code wiki / _hdl_build / nx_f32_hw_matmul_gate.nx

nx_f32_hw_matmul_gate.nx

buildroot/runtime/_hdl_build/nx_f32_hw_matmul_gate.nx

3817 B94 linesdepth 4pulls 7 transitivereach 0 importersview sourcekind gate/prooftopic f32
docsdependenciesstructsconstsfunctions

about

nx_f32_hw_matmul_gate.nx -- proves the hardware-float GEMM is correct. T1 kat : hw 2x2 [[1,2],[3,4]]@[[5,6],[7,8]] == [[19,22],[43,50]] T2 differential : hw == software nx_f32_matmul BIT-FOR-BIT on an 8x8 (binary32 cross-check) T3 bad-dim : m=0 -> NX_HWMM_ERR GREEN only if all three hold. Evidence -> knowledge/status/f32_hw_matmul_gate.log. Sovereign: imports nx_f32_hw_matmul (hw, -> nx_f32_hw SSE) + nx_f32_matmul (sw ref, -> nx_f32) + nx_framed_append + nx_syscalls. license_tier: ORIGINAL

dependencies 4 imports · 0 importers

nx_f32_hw_matmul.nx nx_f32_matmul.nx nx_framed_append.nx nx_syscalls.nx nx_f32_hw_matmul_gate.nx

imports: nx_f32_hw_matmul.nxnx_f32_matmul.nxnx_framed_append.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 hmm_puts sys_write sys_mmap f32_of nx_f32_hw_matmul f32_int nx_f32_matmul hg_row sys_mmap ↻ hg_cat fa_appendz fa_append sys_mmap ↻ sys_openat_append sys_flock fa_write_all sys_write ↻ sys_close fa_len hmm_puts ↻ hg_cat ↻ hg_catn sys_mmap ↻ fa_appendz ↻

structs

none

consts

15const HG_LOG: *u8 = "knowledge/status/f32_hw_matmul_gate.log"

functions

17func hg_cat(dst: *u8, off: i64, s: *u8) -> i64 { var i: i64 = 0; while s[i] != 0 as u8 { dst[off + i] = s[i]; i = i + 1 } return off + i }
called by 2: hg_rowmain
18func hg_catn(dst: *u8, off: i64, v: i64) -> i64
called by 1: main calls 1: sys_mmap
28func hg_row(name: *u8, pass: i64) -> i64
called by 1: main calls 4: sys_mmaphg_catfa_appendzhmm_puts
41func main() -> i64
calls 10: hmm_putssys_mmapf32_ofnx_f32_hw_matmulf32_intnx_f32_matmul+4