nx_abs_ab.nx
buildroot/runtime/nx_abs_ab.nx
about
nx_abs_ab.nx -- rigorous SAME-PROCESS A/B of branchy vs branchless abs in
the jitter update. Both variants run the IDENTICAL work (only the abs
differs), measured INTERLEAVED with min-of-K so between-batch system-load
drift and upward noise cancel (RACING_TEAM_BENCHMARK_DOCTRINE: separate
signal from confound; min is the least-disturbed estimate). This is the
honest way to decide a ~2-cycle question -- single-run amortized numbers
across separate processes are dominated by load variance, not the code.
license_tier: ORIGINAL
dependencies 1 imports · 0 importers
imports: nx_syscalls.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 K_MAGIC_2654435761: i64 = 2654435761 |
| 13 | const K_MAGIC_2000000: i64 = 2000000 |
| 14 | const K_MAGIC_200000: i64 = 200000 |
| 15 | const K_MAGIC_1000000000000: i64 = 1000000000000 |
functions
| 17 | func bdec(n: i64) -> i64 |
| 30 | func run_branchy(n: i64) -> i64 called by 1: main |
| 50 | func run_free(n: i64) -> i64 called by 1: main |
| 70 | func main() -> i64 |