Running a drawing
Every netlist is public, and the browser can execute it gate by gate. That is what makes a card a circuit rather than an image of one.
What the simulator does
The netlist is a flat byte string. Each gate is seven bytes: one operation byte, then three bytes for input A and three for input B. Node 0 and 1 are constants; inputs start at node 2.
The simulator decodes that string in your browser, wires the nodes, and evaluates the circuit for the inputs you set. Nothing is sent to a server, and no wallet is needed.
Reading the output
For a 4-bit adder, inputs A and B are two 4-bit numbers; the outputs are the sum and the carry. For a comparator, the outputs are three bits: equal, greater, less. For the 12×12 multiplier, two 12-bit inputs produce a 24-bit product.
| Drawing | Inputs | Outputs | What it computes |
|---|---|---|---|
| MUL12 | 24 | 24 | multiplies two 12-bit numbers |
| ALU16 | 35 | 16 | eight operations on 16-bit words |
| DIV8 | 16 | 16 | quotient and remainder of 8-bit division |
| MUL8 | 16 | 16 | multiplies two 8-bit numbers |
| CMP64 | 128 | 3 | equal / greater / less for 64-bit words |
| CMP4 | 8 | 3 | equal / greater / less for 4-bit words |
| EQ8 | 16 | 1 | 1 when two 8-bit words match |
| SUB4 | 8 | 5 | difference and borrow |
| ADD4 | 9 | 5 | adds two 4-bit numbers with carry |
| MUL2 | 4 | 4 | multiplies two 2-bit numbers |
| SHIFT4 | 6 | 4 | shifts a 4-bit word by 0–3 |
The gate array
Under the inputs the page draws one small block per gate, in netlist order, and lights them with their real values as evaluation runs. The length of that strip is the gate count. A multiplier's strip is 53 times longer than a shifter's, which is the most direct explanation of why it costs more.
What this proves
- The drawing is a real circuit, not an illustration.
- The gate count in the price is the number of gates in the netlist.
- The card you buy is the same netlist that runs in the page — the chain stores it byte for byte.
Limits
The simulator evaluates the netlist; it does not prove the drawing is the most efficient implementation of that function. A 976-gate divider and a 400-gate divider could compute the same thing — the price follows the gates, not the elegance.