code wiki / (root) / nx_i8dot_ab.nx

nx_i8dot_ab.nx

buildroot/runtime/nx_i8dot_ab.nx

1872 B50 linesdepth 4pulls 6 transitivereach 0 importersview sourcekind tool
docsdependenciesstructsconstsfunctions

about

nx_i8dot_ab.nx -- same-process kernel A/B: SSE __f32_i8dot32 vs AVX2 __f32_i8dot32a. Tight loop, same data -> noise-immune ns/call ratio + exact-int equality check (they differ in FP order for random data, but exact-int inputs are order-independent -> must match). expect_exit: 0

dependencies 5 imports · 0 importers

nx_syscalls.nx nx_tier.nx nx_f32.nx nx_f32_cvt.nx nx_fmt.nx nx_i8dot_ab.nx

imports: nx_syscalls.nxnx_tier.nxnx_f32.nxnx_f32_cvt.nxnx_fmt.nx

imported by: nobody (leaf or entry point)

call flow from main pre-order; caps 40 nodes / depth 6 declared; ↻ = already shown

main sys_mmap st4 nx_i32_to_f32 fmt_puts fmt_puts_fd sys_write ab_nl fmt_puts ↻ sys_now_us sys_mmap ↻ sys_clock_gettime_mono fmt_putn fmt_putn_fd sys_write ↻

structs

none

consts

10const K_MAGIC_20000000: i64 = 20000000

functions

12func ab_nl() -> i64 { fmt_puts("\n" as *u8); return 0 }
called by 1: main calls 1: fmt_puts
13func st4(p: *u8, idx: i64, bits: i64) -> i64
called by 1: main
17func main() -> i64