code wiki / _hdl_build / nx_nofloat_mxfp8_gate.nx

nx_nofloat_mxfp8_gate.nx

buildroot/runtime/_hdl_build/nx_nofloat_mxfp8_gate.nx

3502 B71 linesdepth 3pulls 4 transitivereach 0 importersview sourcekind gate/prooftopic nofloat
docsdependenciesstructsconstsfunctions

about

nx_nofloat_mxfp8_gate.nx -- gate: MXFP8 (OCP E4M3 + E5M2) dequant to Q16, EXACT known values. T1 E4M3 normals decode to exact Q16 (1.0, 1.5, 2.0, 0.5, -1.0) T2 E4M3 subnormal (2^-7) + NaN (E=1111,M=111) -> 0 T3 E5M2 normals decode to exact Q16 (1.0, 1.25, 0.5) T4 E5M2 subnormal (2^-16) + inf (E=11111,M=0) -> 0 T5 E8M0 shared block scale: same element, scale 2^+1 doubles, 2^-1 halves T6 block decode: 32-element E4M3 block all decode + bit-identical on a second pass (deterministic) license_tier: ORIGINAL No hw writes (Rule 26). expect_exit: 0

dependencies 3 imports · 0 importers

nx_nofloat_mxfp8.nx nx_gate_verdict.nx nx_syscalls.nx nx_nofloat_mxfp8_gate.nx

imports: nx_nofloat_mxfp8.nxnx_gate_verdict.nxnx_syscalls.nx

imported by: nobody (leaf or entry point)

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

main gv_ctr sys_mmap nxa_die sys_write sys_exit nxa_lock_take nxa_lock_addr sys_write ↻ nxa_lock_give nxa_lock_addr ↻ nxa_report_overrun sys_write ↻ nxa_dump_printable sys_write ↻ nxa_dump_sizes sys_write ↻ gv_head gv_puts sys_write ↻ mx8_e4m3 mx8_shift gv_check gv_puts ↻ mx8_e5m2 mx8_shift ↻ sys_mmap ↻ mx8_block_e4m3 mx8_e4m3 ↻ gv_verdict gv_note_bare_rate gv_bare_rate gv_at gv_obj_has_n gv_at ↻ gv_puts ↻ gv_num sys_mmap ↻ sys_write ↻ sys_munmap

structs

none

consts

13const G8_S0: i64 = 127 // E8M0 scale exponent for 2^0 = 1

functions

15func main() -> i64