Also were there any 6502 computers that were popular in Japan, or did the MSX and these Z80 machines rule?
Z80 machines ruled, japan oriented 6502 machines does NOT exist.
In Japan, NEC PC-8001, Sharp MZ-80 was born as 8 bit business machine with Z80.
Fujitsu FM-8 was also born with two 6809 CPU. But most hobbiest had selected Z80 machines.
(as a de-fact standard)
Also were there any 6502 computers that were popular in Japan, or did the MSX and these Z80 machines rule?
Z80 machines ruled, japan oriented 6502 machines does NOT exist.
That being true, I guess in Japan the 6502 is relegated to, with diverse minor modifications, being the engine for video games in the NES, SNES and PC Engine — the latter two of which are both definitely graphically impressive.
... in the NES, SNES and PC Engine — the latter two of which are both definitely graphically impressive.
The SNES used the 65C816 at a relatively high clock if I recall correctly. I would say it is a little faster than an R800. Has 24-bit addressing and corrects some limitations of the 6502. But the rest of the hardware, DMA, Video, Timers, RAM are really what makes it "impressive." Things like number of sprites and sizes, planes, commands, etc...
The 6502 based machines are: Family Computer (aka NES), PC-Engine, and Super FamiCom (aka SNES) with its 16bit version of 65C816.
Just the only game consoles has 6502, no japanese personal computer didn't.
In Japan, all computers are made by big company (very simillar to IBM), so these're very business like, so Z80 ruled.
And Japanese font (Kanji) needs at least 16x16 (or 12x12) dots to draw, so high resolution screen was needed.
Most machine have 640x200 dots screen, and it's very popular, but this was a big overhead for Z80.
(640x200/8 = 16KB, and three planes are needed for 8 colors screen => 48KB for video ram, and it located to main memory area with bank switch method.)
All of machines are Z80 based: PC-8001/8801/6001/6601 (NEC), MZ-80/700/1200/2000/2500 (Sharp), X1 (Sharp), M-5 (Sord), SMC-777 (Sony), MSX, and so on.
I had a quick check; the SNES's CPU isn't much more 16-bit than the Z80: it's an 8-bit data bus with some 16-bit internal storage and arithmetic. A 24-bit data bus, but in a 40-pin package with some multiplexing. Only 3.58Mhz; the PC Engine is likely faster with it's fully-8-bit improved 65C02 at double the SNES's clock rate. So now I understand why slow-down is so strongly associated with the Nintendo.
This is definitely a case of the support hardware massively outclassing the CPU. Probably a smart optimisation decision.
Anyway, apologies for the digression. All I was really trying to is observe that in Japan the 6502 was used, so it's not like nobody seriously evaluated it, but not anywhere that the objective was actual, serious computing. Just for games.
Not all Japanese computers used a Z80: There was the VIC-1001 (AKA Vic-20) designed and built in Japan by Commodore Japan that enjoyed a brief success before NEC took over the marked, also Fujitsu's FM7 (and followers) were 6809 based.
Moreover both NEC and Sharp had their own versions of the Z80 CPU (LH0080 & µPD780C), so it made sense that their machines were Z80 based.
... in the NES, SNES and PC Engine — the latter two of which are both definitely graphically impressive.
The SNES used the 65C816 at a relatively high clock if I recall correctly. I would say it is a little faster than an R800. Has 24-bit addressing and corrects some limitations of the 6502. But the rest of the hardware, DMA, Video, Timers, RAM are really what makes it "impressive." Things like number of sprites and sizes, planes, commands, etc...
It's a 16 bit processor. IMHO we are out of the comparison.
they used the 65816 because of the compatibility with 6502
Moreover both NEC and Sharp had their own versions of the Z80 CPU (LH0080 & µPD780C), so it made sense that their machines were Z80 based.
Or more likely the opposite. They had Z80 licensed because they were using it in their machines and their chip customer were also using it.
It's a 16 bit processor. IMHO we are out of the comparison.
Sure. It is called a 16-bit CPU because it has a 16bit accumulator (more versatile than HL) and 16-bit indexing and stack (no comments). It offers better 16bit processing than the Z80 instruction set (a little). What makes the difference in comparison with the 6502 is that is memory is segmented so you can have 64k stack, 64k code, 64k data being addressed simultaneously and independently and a total of 256 64k pages to use. Giving it more capability.
In terms of running 6502 code it is about as fast as a 6502. Compared to an MSX it can access memory in one cycle, as opposed to about 4, in average. So a 3.58MHz 65816 could be about 3.5x faster than the MSX a little more than that if taking full advantage of the 16bit accumulator. The problem is that it needs a memory operand almost every instruction due to the lack of registers. So it is more like 2.5ish speed advantage at the same clock speed. The point is that a 3.58MHz 6502 would have the same advantage. The 16bit here is not really helping that much in performance.
At the same clock 6502 are significantly faster than a Z80. At the same bus speed (many times limited by memory), Z80 is somewhat faster if the registers are well used. But most 6502 were limited to 1 or 2MHz at the time. Achieving 4MHz by middle 80's.
It's a 16 bit processor. IMHO we are out of the comparison.
they used the 65816 because of the compatibility with 6502
Well, for relaxed definitions of a "16-bit" processor, anyway, being a fairly meaningless term.
One processor has an 8-bit data bus, 16-bit address bus, an 8-bit primary accumulator, can perform 8- and 16-bit arithmetic, and does so with a 4-bit ALU. Obviously, it is 8-bit!
Another has a 16-bit data bus, 24-bit address bus, a suite of nearly-orthogonal 32-bit registers, and primarily performs 32-bit arithmetic using a 16-bit ALU. Obviously it is 16-bit! Unless you're talking about the version with the 8-bit data bus. That's 8 bit, obviously.
Yet another has an 8-bit data bus, 24-bit address bus, 8-bit primary accumulators, and performs 8- and 16-bit arithmetic, with a 16-bit ALU. It's 16 bit, of course!
A fourth that springs straight to mind has a 32-bit data bus, a 26-bit address bus, and an orthogonal set of 32-bit registers with a 32-bit ALU. This one is 32 bit!
So the rule is: classify as whatever the marketing department picks. After all, it's not the ALU, the primary register size, the data bus size, the usual data word size or any function of the address bus size.