256 bytes starting at 0x00 (like the 6502 zero page). The opcodes that operate on zero page are shorter and thus faster.

"Linker"?

In theory, the motivation for position independent code was to support the development and use of software libraries that could be "plugged in" to an application.

In practice, RAM was often limited to 16 KB; software reuse that I'm familiar with on a 6809 platform was at the source-code level and optimized by the programmer.

I remember editing and assembling, but not compiling or linking.

That said, I believe Motorola wrote some floating-point libraries.

I was a kid on a Tandy Color Computer, and the $49.95 EDTASM cartridge was a huge investment for our family. So my point of view could be way off... but the simplicity of the Color Computer with the design of the 6809 made programming delightful. (20 years later, my enjoyment in programming the Palm Pilot felt like that... although by then I could use C as a fancy macro assembler.)

Larger and later systems could use OS-9, which reasonably resembled UNIX and maybe supported a C compiler.

On older hardware such as this it could e.g. let you write a multitasking environment that supported shared libraries without use of an MMU (though you'd hit memory constraints pretty quickly on a Z80-era cpu!).

I'm not familiar with the instruction sets of the 6809 but I could also see more compact opcodes, e.g. a JMP with a relative offset can be encoded smaller than JMP with an absolute address.

In modern terms PIC is used for ASLR and is therefore a security requirement. Some arches (I'm most familiar with arm64) are entirely designed around PIC and you need extra hoops to do anything in absolute terms.

In general, yes, the linker (if there is one) would have less to do.

Position independent code (PIC) on the 6809 is pretty easy [1], but it does increase the code side a bit, but the resulting code can be placed anywhere in memory with no changes and still work. As mentioned, Motorola intended to sell a ROM with IEEE-754 floating point routines for the 6809 (as the MC6839) that was PIC. As far as I could tell, they never did sell the ROM, but they did provide it (with source) for anyone to use.

[1] Relative branches instead of absolute jumps, using the index registers to address memory, as well as addressing relative to the program counter. You can still do jump tables, but instead of a list of addresses, they're just a list of relative jump instructions. That type of thing.

Technically it wasn’t the linker that’s simplified, it’s the loader.

Modern systems don’t worry about PIC code, they have virtual memory so everyone sees memory the same. The virtual memory system manages the relocation automatically.

OS/9 relies pretty much entirely on PIC code, that made the loader and multi-tasking easy.

Original MacOS also relied on PIC. For similar reasons, and it’s partly why code segments were limited to 32k.

Then you have things like the original 8086. As long as you stick with the “tiny”/“small” memory models, everything was relative to the segment registers, so code and data could be moved easily.

In contrast you had systems like the Apple IIGS. The 65816 does not support PIC well, so code segments carry a relocation table that allows the segment loader to relocate code during transfer from disk. The creation of segment and relocation table is the job of the linker.

I also found the 6502 far more enjoyable. It feels like a refined, minimal design, since the Z-80 does bolt things on to the 8080 (which I've also coded a lot for, also not as enjoyable). There tends to be one straight-forward way to code things, that's also the most efficient, barring changing the algorithm or twisting the design. Instructions tend to use one cycle per memory access, and memory accesses are mostly as expected, e.g. load A from 16-bit address is four cycles: the LDA opcode, two-byte address, and byte loaded. The Z-80 suffers in this regard because like the x86, it uses 1-2 bytes for the bolted-on instructions and modes, so some seemingly similar instructions can use a different number of cycles (e.g. ld hl,nn and ld ix,nn take different times, even though they both load a 16-bit value into a register).

> Programming 6502 seems simpler than learning lets say JS framework or to learn just about anything modern.

Vast, complex, changing APIs make programming unfun. Retro machines were fun because the things to learn were concise; the rest was up to your thinking.

Was hoping this would be one of the the chips covered.