01-09-2014, 11:26 PM
Man does this look promising. But, there are few things that bugged me a little bit. First comes the logic operations. Having 18 logic op is overkill, kinda useless, and it opposes the idea behind risc. You can have the basic nand, xnor and be good to go. Altought it would take more than a cycle to do other operations, a cpu will use them 1,2% (actual fact from risc research, mostly used operations are OR and AND).
By removing those Operations from the IS, you can have much more space for other stuff (like interrupts, routine calls, rotation, pair registers..)!
Secondly, to me, having unconditional and branch if equal, at the same IS, is very inefficient. You can just test if zero register equals to it self, ehich will always return true. In real life, having as much jumps possible is a nice thing. But in mc, maybe you could have optimized a little bit more!
I guess it is a bit late to change stuff around, but do not get me wrong. Von neumann and signed data and all other fancy stuff looks just great! I hope hearing more about this soon, and maybe a v3?
By removing those Operations from the IS, you can have much more space for other stuff (like interrupts, routine calls, rotation, pair registers..)!
Secondly, to me, having unconditional and branch if equal, at the same IS, is very inefficient. You can just test if zero register equals to it self, ehich will always return true. In real life, having as much jumps possible is a nice thing. But in mc, maybe you could have optimized a little bit more!
I guess it is a bit late to change stuff around, but do not get me wrong. Von neumann and signed data and all other fancy stuff looks just great! I hope hearing more about this soon, and maybe a v3?