code wiki / _hdl_build / nx_vcodec_pskip_probe.nx

nx_vcodec_pskip_probe.nx

buildroot/runtime/_hdl_build/nx_vcodec_pskip_probe.nx

4834 B94 linesdepth 6pulls 16 transitivereach 0 importersview sourcekind probetopic vcodec
docsdependenciesstructsconstsfunctions

about

nx_vcodec_pskip_probe.nx -- SIZE THE P-SKIP GAP (seq1105 rung-1 diagnostic). x264 skips 41.9% of foreman P-MBs; its P-SKIP is a MOTION-COMPENSATED copy at the predicted MV with zero residual. OUR skip is ZERO-MV-only (dosk iff vm_sad_zero <= sad_thresh) -- on panning content it may never fire. This probe replays the SHIPPED P chain (rct9, emode 673) on a y4m and, per P frame, counts against the SAME pre-encode reference the encoder saw: ourskip = MBs with zero-MV SAD <= qp*94 (what the shipped gate can skip today) mcskip = MBs with best-searched SAD <= qp*94 (what an MC-skip/merge mode could skip) The mcskip-ourskip delta is the addressable population of rung-1. license: ORIGINAL

dependencies 2 imports · 0 importers

nx_syscalls.nx nx_video_codec_wasm.nx nx_vcodec_pskip_probe.nx

imports: nx_syscalls.nxnx_video_codec_wasm.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 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 ↻ sys_read_file sys_openat_rd sys_lseek sys_mmap ↻ sys_read sys_munmap sys_close gw sys_write ↻ frames runq sys_mmap ↻ vc_t8_init nx_dct8_init vc_t8p vv_enc_rc vc_enc_frame_packed_rc rc_enc_init nx_bw_put vc_aq_level vc_mb_mad vc_enc_block_packed_e vc_enc_block_packed_i nx_bw_put ↻

structs

none

consts

12const SW: i64 = 352
13const SH: i64 = 288
14const NF: i64 = 48
15const KEYMUL: i64 = 188
16const PMUL: i64 = 94

functions

18func gw(s: *u8) -> i64 { var n: i64=0; while s[n]!=(0 as u8){n=n+1} sys_write(1,s,n); return 0 }
called by 2: runqmain calls 1: sys_write
19func gn(v: i64) -> i64
called by 1: runq calls 2: sys_mmapsys_write
23func cpb(d: *u8, s: *u8, n: i64) -> i64 { var i: i64=0; while i<n { d[i]=s[i]; i=i+1 } return 0 }
called by 1: runq
25func frames(yuv: *u8, hi: i64, total: i64, fp: *i64, fdata: i64) -> i64
called by 1: main
32func runq(fp: *i64, qp: i64, sz: i64) -> i64
77func main(argc: i64, argv: *i64) -> i64