{"key":"swar-1","variants":[{"name":"base","desc":"latt-gas-1 LattFilter_base.sol rebuilt and replayed here (scalar exact kernel, cap 12): reproduction of the episode-2 port","peak_gas":4594219,"mean_gas":209817,"by_dt":{"1":558394,"2":860654,"3-5":1665604,"6+":4594219},"mean_by_dt":{"0":31858,"1":264073,"2":338926,"3-5":508896,"6+":1036235},"dt0_peak":31915,"max_err_mean_bps":17.9149,"max_err_sd_bps":9.8588,"max_err_pi":0.180818,"share_mean_err_le_0_5bps":0.9992,"share_mean_err_le_0_01bps":0.9979,"p99_mean_err_bps":0.0004,"bytes_runtime":6370,"slots":13,"reverts":0,"feasible":false},{"name":"a","desc":"(a) SWAR: 4 x 64-bit lanes a word, one MUL per four cells per tap, exact 33-tap kernel unrolled, all 12 silent steps exact","peak_gas":2421322,"mean_gas":168432,"by_dt":{"1":402544,"2":544877,"3-5":966893,"6+":2421322},"mean_by_dt":{"0":31889,"1":218068,"2":258527,"3-5":361430,"6+":676283},"dt0_peak":31946,"max_err_mean_bps":17.9151,"max_err_sd_bps":9.8589,"max_err_pi":0.180819,"share_mean_err_le_0_5bps":0.9992,"share_mean_err_le_0_01bps":0.9977,"p99_mean_err_bps":0.0004,"bytes_runtime":10762,"slots":13,"reverts":0,"feasible":false},{"name":"b","desc":"(b) precomputed silent operator: the r-step Gaussian kernel (r = 1..12, 20-bit taps, packed constants) applied once with one cut","peak_gas":894283,"mean_gas":164110,"by_dt":{"1":402348,"2":546640,"3-5":676244,"6+":894283},"mean_by_dt":{"0":31907,"1":218059,"2":259606,"3-5":320517,"6+":389940},"dt0_peak":31964,"max_err_mean_bps":30.2616,"max_err_sd_bps":15.7568,"max_err_pi":0.323361,"share_mean_err_le_0_5bps":0.7869,"share_mean_err_le_0_01bps":0.5403,"p99_mean_err_bps":5.3003,"bytes_runtime":11036,"slots":13,"reverts":0,"feasible":false},{"name":"h2","desc":"(a)+(b): 2 exact SWAR silent steps, the rest as one precomputed r-step kernel and one cut","peak_gas":1180132,"mean_gas":167193,"by_dt":{"1":402544,"2":544877,"3-5":917293,"6+":1180132},"mean_by_dt":{"0":31889,"1":218077,"2":258556,"3-5":359192,"6+":512029},"dt0_peak":31946,"max_err_mean_bps":24.4419,"max_err_sd_bps":12.6092,"max_err_pi":0.263047,"share_mean_err_le_0_5bps":0.9711,"share_mean_err_le_0_01bps":0.8834,"p99_mean_err_bps":1.241,"bytes_runtime":10762,"slots":13,"reverts":0,"feasible":false},{"name":"h3","desc":"(a)+(b): 3 exact silent steps, then one precomputed kernel","peak_gas":1335138,"mean_gas":167728,"by_dt":{"1":402544,"2":544877,"3-5":966914,"6+":1335138},"mean_by_dt":{"0":31889,"1":218073,"2":258525,"3-5":361435,"6+":571698},"dt0_peak":31946,"max_err_mean_bps":23.2753,"max_err_sd_bps":12.0239,"max_err_pi":0.249672,"share_mean_err_le_0_5bps":0.9863,"share_mean_err_le_0_01bps":0.9422,"p99_mean_err_bps":0.6877,"bytes_runtime":10755,"slots":13,"reverts":0,"feasible":false},{"name":"h4","desc":"(a)+(b): 4 exact silent steps, then one precomputed kernel (cheapest variant within the 0.5 bps / 99% accuracy bar)","peak_gas":1487686,"mean_gas":168014,"by_dt":{"1":402544,"2":544877,"3-5":966893,"6+":1487686},"mean_by_dt":{"0":31889,"1":218071,"2":258527,"3-5":361434,"6+":614080},"dt0_peak":31946,"max_err_mean_bps":22.3191,"max_err_sd_bps":11.5638,"max_err_pi":0.238238,"share_mean_err_le_0_5bps":0.9932,"share_mean_err_le_0_01bps":0.971,"p99_mean_err_bps":0.2714,"bytes_runtime":10755,"slots":13,"reverts":0,"feasible":false},{"name":"h6","desc":"(a)+(b): 6 exact silent steps, then one precomputed kernel","peak_gas":1796258,"mean_gas":168261,"by_dt":{"1":402544,"2":544877,"3-5":966893,"6+":1796258},"mean_by_dt":{"0":31889,"1":218069,"2":258527,"3-5":361430,"6+":650899},"dt0_peak":31946,"max_err_mean_bps":20.7508,"max_err_sd_bps":10.8587,"max_err_pi":0.218451,"share_mean_err_le_0_5bps":0.9976,"share_mean_err_le_0_01bps":0.9911,"p99_mean_err_bps":0.0047,"bytes_runtime":10762,"slots":13,"reverts":0,"feasible":false}],"python_only_variants":[{"name":"c_qsd1","desc":"(c) quasi-stationary: 1 exact silent step, then blend toward the cosine shape of the cut interval, decay lambda2^r","share_mean_err_le_0_5bps":0.8086,"max_err_mean_bps":29.532,"p99_mean_err_bps":4.659},{"name":"c_qsd2","desc":"(c) as above, 2 exact steps","share_mean_err_le_0_5bps":0.9152,"max_err_mean_bps":14.733,"p99_mean_err_bps":2.367},{"name":"c_qsd3","desc":"(c) as above, 3 exact steps","share_mean_err_le_0_5bps":0.9634,"max_err_mean_bps":11.897,"p99_mean_err_bps":1.409},{"name":"d_q16","desc":"cells quantised to 16 bits of the largest cell after every convolution (lane-width test)","share_mean_err_le_0_5bps":0.9999,"max_err_mean_bps":2.287,"p99_mean_err_bps":0.036},{"name":"d_q12","desc":"cells quantised to 12 bits","share_mean_err_le_0_5bps":0.9913,"max_err_mean_bps":6.815,"p99_mean_err_bps":0.463},{"name":"d_km11","desc":"kernel truncated to +-11 cells (2.9 sigma), renormalised","share_mean_err_le_0_5bps":0.9481,"max_err_mean_bps":72.089,"p99_mean_err_bps":2.444},{"name":"d_km8","desc":"kernel truncated to +-8 cells","share_mean_err_le_0_5bps":0.5815,"max_err_mean_bps":90.751,"p99_mean_err_bps":30.232}],"bench_gas":{"swar_24in_24out":45756,"swar_24in_56out":79325,"swar_multi9_24in_24out":164144,"exact_24in_24out":83576,"exact_24in_56out":194904,"cut_24":4071,"norm_24":9585,"moments_24":4782,"recentre_24":26895,"densAfter":7227,"lnWad":1952,"expNeg":2308,"sqrtU":424,"alloc_121":1465,"store_24_first":105032,"store_24_cold_nz":25426,"load_24_cold":20116,"conv8_unrolled_64cells_8x32bit_lanes":12555},"best":"h4","feasible":false,"best_why":"cheapest Solidity variant meeting the accuracy bar (99.32% of callbacks within 0.5 bps); its gas (peak 1.49M, dt=1 peak 403k) is far over 120k","seeds":"replica 10000..10004, latt.py qm_k1e-5.json run blind (latt-gas-1 record.py); recorded edges 443.663520, 571.977293, 526.411631 for 10000..10002 match gates.py; 53268 callbacks; no engine seed used","compiler":"solc 0.8.24 viaIR, optimizer runs 200, evm paris, no inline assembly in the filter (the forge harness uses assembly)","gas_method":"each callback its own transaction (forge --isolate), cold storage, refunds not subtracted; filter only, no fee tables","code_zb64":"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","code_sha256":"c4c1e04f1dcbe4fd677225f0271d8969241d98e1501bc3f48e82b9c25d33129d","notes":"Gas is whole-tx gasUsed: forge --isolate call gas minus 122, equal to anvil receipts on 25 calls; this harness puts latt-gas-1's base at 31.9k for dt=0 and 558k dt=1 peak (it reported 25.1k/518k). No variant fits: SWAR (a) cuts dt=1 mean 264k->218k and peak 558k->403k only; silent steps stay ~45k each. Per-function bench (24-cell support): SWAR step 46k, store 25k, load 20k, recentre 27k, norm 10k, ln 2k. Array-based Solidity costs ~200-400 gas per cell-op (bounds checks, stack-to-memory), so per-cell loops, not MULs, dominate. (b) alone is cheap but wrong (79% within 0.5 bps); hybrids need 4 exact steps for 99%; (c) is worse than (b). Kernel truncation fails (tails carry the arb density); 16-bit cells are fine (99.99%). Next: whole belief as 8 words of 8x32-bit lanes (64 cells, 16-bit masses, 16-bit taps), every op word-parallel on static memory arrays with constant indices: an unrolled 64-cell conv measured 12.6k, ~3-6k per silent step on the interval words; cut by lane masks, sums by one MUL with a ones vector, 4 storage slots. Estimate, not measured: ~110-150k peak with 4 exact steps + one precomputed kernel.","bench_gas_note":"gas inside one forge test frame (not a transaction): one call of each internal function of variant a on a 24-cell support; store/load with storage cooled"}