nonbinary.computer
OR-1 dataflow CPU sketch
1; OR1 Monitor Demo Program
2;
3; Computes (3 + 7) on PE0, writes the result (10) to SM cell 0,
4; then PE1 reads it back and checks if it equals 10 using SWEQ.
5;
6; This exercises:
7; - Seed token injection (const nodes)
8; - Dyadic matching (add waits for both operands)
9; - Cross-PE token routing (PE0 -> PE1)
10; - SM write and read with I-structure semantics
11; - Comparison / switch routing (sweq)
12; - Multiple PEs active across different ticks
13;
14; Expected execution flow:
15; t=0: const seeds arrive at c_three, c_seven, c_expected, c_read_trigger
16; t=0: c_three and c_seven match at adder -> executes ADD(3,7) = 10
17; t=0: adder emits result (10) to writer on PE0
18; t=0: writer is monadic (const addr), emits WRITE(cell=0, data=10) to SM0
19; t=0: SM0 writes cell 0 = 10
20; t=0: c_read_trigger fires reader on PE1 -> READ(cell=0)
21; t=0: SM0 satisfies read (cell already FULL) -> returns 10 to checker on PE1
22; t=0: c_expected (10) and SM read result (10) match at checker -> SWEQ
23; t=0: checker compares: 10 == 10 -> bool_out=true
24;
25; In the monitor you should see:
26; - PE0 cluster: c_three, c_seven, adder, writer all light up
27; - PE1 cluster: c_expected, c_read_trigger, reader, checker light up
28; - SM cell 0 transitions from EMPTY to FULL
29; - Token flow edges animate across PE boundary
30; - Half-match visible on adder/checker before second operand arrives
31
32@system pe=2, sm=1
33
34; PE0: compute 3 + 7 and write result to SM
35&c_three|pe0 <| const, 3
36&c_seven|pe0 <| const, 7
37&adder|pe0 <| add
38&writer|pe0 <| write, 0
39
40&c_three|pe0 |> &adder|pe0:L
41&c_seven|pe0 |> &adder|pe0:R
42&adder|pe0 |> &writer|pe0:L
43
44; PE1: read SM cell 0 and compare against expected value
45&c_expected|pe1 <| const, 10
46&c_read_trigger|pe1 <| const, 0
47&reader|pe1 <| read, 0
48&checker|pe1 <| sweq
49
50&c_read_trigger|pe1 |> &reader|pe1:L
51&reader|pe1 |> &checker|pe1:L
52&c_expected|pe1 |> &checker|pe1:R