[{"data":1,"prerenderedAt":120},["ShallowReactive",2],{"run:lego-phase-f3-baseline-5k":3,"run-findings:lego-phase-f3-baseline-5k":93},{"run":4,"config":20,"metrics":54,"curve":60,"assets":92},{"id":5,"title":5,"subtitle":6,"eyebrow":7,"date":8,"workspace":9,"commit":10,"tags":11,"verdict":19},"lego-phase-f3-baseline-5k","Phase F.3 baseline (SPLAT_RADIX_CPU_SCAN=1, legacy CPU scan) for radix GPU scan A\u002FB","Run summary · P1 Phase F.3 radix GPU prefix scan A\u002FB baseline","2026-05-25","splat","f243e1a",[12,13,14,15,16,17,18],"p1-profile","axis-1","phase-f","f3-baseline","lego-5k","premultiplied","cpu-scan","partial",[21,24,27,30,33,36,39,42,45,48,51],{"key":22,"value":23},"dataset","\u002FUsers\u002Fotkrickey\u002Fdev\u002F3dgs-workspace\u002Fdatasets\u002Fnerf_synthetic\u002Flego",{"key":25,"value":26},"gt_convention","Premultiplied",{"key":28,"value":29},"iterations","5,000",{"key":31,"value":32},"seed","42",{"key":34,"value":35},"capacity","1,000,000 splats",{"key":37,"value":38},"sh_degree","3",{"key":40,"value":41},"loss","L1Ssim",{"key":43,"value":44},"lambda","0.200",{"key":46,"value":47},"ssim","window=7 sigma=1",{"key":49,"value":50},"backend.backward","Simd",{"key":52,"value":53},"backend.loss_path","Gpu",{"psnr":55,"wallclock":56,"wallclock_regress":57,"splats":58,"final_loss":59},31.604379653930664,"1m 55s",false,84840,"1.453633e-2",{"loss":61,"splats":86},{"iters":62,"values":74},[63,64,65,66,67,68,69,70,71,72,73],1,500,1000,1500,2000,2500,3000,3500,4000,4500,5000,[75,76,77,78,79,80,81,82,83,84,85],0.5938448309898376,0.0862184390425682,0.0549686998128891,0.03164608031511307,0.019628599286079407,0.01719672605395317,0.01610240340232849,0.015447881072759628,0.014978615567088127,0.014693493023514748,0.014536334201693535,{"iters":87,"values":88},[63,64,65,66,67,68,69,70,71,72,73],[89,90,91,58,58,58,58,58,58,58,58],5207,842,11494,{"has_renders":57,"has_splat":57},[94],{"id":95,"title":96,"date":8,"status":97,"polarity":98,"category":99,"axes":100,"tags":101,"task_code":111,"related_runs":112,"delta_psnr":114,"delta_wallclock":115,"rank":116,"verdict":117,"impact_summary":118,"detail_path":119},"p1-axis1-phase-f3-radix-gpu-prefix-falsified","Phase F.3 radix GPU prefix scan — bit-exact 実装完成だが Metal implicit fences で +7.4% wallclock \u002F +35-41% per-call regression、audit Tier 2 仮説 falsified","stable","negative","audit",[63],[102,14,103,104,105,106,107,108,109,110,16],"p1-axis1","radix-sort","gpu-prefix-scan","tier-2","falsified","negative-finding","metal-fences","tbdr","ab-test","P1 Phase F.3",[5,113],"lego-phase-f3-gpu-scan-5k","+0.03 dB (parity、bit-exact 経路、session noise 内)","+7.4% (+8.58s @ 5k iter、sanity rerun +4.8% でも regression 確定)","high","audit-falsified-tier-2","audit (p1-axis1-metal-opt-audit) で Tier 2 「radix_sort GPU prefix sum、-0.5-0.8% wallclock、LOW PSNR risk」と分類した候補を empirical 検証。\u003Cstrong>bit-exact 実装は完成\u003C\u002Fstrong> (16-thread single-threadgroup kernel、Apple SIMD prefix exclusive sum + per-digit serial scan、100k random \u002F 500k packed keys \u002F edge cases 6 種で CPU stable sort と byte-for-byte 一致)、しかし 5k Lego smoke で \u003Cstrong>wallclock 115.83s → 124.41s (+7.4%)、ts_fwd_radix_sort 4.768 → 6.733 ms\u002Fcall (+41%)\u003C\u002Fstrong> の net regression。sanity re-run (118.73s \u002F 6.402ms) で再現確認、run-to-run 変動の上。\u003Cstrong>PSNR は parity\u003C\u002Fstrong> (31.604 → 31.635 dB、bit-exact 経路で 0 drift 期待、観測 +0.03 dB は session noise)。\u003Cstrong>Likely mechanism\u003C\u002Fstrong>: StorageModeShared buffer での back-to-back compute encoder 間で Metal が implicit fence を挿入 (hist→scan の buf_hist、scan→scatter の buf_offsets で read-after-write hazard)、TBDR pipeline stall。旧 CPU 経路は buf_hist→buf_offsets 変換を host で実行するため GPU 内 memory dependency が無く、\u003Cstrong>「除去した wait_until_completed」は実は CPU prefix scan と overlap していた active work\u003C\u002Fstrong> だった。教訓: 「CPU 介在を on-GPU に置換」族の audit 予測は overlap の存在を見落とすため systematically overestimate、Tier 2 同 family (backward TBDR、tile-local accumulator) の skip 判断強化。kernel + tests は env \u003Ccode>SPLAT_RADIX_GPU_SCAN=1\u003C\u002Fcode> で opt-in (future workload hedge)。","\u002Ffindings\u002Fp1-axis1-phase-f3-radix-gpu-prefix-falsified\u002F",1782449788228]