Quote:
A lot of the allegedly painful features you describe are common in RISC designs of the era.
True. The SA-1 doesn't have these problems.
Quote:
What exactly do you mean? Can't the CPU just enter a RAM loop?
Yes, but there is "only" 128k WRAM available. Not fun to sacrifice half of it for your program loop. With the SA-1, both your CPU program and SA-1 program can execute right off the game pack ROM. The SA-1 memory controller will stall the SA-1 only when required.
Quote:
You mean like x86 instruction prefixes?
I'm not much of an x86 programmer, but I don't
think it's like that. But seriously, if you can get an x86 on an SNES cart, perhaps an Atom,
please get in touch. I'd love to have one of those.
Quote:
The CPUs in the Nintendo 64 and Sony's PS1, PS2, and PSP systems are based on the MIPS architecture. It has a similar branch delay slot.
The SA-1 has a two-stage pipeline and does not require this.
Quote:
Yeah, SNES homebrew is about to take off any minute now. Aaaaany minute now...
Hahah. That's what we need. Pan's SuperFX scrolling BBS trainer ROM. "Now the sprite letters bounce, change colors wildly
and rotozoom all at once!"
Quote:
The NES/FC gets a lot more attention because, well, fanboyism.
Yeah, it's incredible when you realize there's at least a hundred
active NES emulators, and four SNES ones. I don't know whether to be sad (lack of help on researching new things -- sans occasional miracles ala blargg's S-DSP), or happy (I can actually
do something useful for the scene -- with Nestopia and Nintendulator ... I couldn't come close. I'd off myself by the time I got the 34th input controller. "Pokkun Moguraa Tap-tap mat? ... WHY?!")
Ah, and I still haven't gotten around to emulating Felon's banana register yet :/
Quote:
But you've got RAM, and load/store instructions.
RAM and LINK are hardly an alternative to built-in stack and call intrinsics. Sure, you
can do it. But why not use a processor that already has it natively?
Quote:
If an assembler could generate machine code from the more readable syntax, then why couldn't a debugger do it in reverse?
Well, you can split things up easily, eg:
to r13
loop
with r15
Hard to group to r13 + with r15 into move r13,r15; but I guess the same is true when you're writing that assembler code in the first place. Fair enough, I rescind that argument.
Quote:
I'm not saying the SuperFX is the best thing ever. I just find these kinds of co-processors interesting. Whether or not anything useful will come out of it is irrelevant to me.
Yes, it's no doubt a neat chip. And it can obviously do very cool things. The whole point of my argument was that the SA-1 was better in virtually every way. That's all. If you don't want to use the best, why not use the OBC1 or DSP-2? :P