nx_nofloat_multistep_gate.nx
buildroot/runtime/nx_nofloat_multistep_gate.nx
about
nx_nofloat_multistep_gate.nx -- WHY no-float wins the MULTI-STEP-LOGIC arena (operator thesis 2026-06-18).
The exceed isn't raw IQ -- it's that a long chain of EXACT, reproducible steps lets logic compound
reliably, where float DRIFTS step-by-step until the chain is wrong. Proof: a 1000-step chain where each
step is the trivial logical op "(s + BIG) - BIG + 1" = net +1. Integer does it EXACTLY (s = N every
length); float (BIG=2^24, ulp=2) loses each +1 inside the BIG+-BIG round-off -> s STAYS ~1 forever, the
error compounds, the chain is wrong -- and the gap GROWS with chain length. This is the capability that
matters for multi-step functional/math reasoning + for a system that builds capabilities PIECE BY PIECE
(each proven step must feed the next on solid ground). No hw writes (Rule 26). expect_exit: 0 tier: ORIGINAL
dependencies 2 imports · 0 importers
imports: nx_syscalls.nxnx_gate_verdict.nx
imported by: nobody (leaf or entry point)
call flow from main pre-order; caps 40 nodes / depth 6 declared; ↻ = already shown
structs
| none |
consts
| 12 | const BIG: i64 = 16777216 // 2^24 -- ulp here is 2, so +1 vanishes in float |
functions
| 14 | func ms_puts(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} sys_write(1,s,n); return 0 } |
| 15 | func ms_num(v: i64) -> i64 { let b: *u8=sys_mmap(28); var m: i64=v; if m<0{m=0-m;sys_write(1,"-" as *u8,1)} let t: *u8=sys_mmap(28); var k: i64=0; if m==0{t[0]=48 as u8;k=1} while m>0{t[k]=(48+(m%10)) as u8;m=m/10;k=k+1} var i: i64=0; while i<k{b[i]=t[k-1-i];i=i+1} sys_write(1,b,k); return 0 } |
| 18 | func int_chain(N: i64) -> i64 { var s: i64=0; var n: i64=0; while n<N { s=(s+BIG)-BIG+1; n=n+1 } return s } called by 1: main |
| 20 | func flt_chain(N: i64) -> i64 called by 1: main |
| 30 | func main() -> i64 |