/sci/ - Science & Math

>>12040840
so you really only need to know how a NAND gate works, even tho modern computers uses many different kinds of gates. a cool fact that you find in boolean logic is that you can construct all other logic using either just NANDs or NORs. im pretty sure NANDs also happen to be the absolute cheapest gates, as you can construct them very simply with 2 transistors. in terms of memory cost tho, NAND memory is relatively expensive compared to just using magnetic HDDs or a transistor/capacitor combo (RAM).

so the pic related is using BJT transistors which are different than MOSFETs but BJT NAND gates are as simple as it gets (MOSFET NANDs require 2 extra gates, that are p-channel if i recall), and they too only use n-type and p-type shenanigans. if you apply a "1" to either A or B it opens the gate. so if you apply a "1" to both A and B it will make the voltage connect to ground so the output is 0. you can only make the output go to zero when you input a 1 at both A and B.

now that you understand gates. you can now build adders that will add two n-bit numbers. you can then build an ALU (brain of the cpu kinda) which will do lots of different things like increment a binary number which can represent a memory address or add two together which will represent data to be stored at a certain memory addresse which will represent some monitor output for graphics or some shit

>>11542897
you don't really need to know but they're stupidly simple. it's a basic property of boolean logic that you can construct ALL other logic out of NAND logic. so you could construct a computer using only NAND gates that can do everything any other computer can do (albeit it won't be as fast, surely).

a NAND gate is extremely simple, just two transistors like so (pic related).