igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity

adv-accept: first live results (five low-ratio programs, plant and control), logs copied; Q2 gap tool

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-07 18:43:12 +00:00
parent fe9dac31e5
commit e0d9ed9526
5 changed files with 639 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

47
docs/analysis/cryptanalysis/logs/adv-accept/live-control-p2.log Normal file
View file

@ -0,0 +1,47 @@
[2026-10-07T18:30:09Z] attack-f8 0.1.0 (igneum-pow 4); args ["warps", "--program", "2", "--nonces", "1000000", "--threads", "32", "--out", "/srv/builds/igneum-wt-adv-accept/adv/live", "--validate", "sample"]
[2026-10-07T18:30:09Z] census warps: programs [2] day 20730 nonces 1000000 threads 32 plant none validate sample check_every 997 diag false
[2026-10-07T18:30:24Z] day 20730: cache filled in 15.39 s, fnv 448274a57f508cbc
[2026-10-07T18:30:25Z] table: 16777216 items derived with their 8 lines in 1.3 s; library comparison sample (0 mismatches)
[2026-10-07T18:30:30Z] program p2-attack-f8: epoch seed 59cef1aaa406506685f4355933868219bedece8dbd4e84c281724db9313bfdfa era seed 9cba001faa1e017235841c72f560ebd1b93f6d7aea5f9a598f7d3080e8a11f69 id 8a0047b2eb071ade attempt 5 class mx8-era35d128b9+sh256x27 op mix load=16 mul=8 add=6 mulhi=6 shfl=6 mad=4 or=4 rotl=4 rotr=4 xor=4 sub=2; era stride mul 0xbf0fab63 rot 4 interleave [2, 3, 7, 15] windows 2:2:0 3:0:0 7:0:0 9:0:0 13:2:1 17:1:1 21:2:0 22:1:0 23:0:0 27:1:0 30:2:2 40:1:0 44:2:1 55:0:0 62:0:0 63:2:0
[2026-10-07T18:30:30Z] load site instr 2: src r1 win 2 off 0; last base writer add at 1; writers back to the last injecting one: add@1; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 3: src r5 win 0 off 0; last base writer mad at 0; writers back to the last injecting one: mad@0; shadow writes of r5 per rep: add=3 mad=4 mul=5 or=5 rotl=1 rotr=4 shfl=4 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 7: src r1 win 0 off 0; last base writer shfl at 5; writers back to the last injecting one: shfl@5; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 9: src r3 win 0 off 0; last base writer add at 6; writers back to the last injecting one: add@6; shadow writes of r3 per rep: add=4 mad=3 mul=1 mulhi=2 rotl=1 rotr=2 shfl=3 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 13: src r0 win 2 off 1; last base writer xor at 12; writers back to the last injecting one: xor@12; shadow writes of r0 per rep: add=5 mad=6 mul=1 mulhi=3 or=1 rotl=1 rotr=7 shfl=8 sub=2
[2026-10-07T18:30:30Z] load site instr 17: src r5 win 1 off 1; last base writer xor at 14; writers back to the last injecting one: xor@14; shadow writes of r5 per rep: add=3 mad=4 mul=5 or=5 rotl=1 rotr=4 shfl=4 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 21: src r4 win 2 off 0; last base writer shfl at 15; writers back to the last injecting one: shfl@15; shadow writes of r4 per rep: add=4 mad=4 mulhi=1 or=1 rotl=4 rotr=1 shfl=2 sub=8 xor=6
[2026-10-07T18:30:30Z] load site instr 22: src r5 win 1 off 0; last base writer add at 20; writers back to the last injecting one: add@20; shadow writes of r5 per rep: add=3 mad=4 mul=5 or=5 rotl=1 rotr=4 shfl=4 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 23: src r1 win 0 off 0; last base writer load at 13; writers back to the last injecting one: load@13; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 27: src r2 win 1 off 0; last base writer xor at 26; writers back to the last injecting one: xor@26; shadow writes of r2 per rep: add=6 mad=2 mul=4 mulhi=3 or=3 rotl=5 rotr=3 shfl=3 sub=3 xor=6
[2026-10-07T18:30:30Z] load site instr 30: src r2 win 2 off 2; last base writer rotr at 29; writers back to the last injecting one: rotr@29 xor@26; shadow writes of r2 per rep: add=6 mad=2 mul=4 mulhi=3 or=3 rotl=5 rotr=3 shfl=3 sub=3 xor=6
[2026-10-07T18:30:30Z] load site instr 40: src r1 win 1 off 0; last base writer rotl at 32; writers back to the last injecting one: rotl@32 rotl@31 load@30; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 44: src r4 win 2 off 1; last base writer load at 27; writers back to the last injecting one: load@27; shadow writes of r4 per rep: add=4 mad=4 mulhi=1 or=1 rotl=4 rotr=1 shfl=2 sub=8 xor=6
[2026-10-07T18:30:30Z] load site instr 55: src r4 win 0 off 0; last base writer mad at 51; writers back to the last injecting one: mad@51; shadow writes of r4 per rep: add=4 mad=4 mulhi=1 or=1 rotl=4 rotr=1 shfl=2 sub=8 xor=6
[2026-10-07T18:30:30Z] load site instr 62: src r0 win 0 off 0; last base writer add at 53; writers back to the last injecting one: add@53; shadow writes of r0 per rep: add=5 mad=6 mul=1 mulhi=3 or=1 rotl=1 rotr=7 shfl=8 sub=2
[2026-10-07T18:30:30Z] load site instr 63: src r1 win 2 off 0; last base writer load at 62; writers back to the last injecting one: load@62; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] window layer: site item windows (first, items) (0x0,2^22) (0x0,2^24) (0x0,2^24) (0x0,2^24) (0x400000,2^22) (0x800000,2^23) (0x0,2^22) (0x0,2^23) (0x0,2^24) (0x0,2^23) (0x800000,2^22) (0x0,2^23) (0x400000,2^22) (0x0,2^24) (0x0,2^24) (0x0,2^22); expected reads per item by quarter 11.444 9.537 5.722 3.815 (flat uniform 7.629)
[2026-10-07T18:30:33Z] 31250 warps (1000000 nonces) interpreted in 3.7 s (3.780 ms per warp per thread); Epoch::hash_warp agreement on 95 warps: 0 mismatches
[2026-10-07T18:30:33Z] p2-attack-f8 distinct lines per hash: n 1000000 min 1014 p1 1023 median 1024 p99 1024 max 1024 mean 1023.8709
[2026-10-07T18:30:33Z] p2-attack-f8 distinct items per hash: n 1000000 min 127 p1 128 median 128 p99 128 max 128 mean 127.9994
[2026-10-07T18:30:33Z] p2-attack-f8 distinct lines per warp: n 31250 min 32576 p1 32604 median 32636 p99 32663 max 32686 mean 32635.7294
[2026-10-07T18:30:33Z] p2-attack-f8 distinct items per warp: n 31250 min 4091 p1 4093 median 4096 p99 4096 max 4096 mean 4095.4226
[2026-10-07T18:30:33Z] uniform expectation: distinct lines per hash 1023.875 of 1024 reads, per warp 32640.3 of 32768 reads (2^22 lines); distinct items per hash 127.9995 of 128, per warp 4095.50 of 4096 (2^24 items)
[2026-10-07T18:30:33Z] positions: largest hi16-bucket (256 items) share 0.0104% at p16 (iteration 1, site 0, instr 2); windowed expectation 0.0061%; largest saturated-source share 0.0000% at p0 (site 0, instr 2)
[2026-10-07T18:30:33Z] acceptance-style (2,048 evaluations, the rule's sample size): the most repeated address at one position 2 of 2048 at p17 (site 1, instr 3); saturated sources at one position 0 of 2048 at p0 (site 0, instr 2); uniform expectation: repeats 1 to 2, saturated 0
[2026-10-07T18:30:34Z] p2-attack-f8 item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 32 (bin 1185685) z_max +8.82 min 0 z_min -2.76 chi2/dof 2.19250 chi2_z +3453.84 top0.1% 0.29907% top0.5% 1.34679% top1% 2.55754%
[2026-10-07T18:30:34Z] p2-attack-f8 item histogram against the window density: full 2^24 chi2/dof 1.00015 largest 23 (item 0x8c49bb) at +7.22 sigma smallest at -3.38; buckets64 chi2/dof 0.99742 largest 859 (bucket 24423) at +4.68 sigma smallest at -4.55
[2026-10-07T18:30:34Z] p2-attack-f8 WINDOWED CONTROL item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 37 (bin 3356675) z_max +10.63 min 0 z_min -2.76 chi2/dof 2.19210 chi2_z +3452.68 top0.1% 0.29918% top0.5% 1.34673% top1% 2.55749%
[2026-10-07T18:30:34Z] p2-attack-f8 WINDOWED CONTROL against the window density: full chi2/dof 0.99966 largest +7.55 sigma; buckets64 chi2/dof 1.00387 largest +4.49 smallest -4.89
[2026-10-07T18:30:34Z] 6-sigma p2-attack-f8 items buckets64 against the window density: largest bucket +4.68 sigma, smallest -4.55 -> within 6 sigma
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): f 0.1% S_f 0.29907% E_f(control) 0.29918% X_f -0.00012% X_f/f -0.0012 -> no hot set
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): f 0.5% S_f 1.34679% E_f(control) 1.34673% X_f +0.00006% X_f/f +0.0001 -> no hot set
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): f 1.0% S_f 2.55754% E_f(control) 2.55749% X_f +0.00005% X_f/f +0.0000 -> no hot set
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): verdict clear
[2026-10-07T18:30:34Z] p2-attack-f8 FLAT CONTROL item histogram: bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 28 (bin 13310856) z_max +7.37 min 0 z_min -2.76 chi2/dof 0.99990 chi2_z -0.29 top0.1% 0.24291% top0.5% 1.09940% top1% 2.08772%
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): f 0.1% S_f 0.29907% E_f(control) 0.24291% X_f +0.05616% X_f/f +0.5616 -> no hot set
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): f 0.5% S_f 1.34679% E_f(control) 1.09940% X_f +0.24739% X_f/f +0.4948 -> no hot set
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): f 1.0% S_f 2.55754% E_f(control) 2.08772% X_f +0.46981% X_f/f +0.4698 -> no hot set
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): verdict clear
[2026-10-07T18:30:34Z] ratio p2-attack-f8: top 0.1% / 0.5% / 1% share over the WINDOW-MODEL control 0.9996x / 1.0000x / 1.0000x (gate 1.2x at 0.1%: within); over the FLAT control 1.2312x / 1.2250x / 1.2250x
[2026-10-07T18:30:35Z] hottest item p2-attack-f8 (0x2d3b84): 32 reads (0.0000% of all); predicted source none
[2026-10-07T18:30:35Z] program p2-attack-f8 DONE: 1000000 nonces; 6-sigma (windowed) clear; hot-set clear; verdict PASS (no test fired)

235
docs/analysis/cryptanalysis/logs/adv-accept/live-lowratio-s01.log Normal file
View file

@ -0,0 +1,235 @@
[2026-10-07T18:36:58Z] attack-f8 0.1.0 (igneum-pow 4); args ["warps", "--program", "100211", "--nonces", "1000000", "--threads", "32", "--out", "/srv/builds/igneum-wt-adv-accept/adv/live", "--validate", "sample"]
[2026-10-07T18:36:58Z] census warps: programs [100211] day 20730 nonces 1000000 threads 32 plant none validate sample check_every 997 diag false
[2026-10-07T18:36:58Z] day 20730: cache filled in 0.50 s, fnv 448274a57f508cbc
[2026-10-07T18:37:01Z] table: 16777216 items derived with their 8 lines in 2.7 s; library comparison sample (0 mismatches)
[2026-10-07T18:37:06Z] program p100211-attack-f8: epoch seed 3d4686b780097991a4600a0683fca3645e627c67edd53365c4131c7095ecc55d era seed 3e3a196a3504d573eee9656bffb7e3d2dffc03dc10f95f92fd08d0ab245641f7 id d07885a437e237c7 attempt 0 class mx8-eraa8bce51f+sh256x27 op mix load=16 xor=11 mul=10 add=7 mad=6 mulhi=4 shfl=4 rotl=3 or=1 rotr=1 sub=1; era stride mul 0x4babb2b7 rot 31 interleave [1, 6, 8, 13] windows 3:1:1 4:1:0 7:0:0 15:2:0 17:2:2 23:0:0 24:1:0 27:2:3 29:2:1 31:1:0 35:0:0 41:1:0 44:2:1 49:0:0 61:2:1 63:1:0
[2026-10-07T18:37:06Z] load site instr 3: src r4 win 1 off 1; last base writer xor at 2; writers back to the last injecting one: xor@2; shadow writes of r4 per rep: add=6 mad=2 mul=4 mulhi=7 or=1 rotl=6 rotr=5 shfl=3 sub=3 xor=5
[2026-10-07T18:37:06Z] load site instr 4: src r0 win 1 off 0; last base writer load at 3; writers back to the last injecting one: load@3; shadow writes of r0 per rep: add=5 mad=4 mul=5 mulhi=1 or=1 rotl=2 rotr=6 shfl=3 sub=6 xor=4
[2026-10-07T18:37:06Z] load site instr 7: src r2 win 0 off 0; last base writer load at 4; writers back to the last injecting one: load@4; shadow writes of r2 per rep: add=5 mad=4 mul=3 mulhi=1 or=2 rotl=2 rotr=3 shfl=1 sub=3 xor=9
[2026-10-07T18:37:06Z] load site instr 15: src r4 win 2 off 0; last base writer mad at 12; writers back to the last injecting one: mad@12; shadow writes of r4 per rep: add=6 mad=2 mul=4 mulhi=7 or=1 rotl=6 rotr=5 shfl=3 sub=3 xor=5
[2026-10-07T18:37:06Z] load site instr 17: src r3 win 2 off 2; last base writer load at 15; writers back to the last injecting one: load@15; shadow writes of r3 per rep: add=4 mad=1 mul=1 mulhi=2 or=1 rotl=4 rotr=2 shfl=4 sub=1 xor=3
[2026-10-07T18:37:06Z] load site instr 23: src r0 win 0 off 0; last base writer xor at 13; writers back to the last injecting one: xor@13; shadow writes of r0 per rep: add=5 mad=4 mul=5 mulhi=1 or=1 rotl=2 rotr=6 shfl=3 sub=6 xor=4
[2026-10-07T18:37:06Z] load site instr 24: src r7 win 1 off 0; last base writer shfl at 16; writers back to the last injecting one: shfl@16; shadow writes of r7 per rep: add=2 mad=4 mul=3 mulhi=1 or=1 rotl=5 rotr=4 shfl=2 sub=1 xor=4
[2026-10-07T18:37:06Z] load site instr 27: src r5 win 2 off 3; last base writer load at 24; writers back to the last injecting one: load@24; shadow writes of r5 per rep: add=3 mad=5 mul=1 mulhi=2 or=1 rotl=3 rotr=2 shfl=6 sub=1 xor=5
[2026-10-07T18:37:06Z] load site instr 29: src r2 win 2 off 1; last base writer load at 27; writers back to the last injecting one: load@27; shadow writes of r2 per rep: add=5 mad=4 mul=3 mulhi=1 or=2 rotl=2 rotr=3 shfl=1 sub=3 xor=9
[2026-10-07T18:37:06Z] load site instr 31: src r6 win 1 off 0; last base writer xor at 30; writers back to the last injecting one: xor@30; shadow writes of r6 per rep: add=5 mad=4 mul=5 mulhi=2 or=1 rotl=3 rotr=3 shfl=1 sub=4 xor=4
[2026-10-07T18:37:06Z] load site instr 35: src r7 win 0 off 0; last base writer rotl at 33; writers back to the last injecting one: rotl@33 shfl@16; shadow writes of r7 per rep: add=2 mad=4 mul=3 mulhi=1 or=1 rotl=5 rotr=4 shfl=2 sub=1 xor=4
[2026-10-07T18:37:06Z] load site instr 41: src r1 win 1 off 0; last base writer add at 40; writers back to the last injecting one: add@40; shadow writes of r1 per rep: add=3 mad=3 mul=3 mulhi=2 or=3 rotl=5 rotr=2 shfl=5 sub=2 xor=5
[2026-10-07T18:37:06Z] load site instr 44: src r5 win 2 off 1; last base writer load at 35; writers back to the last injecting one: load@35; shadow writes of r5 per rep: add=3 mad=5 mul=1 mulhi=2 or=1 rotl=3 rotr=2 shfl=6 sub=1 xor=5
[2026-10-07T18:37:06Z] load site instr 49: src r4 win 0 off 0; last base writer load at 44; writers back to the last injecting one: load@44; shadow writes of r4 per rep: add=6 mad=2 mul=4 mulhi=7 or=1 rotl=6 rotr=5 shfl=3 sub=3 xor=5
[2026-10-07T18:37:06Z] load site instr 61: src r3 win 2 off 1; last base writer xor at 58; writers back to the last injecting one: xor@58; shadow writes of r3 per rep: add=4 mad=1 mul=1 mulhi=2 or=1 rotl=4 rotr=2 shfl=4 sub=1 xor=3
[2026-10-07T18:37:06Z] load site instr 63: src r4 win 1 off 0; last base writer shfl at 62; writers back to the last injecting one: shfl@62; shadow writes of r4 per rep: add=6 mad=2 mul=4 mulhi=7 or=1 rotl=6 rotr=5 shfl=3 sub=3 xor=5
[2026-10-07T18:37:06Z] window layer: site item windows (first, items) (0x800000,2^23) (0x0,2^23) (0x0,2^24) (0x0,2^22) (0x800000,2^22) (0x0,2^24) (0x0,2^23) (0xc00000,2^22) (0x400000,2^22) (0x0,2^23) (0x0,2^24) (0x0,2^23) (0x400000,2^22) (0x0,2^24) (0x400000,2^22) (0x0,2^23); expected reads per item by quarter 8.583 12.398 4.768 4.768 (flat uniform 7.629)
[2026-10-07T18:37:09Z] 31250 warps (1000000 nonces) interpreted in 3.4 s (3.514 ms per warp per thread); Epoch::hash_warp agreement on 95 warps: 0 mismatches
[2026-10-07T18:37:09Z] p100211-attack-f8 distinct lines per hash: n 1000000 min 1013 p1 1023 median 1024 p99 1024 max 1024 mean 1023.8706
[2026-10-07T18:37:09Z] p100211-attack-f8 distinct items per hash: n 1000000 min 127 p1 128 median 128 p99 128 max 128 mean 127.9994
[2026-10-07T18:37:09Z] p100211-attack-f8 distinct lines per warp: n 31250 min 32580 p1 32604 median 32636 p99 32664 max 32682 mean 32635.6451
[2026-10-07T18:37:09Z] p100211-attack-f8 distinct items per warp: n 31250 min 4089 p1 4093 median 4096 p99 4096 max 4096 mean 4095.4165
[2026-10-07T18:37:09Z] uniform expectation: distinct lines per hash 1023.875 of 1024 reads, per warp 32640.3 of 32768 reads (2^22 lines); distinct items per hash 127.9995 of 128, per warp 4095.50 of 4096 (2^24 items)
[2026-10-07T18:37:09Z] positions: largest hi16-bucket (256 items) share 0.0193% at p89 (iteration 5, site 9, instr 31); windowed expectation 0.0031%; largest saturated-source share 0.0000% at p0 (site 0, instr 3)
[2026-10-07T18:37:09Z] acceptance-style (2,048 evaluations, the rule's sample size): the most repeated address at one position 2 of 2048 at p9 (site 9, instr 31); saturated sources at one position 0 of 2048 at p0 (site 0, instr 3); uniform expectation: repeats 1 to 2, saturated 0
[2026-10-07T18:37:09Z] p100211-attack-f8 item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 1238 (bin 4539781) z_max +445.44 min 0 z_min -2.76 chi2/dof 2.36870 chi2_z +3964.17 top0.1% 0.34137% top0.5% 1.46293% top1% 2.74616%
[2026-10-07T18:37:09Z] p100211-attack-f8 item histogram against the window density: full 2^24 chi2/dof 1.04292 largest 1238 (item 0x454585) at +348.08 sigma smallest at -3.52; buckets64 chi2/dof 1.03150 largest 1989 (bucket 70934) at +42.44 sigma smallest at -4.02
[2026-10-07T18:37:10Z] p100211-attack-f8 WINDOWED CONTROL item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 37 (bin 6537897) z_max +10.63 min 0 z_min -2.76 chi2/dof 2.31214 chi2_z +3800.36 top0.1% 0.31534% top0.5% 1.42196% top1% 2.69721%
[2026-10-07T18:37:10Z] p100211-attack-f8 WINDOWED CONTROL against the window density: full chi2/dof 1.00002 largest +6.99 sigma; buckets64 chi2/dof 0.99932 largest +4.68 smallest -4.45
[2026-10-07T18:37:10Z] 6-sigma p100211-attack-f8 items buckets64 against the window density: largest bucket +42.44 sigma, smallest -4.02 -> FLAGGED (beyond 6 sigma)
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (windowed control): f 0.1% S_f 0.34137% E_f(control) 0.31534% X_f +0.02603% X_f/f +0.2603 -> no hot set
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (windowed control): f 0.5% S_f 1.46293% E_f(control) 1.42196% X_f +0.04096% X_f/f +0.0819 -> no hot set
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (windowed control): f 1.0% S_f 2.74616% E_f(control) 2.69721% X_f +0.04895% X_f/f +0.0489 -> no hot set
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (windowed control): verdict clear
[2026-10-07T18:37:10Z] p100211-attack-f8 FLAT CONTROL item histogram: bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 27 (bin 9914800) z_max +7.01 min 0 z_min -2.76 chi2/dof 1.00012 chi2_z +0.35 top0.1% 0.24321% top0.5% 1.09954% top1% 2.08777%
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (flat control, the auditor's first view): f 0.1% S_f 0.34137% E_f(control) 0.24321% X_f +0.09815% X_f/f +0.9815 -> no hot set
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (flat control, the auditor's first view): f 0.5% S_f 1.46293% E_f(control) 1.09954% X_f +0.36339% X_f/f +0.7268 -> no hot set
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (flat control, the auditor's first view): f 1.0% S_f 2.74616% E_f(control) 2.08777% X_f +0.65839% X_f/f +0.6584 -> no hot set
[2026-10-07T18:37:10Z] hot-set p100211-attack-f8 items (flat control, the auditor's first view): verdict clear
[2026-10-07T18:37:10Z] ratio p100211-attack-f8: top 0.1% / 0.5% / 1% share over the WINDOW-MODEL control 1.0825x / 1.0288x / 1.0181x (gate 1.2x at 0.1%: within); over the FLAT control 1.4036x / 1.3305x / 1.3154x
[2026-10-07T18:37:10Z] hottest item p100211-attack-f8 (0x454585): 1238 reads (0.0010% of all); predicted source site-instr4:r0:one-zero-bit:last-writer-load@3
[2026-10-07T18:37:10Z] program p100211-attack-f8 DONE: 1000000 nonces; 6-sigma (windowed) FLAGGED; hot-set clear; verdict FLAGGED
[2026-10-07T18:37:11Z] attack-f8 0.1.0 (igneum-pow 4); args ["warps", "--program", "100629", "--nonces", "1000000", "--threads", "32", "--out", "/srv/builds/igneum-wt-adv-accept/adv/live", "--validate", "sample"]
[2026-10-07T18:37:11Z] census warps: programs [100629] day 20730 nonces 1000000 threads 32 plant none validate sample check_every 997 diag false
[2026-10-07T18:37:11Z] day 20730: cache filled in 0.66 s, fnv 448274a57f508cbc
[2026-10-07T18:37:14Z] table: 16777216 items derived with their 8 lines in 2.7 s; library comparison sample (0 mismatches)
[2026-10-07T18:37:19Z] program p100629-attack-f8: epoch seed 68bfab501dbcaf5685d1cfea27ebe7d0bbeb3aeae03591561bd3eabc3d6510a2 era seed 24152e11d7c8f0e60319ba0ca3b83ace33a2f22cffd58ad2a8168dc4283352b5 id 37c849d9741c5994 attempt 0 class mx8-era31b72e61+sh256x27 op mix load=16 add=10 mad=8 mul=6 rotl=5 or=4 shfl=4 xor=4 mulhi=3 rotr=2 sub=2; era stride mul 0xc8c38905 rot 3 interleave [3, 4, 5, 10] windows 4:1:1 5:1:1 9:2:0 13:1:1 15:1:1 16:2:3 17:0:0 18:0:0 19:2:2 33:2:0 34:1:1 38:2:0 40:1:0 47:0:0 51:0:0 54:0:0
[2026-10-07T18:37:19Z] load site instr 4: src r0 win 1 off 1; last base writer mad at 2; writers back to the last injecting one: mad@2; shadow writes of r0 per rep: add=3 mad=3 mul=3 mulhi=2 or=2 rotl=4 rotr=2 sub=5 xor=5
[2026-10-07T18:37:19Z] load site instr 5: src r1 win 1 off 1; last base writer load at 4; writers back to the last injecting one: load@4; shadow writes of r1 per rep: add=9 mad=1 mul=2 mulhi=4 or=4 rotl=1 rotr=2 shfl=1 sub=2 xor=3
[2026-10-07T18:37:19Z] load site instr 9: src r1 win 2 off 0; last base writer add at 8; writers back to the last injecting one: add@8; shadow writes of r1 per rep: add=9 mad=1 mul=2 mulhi=4 or=4 rotl=1 rotr=2 shfl=1 sub=2 xor=3
[2026-10-07T18:37:19Z] load site instr 13: src r2 win 1 off 1; last base writer load at 9; writers back to the last injecting one: load@9; shadow writes of r2 per rep: add=1 mad=3 mul=2 mulhi=1 or=3 rotl=5 rotr=3 shfl=3 sub=4 xor=8
[2026-10-07T18:37:19Z] load site instr 15: src r1 win 1 off 1; last base writer load at 13; writers back to the last injecting one: load@13; shadow writes of r1 per rep: add=9 mad=1 mul=2 mulhi=4 or=4 rotl=1 rotr=2 shfl=1 sub=2 xor=3
[2026-10-07T18:37:19Z] load site instr 16: src r3 win 2 off 3; last base writer mad at 12; writers back to the last injecting one: mad@12; shadow writes of r3 per rep: add=5 mad=3 mul=6 mulhi=3 or=1 rotl=3 rotr=5 shfl=6 sub=3 xor=5
[2026-10-07T18:37:19Z] load site instr 17: src r2 win 0 off 0; last base writer load at 15; writers back to the last injecting one: load@15; shadow writes of r2 per rep: add=1 mad=3 mul=2 mulhi=1 or=3 rotl=5 rotr=3 shfl=3 sub=4 xor=8
[2026-10-07T18:37:19Z] load site instr 18: src r6 win 0 off 0; last base writer load at 16; writers back to the last injecting one: load@16; shadow writes of r6 per rep: add=4 mad=2 mul=6 mulhi=3 or=3 rotl=4 rotr=2 shfl=3 sub=1 xor=3
[2026-10-07T18:37:19Z] load site instr 19: src r5 win 2 off 2; last base writer load at 17; writers back to the last injecting one: load@17; shadow writes of r5 per rep: add=5 mad=2 mul=3 mulhi=5 or=1 rotl=2 rotr=3 shfl=5 sub=6 xor=1
[2026-10-07T18:37:19Z] load site instr 33: src r5 win 2 off 0; last base writer add at 31; writers back to the last injecting one: add@31; shadow writes of r5 per rep: add=5 mad=2 mul=3 mulhi=5 or=1 rotl=2 rotr=3 shfl=5 sub=6 xor=1
[2026-10-07T18:37:19Z] load site instr 34: src r1 win 1 off 1; last base writer xor at 27; writers back to the last injecting one: xor@27; shadow writes of r1 per rep: add=9 mad=1 mul=2 mulhi=4 or=4 rotl=1 rotr=2 shfl=1 sub=2 xor=3
[2026-10-07T18:37:19Z] load site instr 38: src r6 win 2 off 0; last base writer add at 35; writers back to the last injecting one: add@35; shadow writes of r6 per rep: add=4 mad=2 mul=6 mulhi=3 or=3 rotl=4 rotr=2 shfl=3 sub=1 xor=3
[2026-10-07T18:37:19Z] load site instr 40: src r3 win 1 off 0; last base writer load at 38; writers back to the last injecting one: load@38; shadow writes of r3 per rep: add=5 mad=3 mul=6 mulhi=3 or=1 rotl=3 rotr=5 shfl=6 sub=3 xor=5
[2026-10-07T18:37:19Z] load site instr 47: src r1 win 0 off 0; last base writer rotl at 44; writers back to the last injecting one: rotl@44 xor@27; shadow writes of r1 per rep: add=9 mad=1 mul=2 mulhi=4 or=4 rotl=1 rotr=2 shfl=1 sub=2 xor=3
[2026-10-07T18:37:19Z] load site instr 51: src r7 win 0 off 0; last base writer load at 40; writers back to the last injecting one: load@40; shadow writes of r7 per rep: add=7 mad=2 mul=1 mulhi=4 or=2 rotr=5 shfl=2 sub=2 xor=5
[2026-10-07T18:37:19Z] load site instr 54: src r3 win 0 off 0; last base writer load at 51; writers back to the last injecting one: load@51; shadow writes of r3 per rep: add=5 mad=3 mul=6 mulhi=3 or=1 rotl=3 rotr=5 shfl=6 sub=3 xor=5
[2026-10-07T18:37:19Z] window layer: site item windows (first, items) (0x800000,2^23) (0x800000,2^23) (0x0,2^22) (0x800000,2^23) (0x800000,2^23) (0xc00000,2^22) (0x0,2^24) (0x0,2^24) (0x800000,2^22) (0x0,2^22) (0x800000,2^23) (0x0,2^22) (0x0,2^23) (0x0,2^24) (0x0,2^24) (0x0,2^24); expected reads per item by quarter 9.060 3.338 9.060 9.060 (flat uniform 7.629)
[2026-10-07T18:37:23Z] 31250 warps (1000000 nonces) interpreted in 3.8 s (3.852 ms per warp per thread); Epoch::hash_warp agreement on 95 warps: 0 mismatches
[2026-10-07T18:37:23Z] p100629-attack-f8 distinct lines per hash: n 1000000 min 1014 p1 1023 median 1024 p99 1024 max 1024 mean 1023.8713
[2026-10-07T18:37:23Z] p100629-attack-f8 distinct items per hash: n 1000000 min 127 p1 128 median 128 p99 128 max 128 mean 127.9995
[2026-10-07T18:37:23Z] p100629-attack-f8 distinct lines per warp: n 31250 min 32565 p1 32605 median 32636 p99 32664 max 32681 mean 32635.9486
[2026-10-07T18:37:23Z] p100629-attack-f8 distinct items per warp: n 31250 min 4090 p1 4093 median 4096 p99 4096 max 4096 mean 4095.4480
[2026-10-07T18:37:23Z] uniform expectation: distinct lines per hash 1023.875 of 1024 reads, per warp 32640.3 of 32768 reads (2^22 lines); distinct items per hash 127.9995 of 128, per warp 4095.50 of 4096 (2^24 items)
[2026-10-07T18:37:23Z] positions: largest hi16-bucket (256 items) share 0.0104% at p121 (iteration 7, site 9, instr 33); windowed expectation 0.0061%; largest saturated-source share 0.0001% at p5 (site 5, instr 16)
[2026-10-07T18:37:23Z] acceptance-style (2,048 evaluations, the rule's sample size): the most repeated address at one position 2 of 2048 at p4 (site 4, instr 15); saturated sources at one position 0 of 2048 at p0 (site 0, instr 4); uniform expectation: repeats 1 to 2, saturated 0
[2026-10-07T18:37:23Z] p100629-attack-f8 item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 30 (bin 8487097) z_max +8.10 min 0 z_min -2.76 chi2/dof 1.81617 chi2_z +2363.89 top0.1% 0.27111% top0.5% 1.22433% top1% 2.32916%
[2026-10-07T18:37:23Z] p100629-attack-f8 item histogram against the window density: full 2^24 chi2/dof 1.01006 largest 30 (item 0x8180b9) at +6.96 sigma smallest at -3.01; buckets64 chi2/dof 1.01223 largest 296 (bucket 98030) at +5.64 sigma smallest at -4.69
[2026-10-07T18:37:23Z] p100629-attack-f8 WINDOWED CONTROL item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 29 (bin 9724837) z_max +7.74 min 0 z_min -2.76 chi2/dof 1.80496 chi2_z +2331.41 top0.1% 0.26813% top0.5% 1.21510% top1% 2.31449%
[2026-10-07T18:37:23Z] p100629-attack-f8 WINDOWED CONTROL against the window density: full chi2/dof 1.00037 largest +6.93 sigma; buckets64 chi2/dof 0.99550 largest +4.74 smallest -4.31
[2026-10-07T18:37:23Z] 6-sigma p100629-attack-f8 items buckets64 against the window density: largest bucket +5.64 sigma, smallest -4.69 -> within 6 sigma
[2026-10-07T18:37:23Z] hot-set p100629-attack-f8 items (windowed control): f 0.1% S_f 0.27111% E_f(control) 0.26813% X_f +0.00298% X_f/f +0.0298 -> no hot set
[2026-10-07T18:37:23Z] hot-set p100629-attack-f8 items (windowed control): f 0.5% S_f 1.22433% E_f(control) 1.21510% X_f +0.00923% X_f/f +0.0185 -> no hot set
[2026-10-07T18:37:23Z] hot-set p100629-attack-f8 items (windowed control): f 1.0% S_f 2.32916% E_f(control) 2.31449% X_f +0.01468% X_f/f +0.0147 -> no hot set
[2026-10-07T18:37:23Z] hot-set p100629-attack-f8 items (windowed control): verdict clear
[2026-10-07T18:37:24Z] p100629-attack-f8 FLAT CONTROL item histogram: bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 26 (bin 1359050) z_max +6.65 min 0 z_min -2.76 chi2/dof 1.00082 chi2_z +2.36 top0.1% 0.24290% top0.5% 1.09909% top1% 2.08683%
[2026-10-07T18:37:24Z] hot-set p100629-attack-f8 items (flat control, the auditor's first view): f 0.1% S_f 0.27111% E_f(control) 0.24290% X_f +0.02821% X_f/f +0.2821 -> no hot set
[2026-10-07T18:37:24Z] hot-set p100629-attack-f8 items (flat control, the auditor's first view): f 0.5% S_f 1.22433% E_f(control) 1.09909% X_f +0.12524% X_f/f +0.2505 -> no hot set
[2026-10-07T18:37:24Z] hot-set p100629-attack-f8 items (flat control, the auditor's first view): f 1.0% S_f 2.32916% E_f(control) 2.08683% X_f +0.24233% X_f/f +0.2423 -> no hot set
[2026-10-07T18:37:24Z] hot-set p100629-attack-f8 items (flat control, the auditor's first view): verdict clear
[2026-10-07T18:37:24Z] ratio p100629-attack-f8: top 0.1% / 0.5% / 1% share over the WINDOW-MODEL control 1.0111x / 1.0076x / 1.0063x (gate 1.2x at 0.1%: within); over the FLAT control 1.1161x / 1.1139x / 1.1161x
[2026-10-07T18:37:24Z] hottest item p100629-attack-f8 (0x8180b9): 30 reads (0.0000% of all); predicted source none
[2026-10-07T18:37:24Z] program p100629-attack-f8 DONE: 1000000 nonces; 6-sigma (windowed) clear; hot-set clear; verdict PASS (no test fired)
[2026-10-07T18:37:24Z] attack-f8 0.1.0 (igneum-pow 4); args ["warps", "--program", "100064", "--nonces", "1000000", "--threads", "32", "--out", "/srv/builds/igneum-wt-adv-accept/adv/live", "--validate", "sample"]
[2026-10-07T18:37:24Z] census warps: programs [100064] day 20730 nonces 1000000 threads 32 plant none validate sample check_every 997 diag false
[2026-10-07T18:37:25Z] day 20730: cache filled in 0.61 s, fnv 448274a57f508cbc
[2026-10-07T18:37:28Z] table: 16777216 items derived with their 8 lines in 2.4 s; library comparison sample (0 mismatches)
[2026-10-07T18:37:32Z] program p100064-attack-f8: epoch seed 96bc342fb24875e10a17d093c27cdfe28c4293bfb4e0c96bc302de885e195a89 era seed a860aa82556a6f86d2b45818c33f274427428086c50c5d5b3f7101128c289bdf id 2442abf9d56f6611 attempt 2 class mx8-eraa210d1bd+sh256x27 op mix load=16 sub=9 add=7 xor=7 mul=5 rotl=5 mad=4 rotr=4 shfl=4 or=2 mulhi=1; era stride mul 0x82b69d5d rot 9 interleave [7, 8, 9, 12] windows 2:1:0 9:0:0 10:1:0 15:1:1 16:0:0 20:2:2 21:0:0 25:0:0 28:1:1 30:0:0 31:0:0 33:2:1 34:1:1 35:0:0 41:1:0 51:1:0
[2026-10-07T18:37:32Z] load site instr 2: src r5 win 1 off 0; last base writer rotl at 0; writers back to the last injecting one: rotl@0 xor@62; shadow writes of r5 per rep: add=9 mad=7 mul=1 mulhi=2 or=1 rotl=6 rotr=2 shfl=2 sub=3 xor=6
[2026-10-07T18:37:32Z] load site instr 9: src r1 win 0 off 0; last base writer load at 2; writers back to the last injecting one: load@2; shadow writes of r1 per rep: add=4 mad=2 mul=5 mulhi=1 or=3 rotr=4 shfl=4 sub=3 xor=1
[2026-10-07T18:37:32Z] load site instr 10: src r6 win 1 off 0; last base writer sub at 8; writers back to the last injecting one: sub@8; shadow writes of r6 per rep: add=8 mad=5 mul=2 mulhi=1 or=3 rotl=2 rotr=5 shfl=1 sub=2 xor=1
[2026-10-07T18:37:32Z] load site instr 15: src r6 win 1 off 1; last base writer shfl at 14; writers back to the last injecting one: shfl@14; shadow writes of r6 per rep: add=8 mad=5 mul=2 mulhi=1 or=3 rotl=2 rotr=5 shfl=1 sub=2 xor=1
[2026-10-07T18:37:32Z] load site instr 16: src r7 win 0 off 0; last base writer shfl at 13; writers back to the last injecting one: shfl@13; shadow writes of r7 per rep: add=7 mul=5 mulhi=1 rotl=2 rotr=7 shfl=5 sub=1 xor=4
[2026-10-07T18:37:32Z] load site instr 20: src r2 win 2 off 2; last base writer load at 16; writers back to the last injecting one: load@16; shadow writes of r2 per rep: add=6 mad=1 mul=2 mulhi=1 or=1 rotl=2 rotr=3 shfl=5 sub=5 xor=1
[2026-10-07T18:37:32Z] load site instr 21: src r6 win 0 off 0; last base writer load at 20; writers back to the last injecting one: load@20; shadow writes of r6 per rep: add=8 mad=5 mul=2 mulhi=1 or=3 rotl=2 rotr=5 shfl=1 sub=2 xor=1
[2026-10-07T18:37:32Z] load site instr 25: src r5 win 0 off 0; last base writer load at 10; writers back to the last injecting one: load@10; shadow writes of r5 per rep: add=9 mad=7 mul=1 mulhi=2 or=1 rotl=6 rotr=2 shfl=2 sub=3 xor=6
[2026-10-07T18:37:32Z] load site instr 28: src r2 win 1 off 1; last base writer xor at 26; writers back to the last injecting one: xor@26; shadow writes of r2 per rep: add=6 mad=1 mul=2 mulhi=1 or=1 rotl=2 rotr=3 shfl=5 sub=5 xor=1
[2026-10-07T18:37:32Z] load site instr 30: src r7 win 0 off 0; last base writer load at 28; writers back to the last injecting one: load@28; shadow writes of r7 per rep: add=7 mul=5 mulhi=1 rotl=2 rotr=7 shfl=5 sub=1 xor=4
[2026-10-07T18:37:32Z] load site instr 31: src r4 win 0 off 0; last base writer load at 25; writers back to the last injecting one: load@25; shadow writes of r4 per rep: add=6 mad=3 mul=2 mulhi=4 rotl=4 rotr=4 shfl=3 sub=6 xor=4
[2026-10-07T18:37:32Z] load site instr 33: src r0 win 2 off 1; last base writer mad at 27; writers back to the last injecting one: mad@27; shadow writes of r0 per rep: add=7 mad=1 mul=4 mulhi=2 or=2 rotl=2 rotr=4 shfl=2 sub=3 xor=2
[2026-10-07T18:37:32Z] load site instr 34: src r7 win 1 off 1; last base writer load at 33; writers back to the last injecting one: load@33; shadow writes of r7 per rep: add=7 mul=5 mulhi=1 rotl=2 rotr=7 shfl=5 sub=1 xor=4
[2026-10-07T18:37:32Z] load site instr 35: src r5 win 0 off 0; last base writer load at 34; writers back to the last injecting one: load@34; shadow writes of r5 per rep: add=9 mad=7 mul=1 mulhi=2 or=1 rotl=6 rotr=2 shfl=2 sub=3 xor=6
[2026-10-07T18:37:32Z] load site instr 41: src r2 win 1 off 0; last base writer add at 37; writers back to the last injecting one: add@37; shadow writes of r2 per rep: add=6 mad=1 mul=2 mulhi=1 or=1 rotl=2 rotr=3 shfl=5 sub=5 xor=1
[2026-10-07T18:37:32Z] load site instr 51: src r0 win 1 off 0; last base writer rotl at 46; writers back to the last injecting one: rotl@46 mad@27; shadow writes of r0 per rep: add=7 mad=1 mul=4 mulhi=2 or=2 rotl=2 rotr=4 shfl=2 sub=3 xor=2
[2026-10-07T18:37:32Z] window layer: site item windows (first, items) (0x0,2^23) (0x0,2^24) (0x0,2^23) (0x800000,2^23) (0x0,2^24) (0x800000,2^22) (0x0,2^24) (0x0,2^24) (0x800000,2^23) (0x0,2^24) (0x0,2^24) (0x400000,2^22) (0x800000,2^23) (0x0,2^24) (0x0,2^23) (0x0,2^23); expected reads per item by quarter 7.153 9.060 8.106 6.199 (flat uniform 7.629)
[2026-10-07T18:37:36Z] 31250 warps (1000000 nonces) interpreted in 3.7 s (3.777 ms per warp per thread); Epoch::hash_warp agreement on 95 warps: 0 mismatches
[2026-10-07T18:37:36Z] p100064-attack-f8 distinct lines per hash: n 1000000 min 1014 p1 1023 median 1024 p99 1024 max 1024 mean 1023.8712
[2026-10-07T18:37:36Z] p100064-attack-f8 distinct items per hash: n 1000000 min 127 p1 128 median 128 p99 128 max 128 mean 127.9995
[2026-10-07T18:37:36Z] p100064-attack-f8 distinct lines per warp: n 31250 min 32571 p1 32605 median 32637 p99 32664 max 32682 mean 32636.2260
[2026-10-07T18:37:36Z] p100064-attack-f8 distinct items per warp: n 31250 min 4090 p1 4093 median 4096 p99 4096 max 4096 mean 4095.4805
[2026-10-07T18:37:36Z] uniform expectation: distinct lines per hash 1023.875 of 1024 reads, per warp 32640.3 of 32768 reads (2^22 lines); distinct items per hash 127.9995 of 128, per warp 4095.50 of 4096 (2^24 items)
[2026-10-07T18:37:36Z] positions: largest hi16-bucket (256 items) share 0.0484% at p43 (iteration 2, site 11, instr 33); windowed expectation 0.0061%; largest saturated-source share 0.0000% at p0 (site 0, instr 2)
[2026-10-07T18:37:36Z] acceptance-style (2,048 evaluations, the rule's sample size): the most repeated address at one position 2 of 2048 at p5 (site 5, instr 20); saturated sources at one position 0 of 2048 at p0 (site 0, instr 2); uniform expectation: repeats 1 to 2, saturated 0
[2026-10-07T18:37:36Z] p100064-attack-f8 item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 54 (bin 4194344) z_max +16.79 min 0 z_min -2.76 chi2/dof 1.26520 chi2_z +768.11 top0.1% 0.29988% top0.5% 1.27974% top1% 2.37611%
[2026-10-07T18:37:36Z] p100064-attack-f8 item histogram against the window density: full 2^24 chi2/dof 1.10304 largest 54 (item 0x400028) at +14.93 sigma smallest at -3.01; buckets64 chi2/dof 7.60238 largest 2441 (bucket 65536) at +77.29 sigma smallest at -7.51
[2026-10-07T18:37:36Z] p100064-attack-f8 WINDOWED CONTROL item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 27 (bin 5834067) z_max +7.01 min 0 z_min -2.76 chi2/dof 1.14878 chi2_z +430.91 top0.1% 0.25551% top0.5% 1.15143% top1% 2.18725%
[2026-10-07T18:37:36Z] p100064-attack-f8 WINDOWED CONTROL against the window density: full chi2/dof 0.99986 largest +6.30 sigma; buckets64 chi2/dof 1.00273 largest +4.68 smallest -4.35
[2026-10-07T18:37:36Z] 6-sigma p100064-attack-f8 items buckets64 against the window density: largest bucket +77.29 sigma, smallest -7.51 -> FLAGGED (beyond 6 sigma)
[2026-10-07T18:37:36Z] hot-set p100064-attack-f8 items (windowed control): f 0.1% S_f 0.29988% E_f(control) 0.25551% X_f +0.04437% X_f/f +0.4437 -> no hot set
[2026-10-07T18:37:36Z] hot-set p100064-attack-f8 items (windowed control): f 0.5% S_f 1.27974% E_f(control) 1.15143% X_f +0.12831% X_f/f +0.2566 -> no hot set
[2026-10-07T18:37:36Z] hot-set p100064-attack-f8 items (windowed control): f 1.0% S_f 2.37611% E_f(control) 2.18725% X_f +0.18885% X_f/f +0.1889 -> no hot set
[2026-10-07T18:37:36Z] hot-set p100064-attack-f8 items (windowed control): verdict clear
[2026-10-07T18:37:37Z] p100064-attack-f8 FLAT CONTROL item histogram: bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 28 (bin 3756411) z_max +7.37 min 0 z_min -2.76 chi2/dof 1.00062 chi2_z +1.80 top0.1% 0.24291% top0.5% 1.09902% top1% 2.08725%
[2026-10-07T18:37:37Z] hot-set p100064-attack-f8 items (flat control, the auditor's first view): f 0.1% S_f 0.29988% E_f(control) 0.24291% X_f +0.05697% X_f/f +0.5697 -> no hot set
[2026-10-07T18:37:37Z] hot-set p100064-attack-f8 items (flat control, the auditor's first view): f 0.5% S_f 1.27974% E_f(control) 1.09902% X_f +0.18072% X_f/f +0.3614 -> no hot set
[2026-10-07T18:37:37Z] hot-set p100064-attack-f8 items (flat control, the auditor's first view): f 1.0% S_f 2.37611% E_f(control) 2.08725% X_f +0.28886% X_f/f +0.2889 -> no hot set
[2026-10-07T18:37:37Z] hot-set p100064-attack-f8 items (flat control, the auditor's first view): verdict clear
[2026-10-07T18:37:37Z] ratio p100064-attack-f8: top 0.1% / 0.5% / 1% share over the WINDOW-MODEL control 1.1737x / 1.1114x / 1.0863x (gate 1.2x at 0.1%: within); over the FLAT control 1.2346x / 1.1644x / 1.1384x
[2026-10-07T18:37:37Z] hottest item p100064-attack-f8 (0x400028): 54 reads (0.0000% of all); predicted source none
[2026-10-07T18:37:37Z] program p100064-attack-f8 DONE: 1000000 nonces; 6-sigma (windowed) FLAGGED; hot-set clear; verdict FLAGGED
[2026-10-07T18:37:37Z] attack-f8 0.1.0 (igneum-pow 4); args ["warps", "--program", "100767", "--nonces", "1000000", "--threads", "32", "--out", "/srv/builds/igneum-wt-adv-accept/adv/live", "--validate", "sample"]
[2026-10-07T18:37:37Z] census warps: programs [100767] day 20730 nonces 1000000 threads 32 plant none validate sample check_every 997 diag false
[2026-10-07T18:37:38Z] day 20730: cache filled in 0.65 s, fnv 448274a57f508cbc
[2026-10-07T18:37:40Z] table: 16777216 items derived with their 8 lines in 2.3 s; library comparison sample (0 mismatches)
[2026-10-07T18:37:45Z] program p100767-attack-f8: epoch seed 74484b391756fc49568cdd716bcfd839a55d31a45633c223c65c7f2ab4617fd4 era seed 7a65a05391dcd87b8fdaf7f74813aa17c6e13f7c56a03da2aaa6176987ef038b id 9d68e6286fc817d4 attempt 2 class mx8-era763e5847+sh256x27 op mix load=16 add=8 mad=6 mul=6 rotl=6 mulhi=5 xor=5 rotr=4 shfl=4 or=2 sub=2; era stride mul 0x54387173 rot 28 interleave [4, 11, 13, 15] windows 6:1:0 9:0:0 11:2:1 12:0:0 13:1:1 19:1:1 23:2:0 32:2:0 36:1:1 39:1:1 41:0:0 45:1:0 50:2:3 58:0:0 59:0:0 63:1:0
[2026-10-07T18:37:45Z] load site instr 6: src r6 win 1 off 0; last base writer mad at 4; writers back to the last injecting one: mad@4; shadow writes of r6 per rep: add=5 mad=7 mul=1 mulhi=1 rotl=3 rotr=2 shfl=3 sub=1 xor=6
[2026-10-07T18:37:45Z] load site instr 9: src r7 win 0 off 0; last base writer load at 6; writers back to the last injecting one: load@6; shadow writes of r7 per rep: add=5 mad=5 mul=4 mulhi=5 rotl=1 rotr=5 shfl=3 sub=4 xor=5
[2026-10-07T18:37:45Z] load site instr 11: src r5 win 2 off 1; last base writer xor at 5; writers back to the last injecting one: xor@5; shadow writes of r5 per rep: add=10 mul=5 mulhi=1 rotl=3 rotr=1 shfl=4 sub=3 xor=4
[2026-10-07T18:37:45Z] load site instr 12: src r3 win 0 off 0; last base writer load at 9; writers back to the last injecting one: load@9; shadow writes of r3 per rep: mad=6 mul=2 mulhi=4 or=1 rotl=4 rotr=2 xor=4
[2026-10-07T18:37:45Z] load site instr 13: src r2 win 1 off 1; last base writer load at 12; writers back to the last injecting one: load@12; shadow writes of r2 per rep: add=3 mad=2 mul=4 mulhi=2 or=6 rotl=1 rotr=1 shfl=5 sub=2 xor=4
[2026-10-07T18:37:45Z] load site instr 19: src r4 win 1 off 1; last base writer add at 7; writers back to the last injecting one: add@7; shadow writes of r4 per rep: add=5 mad=5 mul=4 mulhi=5 or=1 rotl=4 rotr=5 shfl=3 sub=5 xor=2
[2026-10-07T18:37:45Z] load site instr 23: src r6 win 2 off 0; last base writer mad at 22; writers back to the last injecting one: mad@22; shadow writes of r6 per rep: add=5 mad=7 mul=1 mulhi=1 rotl=3 rotr=2 shfl=3 sub=1 xor=6
[2026-10-07T18:37:45Z] load site instr 32: src r0 win 2 off 0; last base writer xor at 20; writers back to the last injecting one: xor@20; shadow writes of r0 per rep: add=7 mad=1 mul=3 mulhi=3 or=3 rotl=3 rotr=2 shfl=5 sub=2 xor=3
[2026-10-07T18:37:45Z] load site instr 36: src r7 win 1 off 1; last base writer rotr at 31; writers back to the last injecting one: rotr@31 load@6; shadow writes of r7 per rep: add=5 mad=5 mul=4 mulhi=5 rotl=1 rotr=5 shfl=3 sub=4 xor=5
[2026-10-07T18:37:45Z] load site instr 39: src r7 win 1 off 1; last base writer rotl at 38; writers back to the last injecting one: rotl@38 add@37; shadow writes of r7 per rep: add=5 mad=5 mul=4 mulhi=5 rotl=1 rotr=5 shfl=3 sub=4 xor=5
[2026-10-07T18:37:45Z] load site instr 41: src r4 win 0 off 0; last base writer rotl at 40; writers back to the last injecting one: rotl@40 load@32; shadow writes of r4 per rep: add=5 mad=5 mul=4 mulhi=5 or=1 rotl=4 rotr=5 shfl=3 sub=5 xor=2
[2026-10-07T18:37:45Z] load site instr 45: src r5 win 1 off 0; last base writer rotr at 27; writers back to the last injecting one: rotr@27 shfl@24; shadow writes of r5 per rep: add=10 mul=5 mulhi=1 rotl=3 rotr=1 shfl=4 sub=3 xor=4
[2026-10-07T18:37:45Z] load site instr 50: src r7 win 2 off 3; last base writer load at 45; writers back to the last injecting one: load@45; shadow writes of r7 per rep: add=5 mad=5 mul=4 mulhi=5 rotl=1 rotr=5 shfl=3 sub=4 xor=5
[2026-10-07T18:37:45Z] load site instr 58: src r3 win 0 off 0; last base writer load at 50; writers back to the last injecting one: load@50; shadow writes of r3 per rep: mad=6 mul=2 mulhi=4 or=1 rotl=4 rotr=2 xor=4
[2026-10-07T18:37:45Z] load site instr 59: src r2 win 0 off 0; last base writer xor at 46; writers back to the last injecting one: xor@46; shadow writes of r2 per rep: add=3 mad=2 mul=4 mulhi=2 or=6 rotl=1 rotr=1 shfl=5 sub=2 xor=4
[2026-10-07T18:37:45Z] load site instr 63: src r1 win 1 off 0; last base writer rotr at 62; writers back to the last injecting one: rotr@62 add@56; shadow writes of r1 per rep: add=5 mad=4 mul=6 mulhi=3 or=2 rotl=2 rotr=3 shfl=5 sub=3 xor=2
[2026-10-07T18:37:45Z] window layer: site item windows (first, items) (0x0,2^23) (0x0,2^24) (0x400000,2^22) (0x0,2^24) (0x800000,2^23) (0x800000,2^23) (0x0,2^22) (0x0,2^22) (0x800000,2^23) (0x800000,2^23) (0x0,2^24) (0x0,2^23) (0xc00000,2^22) (0x0,2^24) (0x0,2^24) (0x0,2^23); expected reads per item by quarter 9.060 7.153 6.199 8.106 (flat uniform 7.629)
[2026-10-07T18:37:49Z] 31250 warps (1000000 nonces) interpreted in 4.2 s (4.322 ms per warp per thread); Epoch::hash_warp agreement on 95 warps: 0 mismatches
[2026-10-07T18:37:49Z] p100767-attack-f8 distinct lines per hash: n 1000000 min 976 p1 1008 median 1024 p99 1024 max 1024 mean 1021.6579
[2026-10-07T18:37:49Z] p100767-attack-f8 distinct items per hash: n 1000000 min 122 p1 126 median 128 p99 128 max 128 mean 127.7228
[2026-10-07T18:37:49Z] p100767-attack-f8 distinct lines per warp: n 31250 min 32426 p1 32500 median 32567 p99 32622 max 32658 mean 32565.9023
[2026-10-07T18:37:49Z] p100767-attack-f8 distinct items per warp: n 31250 min 4070 p1 4079 median 4087 p99 4093 max 4096 mean 4086.6272
[2026-10-07T18:37:49Z] uniform expectation: distinct lines per hash 1023.875 of 1024 reads, per warp 32640.3 of 32768 reads (2^22 lines); distinct items per hash 127.9995 of 128, per warp 4095.50 of 4096 (2^24 items)
[2026-10-07T18:37:49Z] positions: largest hi16-bucket (256 items) share 0.0283% at p22 (iteration 1, site 6, instr 23); windowed expectation 0.0061%; largest saturated-source share 0.0130% at p22 (site 6, instr 23)
[2026-10-07T18:37:49Z] acceptance-style (2,048 evaluations, the rule's sample size): the most repeated address at one position 2 of 2048 at p118 (site 6, instr 23); saturated sources at one position 1 of 2048 at p102 (site 6, instr 23); uniform expectation: repeats 1 to 2, saturated 0
[2026-10-07T18:37:49Z] p100767-attack-f8 item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 1677 (bin 0) z_max +604.38 min 0 z_min -2.76 chi2/dof 1.24228 chi2_z +701.71 top0.1% 0.34274% top0.5% 1.26980% top1% 2.32590%
[2026-10-07T18:37:49Z] p100767-attack-f8 item histogram against the window density: full 2^24 chi2/dof 1.07966 largest 1677 (item 0x0) at +554.14 sigma smallest at -3.01; buckets64 chi2/dof 1.05456 largest 2217 (bucket 0) at +67.99 sigma smallest at -4.21
[2026-10-07T18:37:50Z] p100767-attack-f8 WINDOWED CONTROL item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 27 (bin 919726) z_max +7.01 min 0 z_min -2.76 chi2/dof 1.14901 chi2_z +431.58 top0.1% 0.25578% top0.5% 1.15209% top1% 2.18797%
[2026-10-07T18:37:50Z] p100767-attack-f8 WINDOWED CONTROL against the window density: full chi2/dof 1.00017 largest +6.64 sigma; buckets64 chi2/dof 0.99639 largest +4.40 smallest -4.40
[2026-10-07T18:37:50Z] 6-sigma p100767-attack-f8 items buckets64 against the window density: largest bucket +67.99 sigma, smallest -4.21 -> FLAGGED (beyond 6 sigma)
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (windowed control): f 0.1% S_f 0.34274% E_f(control) 0.25578% X_f +0.08696% X_f/f +0.8696 -> no hot set
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (windowed control): f 0.5% S_f 1.26980% E_f(control) 1.15209% X_f +0.11771% X_f/f +0.2354 -> no hot set
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (windowed control): f 1.0% S_f 2.32590% E_f(control) 2.18797% X_f +0.13793% X_f/f +0.1379 -> no hot set
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (windowed control): verdict clear
[2026-10-07T18:37:50Z] p100767-attack-f8 FLAT CONTROL item histogram: bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 26 (bin 10438708) z_max +6.65 min 0 z_min -2.76 chi2/dof 0.99974 chi2_z -0.75 top0.1% 0.24260% top0.5% 1.09881% top1% 2.08683%
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (flat control, the auditor's first view): f 0.1% S_f 0.34274% E_f(control) 0.24260% X_f +0.10014% X_f/f +1.0014 -> HOT SET
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (flat control, the auditor's first view): f 0.5% S_f 1.26980% E_f(control) 1.09881% X_f +0.17099% X_f/f +0.3420 -> no hot set
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (flat control, the auditor's first view): f 1.0% S_f 2.32590% E_f(control) 2.08683% X_f +0.23907% X_f/f +0.2391 -> no hot set
[2026-10-07T18:37:50Z] hot-set p100767-attack-f8 items (flat control, the auditor's first view): verdict FLAGGED
[2026-10-07T18:37:50Z] ratio p100767-attack-f8: top 0.1% / 0.5% / 1% share over the WINDOW-MODEL control 1.3400x / 1.1022x / 1.0630x (gate 1.2x at 0.1%: BEYOND); over the FLAT control 1.4128x / 1.1556x / 1.1146x
[2026-10-07T18:37:50Z] hottest item p100767-attack-f8 (0x000000): 1677 reads (0.0013% of all); predicted source site-instr6:r6:zero:last-writer-mad@4
[2026-10-07T18:37:50Z] program p100767-attack-f8 DONE: 1000000 nonces; 6-sigma (windowed) FLAGGED; hot-set clear; verdict FLAGGED
[2026-10-07T18:37:51Z] attack-f8 0.1.0 (igneum-pow 4); args ["warps", "--program", "100159", "--nonces", "1000000", "--threads", "32", "--out", "/srv/builds/igneum-wt-adv-accept/adv/live", "--validate", "sample"]
[2026-10-07T18:37:51Z] census warps: programs [100159] day 20730 nonces 1000000 threads 32 plant none validate sample check_every 997 diag false
[2026-10-07T18:37:51Z] day 20730: cache filled in 0.65 s, fnv 448274a57f508cbc
[2026-10-07T18:37:54Z] table: 16777216 items derived with their 8 lines in 2.8 s; library comparison sample (0 mismatches)
[2026-10-07T18:37:59Z] program p100159-attack-f8: epoch seed 1d76edc78af5ddea74900f1e005f77ef1b115c69c583ea6f57792e386806aede era seed b659277d9635ccc59ea759fd747241f440d5dc89947f629a2473ccef349c869b id 573d650159a8b04e attempt 3 class mx8-era9e0e4f28+sh256x27 op mix load=16 mulhi=7 xor=7 mul=6 add=5 mad=5 rotl=4 shfl=4 sub=4 or=3 rotr=3; era stride mul 0xa39d49c9 rot 7 interleave [3, 8, 9, 10] windows 5:2:0 10:0:0 16:2:3 17:2:2 23:2:2 26:2:2 40:0:0 42:2:3 45:1:0 47:2:3 49:2:2 56:0:0 59:2:0 60:2:3 61:0:0 62:0:0
[2026-10-07T18:37:59Z] load site instr 5: src r4 win 2 off 0; last base writer shfl at 0; writers back to the last injecting one: shfl@0; shadow writes of r4 per rep: add=5 mad=3 mul=1 mulhi=1 or=2 rotl=3 rotr=2 shfl=4 sub=1 xor=5
[2026-10-07T18:37:59Z] load site instr 10: src r7 win 0 off 0; last base writer add at 6; writers back to the last injecting one: add@6; shadow writes of r7 per rep: add=9 mad=2 mul=1 mulhi=1 or=2 rotl=3 shfl=4 sub=2 xor=9
[2026-10-07T18:37:59Z] load site instr 16: src r5 win 2 off 3; last base writer xor at 9; writers back to the last injecting one: xor@9; shadow writes of r5 per rep: add=1 mad=4 mul=3 mulhi=1 or=2 rotl=4 rotr=1 shfl=6 sub=1 xor=5
[2026-10-07T18:37:59Z] load site instr 17: src r1 win 2 off 2; last base writer load at 10; writers back to the last injecting one: load@10; shadow writes of r1 per rep: add=3 mad=6 mul=4 mulhi=3 or=4 rotl=3 rotr=4 shfl=3 sub=2 xor=6
[2026-10-07T18:37:59Z] load site instr 23: src r7 win 2 off 2; last base writer load at 16; writers back to the last injecting one: load@16; shadow writes of r7 per rep: add=9 mad=2 mul=1 mulhi=1 or=2 rotl=3 shfl=4 sub=2 xor=9
[2026-10-07T18:37:59Z] load site instr 26: src r5 win 2 off 2; last base writer xor at 25; writers back to the last injecting one: xor@25; shadow writes of r5 per rep: add=1 mad=4 mul=3 mulhi=1 or=2 rotl=4 rotr=1 shfl=6 sub=1 xor=5
[2026-10-07T18:37:59Z] load site instr 40: src r1 win 0 off 0; last base writer shfl at 30; writers back to the last injecting one: shfl@30; shadow writes of r1 per rep: add=3 mad=6 mul=4 mulhi=3 or=4 rotl=3 rotr=4 shfl=3 sub=2 xor=6
[2026-10-07T18:37:59Z] load site instr 42: src r5 win 2 off 3; last base writer mad at 33; writers back to the last injecting one: mad@33; shadow writes of r5 per rep: add=1 mad=4 mul=3 mulhi=1 or=2 rotl=4 rotr=1 shfl=6 sub=1 xor=5
[2026-10-07T18:37:59Z] load site instr 45: src r7 win 1 off 0; last base writer add at 44; writers back to the last injecting one: add@44; shadow writes of r7 per rep: add=9 mad=2 mul=1 mulhi=1 or=2 rotl=3 shfl=4 sub=2 xor=9
[2026-10-07T18:37:59Z] load site instr 47: src r2 win 2 off 3; last base writer load at 45; writers back to the last injecting one: load@45; shadow writes of r2 per rep: add=9 mad=4 mul=3 mulhi=2 or=2 rotl=2 rotr=1 shfl=2 sub=2 xor=4
[2026-10-07T18:37:59Z] load site instr 49: src r5 win 2 off 2; last base writer load at 47; writers back to the last injecting one: load@47; shadow writes of r5 per rep: add=1 mad=4 mul=3 mulhi=1 or=2 rotl=4 rotr=1 shfl=6 sub=1 xor=5
[2026-10-07T18:37:59Z] load site instr 56: src r1 win 0 off 0; last base writer mad at 51; writers back to the last injecting one: mad@51; shadow writes of r1 per rep: add=3 mad=6 mul=4 mulhi=3 or=4 rotl=3 rotr=4 shfl=3 sub=2 xor=6
[2026-10-07T18:37:59Z] load site instr 59: src r7 win 2 off 0; last base writer add at 50; writers back to the last injecting one: add@50; shadow writes of r7 per rep: add=9 mad=2 mul=1 mulhi=1 or=2 rotl=3 shfl=4 sub=2 xor=9
[2026-10-07T18:37:59Z] load site instr 60: src r2 win 2 off 3; last base writer rotr at 54; writers back to the last injecting one: rotr@54 load@45; shadow writes of r2 per rep: add=9 mad=4 mul=3 mulhi=2 or=2 rotl=2 rotr=1 shfl=2 sub=2 xor=4
[2026-10-07T18:37:59Z] load site instr 61: src r6 win 0 off 0; last base writer load at 56; writers back to the last injecting one: load@56; shadow writes of r6 per rep: add=4 mad=3 mul=1 mulhi=4 rotl=1 rotr=2 shfl=2 sub=3 xor=8
[2026-10-07T18:37:59Z] load site instr 62: src r5 win 0 off 0; last base writer load at 60; writers back to the last injecting one: load@60; shadow writes of r5 per rep: add=1 mad=4 mul=3 mulhi=1 or=2 rotl=4 rotr=1 shfl=6 sub=1 xor=5
[2026-10-07T18:37:59Z] window layer: site item windows (first, items) (0x0,2^22) (0x0,2^24) (0xc00000,2^22) (0x800000,2^22) (0x800000,2^22) (0x800000,2^22) (0x0,2^24) (0xc00000,2^22) (0x0,2^23) (0xc00000,2^22) (0x800000,2^22) (0x0,2^24) (0x0,2^22) (0xc00000,2^22) (0x0,2^24) (0x0,2^24); expected reads per item by quarter 7.153 3.338 10.014 10.014 (flat uniform 7.629)
[2026-10-07T18:38:02Z] 31250 warps (1000000 nonces) interpreted in 3.8 s (3.873 ms per warp per thread); Epoch::hash_warp agreement on 95 warps: 0 mismatches
[2026-10-07T18:38:02Z] p100159-attack-f8 distinct lines per hash: n 1000000 min 976 p1 1000 median 1023 p99 1024 max 1024 mean 1019.4789
[2026-10-07T18:38:02Z] p100159-attack-f8 distinct items per hash: n 1000000 min 122 p1 125 median 128 p99 128 max 128 mean 127.4503
[2026-10-07T18:38:02Z] p100159-attack-f8 distinct lines per warp: n 31250 min 32340 p1 32411 median 32497 p99 32571 max 32616 mean 32496.3442
[2026-10-07T18:38:02Z] p100159-attack-f8 distinct items per warp: n 31250 min 4059 p1 4068 median 4078 p99 4087 max 4091 mean 4077.8626
[2026-10-07T18:38:02Z] uniform expectation: distinct lines per hash 1023.875 of 1024 reads, per warp 32640.3 of 32768 reads (2^22 lines); distinct items per hash 127.9995 of 128, per warp 4095.50 of 4096 (2^24 items)
[2026-10-07T18:38:02Z] positions: largest hi16-bucket (256 items) share 0.0179% at p23 (iteration 1, site 7, instr 42); windowed expectation 0.0061%; largest saturated-source share 0.0000% at p0 (site 0, instr 5)
[2026-10-07T18:38:02Z] acceptance-style (2,048 evaluations, the rule's sample size): the most repeated address at one position 2 of 2048 at p9 (site 9, instr 47); saturated sources at one position 0 of 2048 at p0 (site 0, instr 5); uniform expectation: repeats 1 to 2, saturated 0
[2026-10-07T18:38:03Z] p100159-attack-f8 item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 37 (bin 12582926) z_max +10.63 min 0 z_min -2.76 chi2/dof 2.09764 chi2_z +3179.10 top0.1% 0.30020% top0.5% 1.34002% top1% 2.53830%
[2026-10-07T18:38:03Z] p100159-attack-f8 item histogram against the window density: full 2^24 chi2/dof 1.08732 largest 37 (item 0xc0000e) at +8.53 sigma smallest at -3.16; buckets64 chi2/dof 6.13245 largest 1589 (bucket 196608) at +37.45 sigma smallest at -7.46
[2026-10-07T18:38:04Z] p100159-attack-f8 WINDOWED CONTROL item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 30 (bin 9414155) z_max +8.10 min 0 z_min -2.76 chi2/dof 1.98344 chi2_z +2848.35 top0.1% 0.28063% top0.5% 1.26922% top1% 2.41408%
[2026-10-07T18:38:04Z] p100159-attack-f8 WINDOWED CONTROL against the window density: full chi2/dof 0.99980 largest +6.38 sigma; buckets64 chi2/dof 0.99973 largest +4.54 smallest -4.49
[2026-10-07T18:38:04Z] 6-sigma p100159-attack-f8 items buckets64 against the window density: largest bucket +37.45 sigma, smallest -7.46 -> FLAGGED (beyond 6 sigma)
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (windowed control): f 0.1% S_f 0.30020% E_f(control) 0.28063% X_f +0.01957% X_f/f +0.1957 -> no hot set
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (windowed control): f 0.5% S_f 1.34002% E_f(control) 1.26922% X_f +0.07081% X_f/f +0.1416 -> no hot set
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (windowed control): f 1.0% S_f 2.53830% E_f(control) 2.41408% X_f +0.12422% X_f/f +0.1242 -> no hot set
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (windowed control): verdict clear
[2026-10-07T18:38:04Z] p100159-attack-f8 FLAT CONTROL item histogram: bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 26 (bin 8617755) z_max +6.65 min 0 z_min -2.76 chi2/dof 0.99990 chi2_z -0.30 top0.1% 0.24308% top0.5% 1.09931% top1% 2.08730%
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (flat control, the auditor's first view): f 0.1% S_f 0.30020% E_f(control) 0.24308% X_f +0.05712% X_f/f +0.5712 -> no hot set
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (flat control, the auditor's first view): f 0.5% S_f 1.34002% E_f(control) 1.09931% X_f +0.24071% X_f/f +0.4814 -> no hot set
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (flat control, the auditor's first view): f 1.0% S_f 2.53830% E_f(control) 2.08730% X_f +0.45100% X_f/f +0.4510 -> no hot set
[2026-10-07T18:38:04Z] hot-set p100159-attack-f8 items (flat control, the auditor's first view): verdict clear
[2026-10-07T18:38:04Z] ratio p100159-attack-f8: top 0.1% / 0.5% / 1% share over the WINDOW-MODEL control 1.0697x / 1.0558x / 1.0515x (gate 1.2x at 0.1%: within); over the FLAT control 1.2350x / 1.2190x / 1.2161x
[2026-10-07T18:38:04Z] hottest item p100159-attack-f8 (0xc0000e): 37 reads (0.0000% of all); predicted source none
[2026-10-07T18:38:04Z] program p100159-attack-f8 DONE: 1000000 nonces; 6-sigma (windowed) FLAGGED; hot-set clear; verdict FLAGGED

47
docs/analysis/cryptanalysis/logs/adv-accept/live-plant-p2.log Normal file
View file

@ -0,0 +1,47 @@
[2026-10-07T18:30:08Z] attack-f8 0.1.0 (igneum-pow 4); args ["warps", "--program", "2", "--nonces", "1000000", "--threads", "32", "--out", "/srv/builds/igneum-wt-adv-accept/adv/live", "--plant", "const-item", "--validate", "sample"]
[2026-10-07T18:30:08Z] census warps: programs [2] day 20730 nonces 1000000 threads 32 plant const-item validate sample check_every 997 diag false
[2026-10-07T18:30:24Z] day 20730: cache filled in 15.39 s, fnv 448274a57f508cbc
[2026-10-07T18:30:25Z] table: 16777216 items derived with their 8 lines in 1.2 s; library comparison skipped under the plant (0 mismatches)
[2026-10-07T18:30:30Z] program p2-attack-f8: epoch seed 59cef1aaa406506685f4355933868219bedece8dbd4e84c281724db9313bfdfa era seed 9cba001faa1e017235841c72f560ebd1b93f6d7aea5f9a598f7d3080e8a11f69 id 8a0047b2eb071ade attempt 5 class mx8-era35d128b9+sh256x27 op mix load=16 mul=8 add=6 mulhi=6 shfl=6 mad=4 or=4 rotl=4 rotr=4 xor=4 sub=2; era stride mul 0xbf0fab63 rot 4 interleave [2, 3, 7, 15] windows 2:2:0 3:0:0 7:0:0 9:0:0 13:2:1 17:1:1 21:2:0 22:1:0 23:0:0 27:1:0 30:2:2 40:1:0 44:2:1 55:0:0 62:0:0 63:2:0
[2026-10-07T18:30:30Z] load site instr 2: src r1 win 2 off 0; last base writer add at 1; writers back to the last injecting one: add@1; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 3: src r5 win 0 off 0; last base writer mad at 0; writers back to the last injecting one: mad@0; shadow writes of r5 per rep: add=3 mad=4 mul=5 or=5 rotl=1 rotr=4 shfl=4 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 7: src r1 win 0 off 0; last base writer shfl at 5; writers back to the last injecting one: shfl@5; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 9: src r3 win 0 off 0; last base writer add at 6; writers back to the last injecting one: add@6; shadow writes of r3 per rep: add=4 mad=3 mul=1 mulhi=2 rotl=1 rotr=2 shfl=3 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 13: src r0 win 2 off 1; last base writer xor at 12; writers back to the last injecting one: xor@12; shadow writes of r0 per rep: add=5 mad=6 mul=1 mulhi=3 or=1 rotl=1 rotr=7 shfl=8 sub=2
[2026-10-07T18:30:30Z] load site instr 17: src r5 win 1 off 1; last base writer xor at 14; writers back to the last injecting one: xor@14; shadow writes of r5 per rep: add=3 mad=4 mul=5 or=5 rotl=1 rotr=4 shfl=4 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 21: src r4 win 2 off 0; last base writer shfl at 15; writers back to the last injecting one: shfl@15; shadow writes of r4 per rep: add=4 mad=4 mulhi=1 or=1 rotl=4 rotr=1 shfl=2 sub=8 xor=6
[2026-10-07T18:30:30Z] load site instr 22: src r5 win 1 off 0; last base writer add at 20; writers back to the last injecting one: add@20; shadow writes of r5 per rep: add=3 mad=4 mul=5 or=5 rotl=1 rotr=4 shfl=4 sub=3 xor=3
[2026-10-07T18:30:30Z] load site instr 23: src r1 win 0 off 0; last base writer load at 13; writers back to the last injecting one: load@13; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 27: src r2 win 1 off 0; last base writer xor at 26; writers back to the last injecting one: xor@26; shadow writes of r2 per rep: add=6 mad=2 mul=4 mulhi=3 or=3 rotl=5 rotr=3 shfl=3 sub=3 xor=6
[2026-10-07T18:30:30Z] load site instr 30: src r2 win 2 off 2; last base writer rotr at 29; writers back to the last injecting one: rotr@29 xor@26; shadow writes of r2 per rep: add=6 mad=2 mul=4 mulhi=3 or=3 rotl=5 rotr=3 shfl=3 sub=3 xor=6
[2026-10-07T18:30:30Z] load site instr 40: src r1 win 1 off 0; last base writer rotl at 32; writers back to the last injecting one: rotl@32 rotl@31 load@30; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] load site instr 44: src r4 win 2 off 1; last base writer load at 27; writers back to the last injecting one: load@27; shadow writes of r4 per rep: add=4 mad=4 mulhi=1 or=1 rotl=4 rotr=1 shfl=2 sub=8 xor=6
[2026-10-07T18:30:30Z] load site instr 55: src r4 win 0 off 0; last base writer mad at 51; writers back to the last injecting one: mad@51; shadow writes of r4 per rep: add=4 mad=4 mulhi=1 or=1 rotl=4 rotr=1 shfl=2 sub=8 xor=6
[2026-10-07T18:30:30Z] load site instr 62: src r0 win 0 off 0; last base writer add at 53; writers back to the last injecting one: add@53; shadow writes of r0 per rep: add=5 mad=6 mul=1 mulhi=3 or=1 rotl=1 rotr=7 shfl=8 sub=2
[2026-10-07T18:30:30Z] load site instr 63: src r1 win 2 off 0; last base writer load at 62; writers back to the last injecting one: load@62; shadow writes of r1 per rep: add=7 mad=6 mul=4 mulhi=6 or=1 rotl=3 rotr=6 shfl=1 sub=2 xor=4
[2026-10-07T18:30:30Z] window layer: site item windows (first, items) (0x0,2^22) (0x0,2^24) (0x0,2^24) (0x0,2^24) (0x400000,2^22) (0x800000,2^23) (0x0,2^22) (0x0,2^23) (0x0,2^24) (0x0,2^23) (0x800000,2^22) (0x0,2^23) (0x400000,2^22) (0x0,2^24) (0x0,2^24) (0x0,2^22); expected reads per item by quarter 11.444 9.537 5.722 3.815 (flat uniform 7.629)
[2026-10-07T18:30:34Z] 31250 warps (1000000 nonces) interpreted in 3.7 s (3.754 ms per warp per thread); Epoch::hash_warp agreement on 0 warps: 0 mismatches (library comparison skipped under the plant)
[2026-10-07T18:30:34Z] p2-attack-f8 distinct lines per hash: n 1000000 min 952 p1 967 median 968 p99 968 max 968 mean 967.8845
[2026-10-07T18:30:34Z] p2-attack-f8 distinct items per hash: n 1000000 min 119 p1 121 median 121 p99 121 max 121 mean 120.9995
[2026-10-07T18:30:34Z] p2-attack-f8 distinct lines per warp: n 31250 min 30551 p1 30582 median 30612 p99 30637 max 30661 mean 30611.6212
[2026-10-07T18:30:34Z] p2-attack-f8 distinct items per warp: n 31250 min 3835 p1 3838 median 3841 p99 3841 max 3841 mean 3840.5035
[2026-10-07T18:30:34Z] uniform expectation: distinct lines per hash 1023.875 of 1024 reads, per warp 32640.3 of 32768 reads (2^22 lines); distinct items per hash 127.9995 of 128, per warp 4095.50 of 4096 (2^24 items)
[2026-10-07T18:30:34Z] positions: largest hi16-bucket (256 items) share 100.0000% at p0 (iteration 0, site 0, instr 2); windowed expectation 0.0061%; largest saturated-source share 0.0000% at p0 (site 0, instr 2)
[2026-10-07T18:30:34Z] acceptance-style (2,048 evaluations, the rule's sample size): the most repeated address at one position 1 of 2048 at p0 (site 0, instr 2); saturated sources at one position 0 of 2048 at p0 (site 0, instr 2); uniform expectation: repeats 1 to 2, saturated 0
[2026-10-07T18:30:34Z] p2-attack-f8 item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 8000007 (bin 4660) z_max +2896309.15 min 0 z_min -2.76 chi2/dof 500001.72317 chi2_z +1448156739.23 top0.1% 6.52236% top0.5% 7.47807% top1% 8.58175%
[2026-10-07T18:30:34Z] p2-attack-f8 item histogram against the window density: full 2^24 chi2/dof 333334.02123 largest 8000007 (item 0x1234) at +2364825.39 sigma smallest at -3.38; buckets64 chi2/dof 333328.12062 largest 8000582 (bucket 72) at +295597.78 sigma smallest at -7.96
[2026-10-07T18:30:34Z] p2-attack-f8 WINDOWED CONTROL item histogram (flat): bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 37 (bin 3356675) z_max +10.63 min 0 z_min -2.76 chi2/dof 2.19210 chi2_z +3452.68 top0.1% 0.29918% top0.5% 1.34673% top1% 2.55749%
[2026-10-07T18:30:34Z] p2-attack-f8 WINDOWED CONTROL against the window density: full chi2/dof 0.99966 largest +7.55 sigma; buckets64 chi2/dof 1.00387 largest +4.49 smallest -4.89
[2026-10-07T18:30:34Z] 6-sigma p2-attack-f8 items buckets64 against the window density: largest bucket +295597.78 sigma, smallest -7.96 -> FLAGGED (beyond 6 sigma)
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): f 0.1% S_f 6.52236% E_f(control) 0.29918% X_f +6.22317% X_f/f +62.2317 -> HOT SET
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): f 0.5% S_f 7.47807% E_f(control) 1.34673% X_f +6.13134% X_f/f +12.2627 -> HOT SET
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): f 1.0% S_f 8.58175% E_f(control) 2.55749% X_f +6.02426% X_f/f +6.0243 -> HOT SET
[2026-10-07T18:30:34Z] hot-set p2-attack-f8 items (windowed control): verdict FLAGGED
[2026-10-07T18:30:35Z] p2-attack-f8 FLAT CONTROL item histogram: bins 16777216 reads 128000000 mean 7.629 sigma 2.762 max 28 (bin 13310856) z_max +7.37 min 0 z_min -2.76 chi2/dof 0.99990 chi2_z -0.29 top0.1% 0.24291% top0.5% 1.09940% top1% 2.08772%
[2026-10-07T18:30:35Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): f 0.1% S_f 6.52236% E_f(control) 0.24291% X_f +6.27945% X_f/f +62.7945 -> HOT SET
[2026-10-07T18:30:35Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): f 0.5% S_f 7.47807% E_f(control) 1.09940% X_f +6.37867% X_f/f +12.7573 -> HOT SET
[2026-10-07T18:30:35Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): f 1.0% S_f 8.58175% E_f(control) 2.08772% X_f +6.49402% X_f/f +6.4940 -> HOT SET
[2026-10-07T18:30:35Z] hot-set p2-attack-f8 items (flat control, the auditor's first view): verdict FLAGGED
[2026-10-07T18:30:35Z] ratio p2-attack-f8: top 0.1% / 0.5% / 1% share over the WINDOW-MODEL control 21.8005x / 5.5528x / 3.3555x (gate 1.2x at 0.1%: BEYOND); over the FLAT control 26.8510x / 6.8020x / 4.1106x
[2026-10-07T18:30:35Z] hottest item p2-attack-f8 (0x001234): 8000007 reads (6.2500% of all); predicted source none
[2026-10-07T18:30:35Z] program p2-attack-f8 DONE: 1000000 nonces; 6-sigma (windowed) FLAGGED; hot-set FLAGGED; verdict FLAGGED

66
docs/analysis/cryptanalysis/report-acceptance-rule.md
View file

@ -79,6 +79,72 @@ upper bound on how concentrated an accepted program can be on the stand-in.
Numbers land here when the run finishes (first results by 8 October 18:00 UTC).
## Results as they land
### Tool validation (box 1, 18:30Z, 10^6 nonces each, about 30 s per run on 32 threads)
| Run | Hot-set test (window-model control) | Top 0.1% share ratio over the window model | 6-sigma | Verdict |
|---|---|---|---|---|
| `adv-live warps --program 2 --plant const-item` (known-failed shape) | f 1%: S_f 8.58%, E_f 2.09%, X_f +6.49% (X_f/f 6.49) HOT SET | 21.80x (flat control 26.85x) | FLAGGED | FLAGGED |
| `adv-live warps --program 2` (clean control) | f 1%: S_f 2.56%, E_f 2.09%, X_f +0.47% (X_f/f 0.47) no hot set | 0.9996x (flat 1.2312x, the window layer) | clear | PASS |
The plant fires and the control is clean, so the live hot-set numbers below are trusted. Logs
adv/live-plant-p2.log and adv/live-control-p2.log on box 1.
### Sweep shard 01, first 864 accepted programs (box 1, seeds 100,002..., read at 18:33Z)
| Metric | Value |
|---|---|
| Attempt histogram 0..9 | 273, 191, 134, 90, 59, 38, 34, 12, 7, 11; max 26 |
| Min per-site distinct-index ratio at 256 units | 0.9986 (floor 0.98 at 2^20; every row above 0.998) |
| Programs with a lossy (or/mul/mulhi) last write to an output register | 715 of 864 (83%) |
| Programs writing fewer than 8 registers | 0 |
| Last-resort programs | 0 |
| Lowest-ratio seeds (live hot set queued) | 100211 (0.9986), 100629 (0.9987), 100064, 100767 (0.9988), 100159 |
Reading: on the stand-in the accepted population sits 0.02 above the 0.98 floor at this sample size; the
floor leaves headroom (Q5), and no accepted program in this sample comes near it. Whether any of the
lossy-last-write programs concentrate reads on the LIVE set is what the live census answers.
### Rule change in the method column (box-hours honesty)
18:27Z to 18:4xZ: shards ran under the capacity layer's SIGSTOP yield and were paused almost the whole
time (a build slot held nearly continuously on both boxes). Ruling at 18:4xZ: yield dropped, every sweep at
nice 10 on cores 8-95 (cores 0-7 reserved for release builds, the seed and the observer). Shards 00 and 01
SIGCONTed and re-pinned at 18:4xZ. Box-hours before the ruling: about 0.1 of running time.
### Live hot set of the five lowest-ratio shard-01 programs (box 1, 18:37Z, 10^6 nonces each)
All five pass the rule (they are the chain's accepted programs). Log logs/adv-accept/live-lowratio-s01.log.
Columns: the f8 gate (top 0.1% share over the window-model control, gate 1.2x), the hot-set excess X_f at
f = 0.1% and 1% (a hot set needs X_f >= f), the windowed 6-sigma test, the hottest item and its traced source.
| Seed | id | attempt | closed ratio (256 u) | top 0.1% over window model | X_0.1% / X_1% | 6-sigma (buckets64) | hottest item, reads, share, source |
|---|---|---|---|---|---|---|---|
| 100211 | d07885a437e237c7 | 0 | 0.9986 | 1.08x | +0.026% / +0.049% | +42.4 sigma FLAGGED | 0x454585, 1238 (0.0010%), site instr 4 reads r0 one-zero-bit, writer load@3 |
| 100629 | 37c849d9741c5994 | 0 | 0.9987 | 1.01x | +0.003% / +0.015% | +5.6 within | 30 reads, none |
| 100064 | 2442abf9d56f6611 | 2 | 0.9988 | 1.17x | +0.044% / +0.189% | +77.3 sigma FLAGGED | 54 reads; the flag is a 256-item bucket at p43 (iteration 2, site 11, instr 33) holding 0.048% of that position's reads against 0.0061% expected (8x) |
| 100767 | 9d68e6286fc817d4 | 2 | 0.9988 | 1.34x BEYOND the 1.2x gate | +0.087% / +0.138% | +68.0 sigma FLAGGED | 0x000000, 1677 (0.0013%), site instr 6 reads r6 = zero, writer mad@4; saturated-source share 0.013% at that position (the (c') limit is 1%) |
| 100159 | 573d650159a8b04e | 3 | 0.9988 | 1.07x | +0.020% / +0.124% | +37.5 sigma FLAGGED | 37 reads, none |
| p2 control (18:30Z) | f8 seed 2 | | | 0.9996x | +0.00% / +0.47% (flat) | clear | 32 reads |
Reading, as an attacker. (1) The stand-in ratio does correlate with live concentration: four of the five
lowest-ratio seeds flag the windowed 6-sigma test at 10^6 nonces where the random control is clear, and one
sits beyond the f8 gate. (2) The concentration is NOT a hot set a chip can use: the top 0.1% of items (16.7k
items, 1 MB) take at most 0.34% of reads against 0.26% expected; the hottest single item takes 0.0013% of
reads. An on-die copy of these items saves under 0.1% of DRAM reads: no row of chip-model-v3.md moves.
(3) The mechanism is the known one, a near-saturated source (zero, or one bit off all ones) at one site in
one iteration, below the (c') 1% limit (0.013% here), so (c') never fires, and the per-site ratio at 0.98 is
loose enough (these sit at 0.9986) that (c'') never fires either. (4) Confirmation at the gate's own 2^24
nonces with attribution is running (adv/live-confirm-16m.log); numbers replace this paragraph when it lands.
### Live hot-set census (RUNNING)
- box 2, 18:5xZ: `adv-live census --programs 2..201 --nonces 1000000 --threads 32` (200 consecutive accepted
programs of shard 00's range: the random sample), log adv/live-census-s00-2-201.log.
- box 1, 18:5xZ: `adv-live warps --program k --nonces 1000000` for the five lowest-ratio shard-01 seeds, log
adv/live-lowratio-s01.log.
## Running notes
- Box-hours and commands are logged in each section with the seed, so every claim is reproducible by

244
tools/attack/adv-accept/src/main.rs
View file

@ -327,9 +327,253 @@ fn main() {
"derive-check" => derive_check(&args[2..]),
"sweep" => sweep(&args[2..]),
"plant" => plant(&args[2..]),
"gap" => gap_cmd(&args[2..]),
_ => {
eprintln!("unknown command {}", args[1]);
std::process::exit(2);
}
}
}
// --------------------------------------------------------------------------------------------------------------
// Q2: the stand-in gap. The rule's dynamic test runs on the closed-form dataset; the chain hashes the live
// memory-hard one. A load index at site s depends on the dataset WORDS loaded earlier in the hash, so the
// per-site index distribution (and the rule's (c) metrics) can differ between the two. This command mirrors
// the interpreter with per-site index capture, validates the mirror against the library's hash_warp on both
// datasets and its closed-form distinct counts against the library's distinct_indices_v4, then reports the
// per-site distinct-index ratio on the live set beside the closed-form one over the rule's own 4096 units,
// and the (c) metrics (distinct addresses per hash, saturated finals, output bias) on both over the (c) units.
// Known-failed shape: `--plant zero-dataset` makes every live word 0, so live indices collapse and the tool
// must report a large gap.
// --------------------------------------------------------------------------------------------------------------
mod gap {
use igneum_pow::accept::{accept_base_nonces_n, distinct_indices_v4, ACCEPT_DATASET_LOG2, ACCEPT_UNITS, ACCEPT_UNITS_DISTINCT_V4};
use igneum_pow::bind::day_bytes;
use igneum_pow::generator::{Instr, Op, Program, ProgramClass, ITERATIONS, LANES};
use igneum_pow::memhard::Layout;
use igneum_pow::verify::{hash_warp, load_index, splitmix32, DatasetMode, DatasetSource, Epoch};
use std::sync::atomic::{AtomicUsize, Ordering};
pub struct Words<'a> {
pub ds: &'a DatasetSource,
pub layout: Layout,
pub zero: bool,
}
impl<'a> Words<'a> {
#[inline]
fn word(&self, idx: u32) -> u32 {
if self.zero { 0 } else { self.ds.word_at(self.layout, idx) }
}
}
/// One warp of the program on `w`, with every load's masked index pushed into `sites[site]`; returns the
/// 32 hashes and the per-lane distinct masked addresses, and counts saturated finals and output bit ones.
pub fn warp(p: &Program, w: &Words, base: u32, sites: &mut [Vec<u32>], lane_distinct: &mut u64, sat: &mut u32, ones: &mut [u32; 64]) -> [u64; LANES] {
let seed = &p.seed;
let mask: u32 = (1u32 << ACCEPT_DATASET_LOG2) - 1;
let era = p.class.era;
let mut r = [[0u32; LANES]; 8];
for lane in 0..LANES {
let nonce = base.wrapping_add(lane as u32);
for i in 0..8 {
let mut x = nonce ^ seed[i];
x = x.wrapping_add(0x9e3779b9u32.wrapping_mul(i as u32 + 1));
x = splitmix32(x);
r[i][lane] = x ^ seed[(i + 1) & 7];
}
}
let loads = p.loads_per_hash();
let mut addrs = vec![0u32; LANES * loads];
let mut nload = 0usize;
let reps = p.shadow_reps();
for _ in 0..ITERATIONS {
let sel = r[0];
let mut site = 0usize;
let shadow = (0..reps).flat_map(|_| p.shadow.iter());
for ins in p.instrs.iter().chain(shadow) {
step(ins, &mut r, &sel, w, mask, era.as_ref(), sites, &mut site, &mut addrs, &mut nload, loads);
}
}
let mut hashes = [0u64; LANES];
for lane in 0..LANES {
for i in 0..8 {
let v = r[i][lane];
*sat += (v == 0 || v == u32::MAX) as u32;
}
let lo = r[0][lane] ^ r[1][lane].rotate_left(7) ^ r[2][lane].rotate_left(14) ^ r[3][lane].rotate_left(21);
let hi = r[4][lane] ^ r[5][lane].rotate_left(9) ^ r[6][lane].rotate_left(18) ^ r[7][lane].rotate_left(27);
let h = ((hi as u64) << 32) | lo as u64;
hashes[lane] = h;
for j in 0..64 {
ones[j] += ((h >> j) & 1) as u32;
}
let sl = &mut addrs[lane * loads..(lane + 1) * loads];
sl.sort_unstable();
sl.dedup();
*lane_distinct += sl.len() as u64;
}
hashes
}
#[allow(clippy::too_many_arguments)]
#[inline(always)]
fn step(ins: &Instr, r: &mut [[u32; LANES]; 8], sel: &[u32; LANES], w: &Words, mask: u32, era: Option<&igneum_pow::generator::EraParams>, sites: &mut [Vec<u32>], site: &mut usize, addrs: &mut [u32], nload: &mut usize, loads: usize) {
let d = ins.dst as usize;
let a = ins.src as usize;
match ins.op {
Op::Add => {
let (imm, imm2, bit) = (ins.imm, ins.imm2, ins.bit as u32);
let src = r[a];
for lane in 0..LANES {
let c = if (sel[lane] >> bit) & 1 != 0 { imm2 } else { imm };
r[d][lane] = r[d][lane].wrapping_add(src[lane]).wrapping_add(c);
}
}
Op::Sub => { let src = r[a]; for lane in 0..LANES { r[d][lane] = r[d][lane].wrapping_sub(src[lane]); } }
Op::Mul => { let src = r[a]; for lane in 0..LANES { r[d][lane] = r[d][lane].wrapping_mul(src[lane]); } }
Op::MulHi => { let src = r[a]; for lane in 0..LANES { r[d][lane] = ((r[d][lane] as u64 * src[lane] as u64) >> 32) as u32; } }
Op::Xor => { let src = r[a]; for lane in 0..LANES { r[d][lane] ^= src[lane]; } }
Op::Or => { let src = r[a]; for lane in 0..LANES { r[d][lane] |= src[lane]; } }
Op::Rotl => { let n = ins.rot; for lane in 0..LANES { r[d][lane] = r[d][lane].rotate_left(n); } }
Op::Rotr => { let src = r[a]; for lane in 0..LANES { r[d][lane] = r[d][lane].rotate_right(src[lane] & 31); } }
Op::Mad => { let src = r[a]; let s2 = r[ins.src2 as usize]; for lane in 0..LANES { r[d][lane] = src[lane].wrapping_mul(s2[lane]).wrapping_add(r[d][lane]); } }
Op::Shfl => { let src = r[a]; let m = ins.mask as usize; for lane in 0..LANES { r[d][lane] ^= src[lane ^ m]; } }
Op::Load => {
assert_eq!(ins.width, 1);
for lane in 0..LANES {
let idx = load_index(era, ins, r[a][lane], mask, ACCEPT_DATASET_LOG2);
sites[*site].push(idx);
addrs[lane * loads + *nload] = idx;
r[d][lane] ^= w.word(idx);
}
*site += 1;
*nload += 1;
}
Op::WLoad | Op::Scratch | Op::Hot => panic!("not a class v4 op"),
}
}
pub struct Side {
pub name: &'static str,
pub distinct: Vec<u32>,
pub ratio: Vec<f64>,
pub distinct_mean_c: f64,
pub saturated_c: u32,
pub bias_max_c: u32,
}
/// Run the rule's units on one dataset side: the first `units` base nonces of the seed's acceptance stream
/// (the (c) metrics over the first ACCEPT_UNITS of them), threaded over units.
pub fn side(p: &Program, w: &Words, name: &'static str, units: usize, threads: usize) -> Side {
let n_sites = p.loads_per_hash() / ITERATIONS;
let bases = accept_base_nonces_n(&p.seed, units);
let next = AtomicUsize::new(0);
let parts: Vec<(Vec<Vec<u32>>, u64, u32, [u32; 64])> = std::thread::scope(|sc| {
let mut hs = Vec::new();
for _ in 0..threads {
let (next, bases) = (&next, &bases);
hs.push(sc.spawn(move || {
let mut sites: Vec<Vec<u32>> = vec![Vec::new(); n_sites];
let (mut ld, mut sat, mut ones) = (0u64, 0u32, [0u32; 64]);
let (mut ld_c, mut sat_c, mut ones_c) = (0u64, 0u32, [0u32; 64]);
loop {
let u = next.fetch_add(1, Ordering::Relaxed);
if u >= bases.len() { break; }
if u < ACCEPT_UNITS {
warp(p, w, bases[u], &mut sites, &mut ld_c, &mut sat_c, &mut ones_c);
} else {
warp(p, w, bases[u], &mut sites, &mut ld, &mut sat, &mut ones);
}
}
(sites, ld_c, sat_c, ones_c)
}));
}
hs.into_iter().map(|h| h.join().unwrap()).collect()
});
let mut sites: Vec<Vec<u32>> = vec![Vec::new(); n_sites];
let (mut ld_c, mut sat_c, mut ones_c) = (0u64, 0u32, [0u32; 64]);
for (s, l, sa, on) in parts {
for (i, v) in s.into_iter().enumerate() { sites[i].extend(v); }
ld_c += l; sat_c += sa;
for j in 0..64 { ones_c[j] += on[j]; }
}
let n = (units * LANES * ITERATIONS) as f64;
let mut distinct = Vec::new();
let mut ratio = Vec::new();
let mut k = 0usize;
for i in p.instrs.iter().filter(|i| i.op == Op::Load) {
let mut v = std::mem::take(&mut sites[k]);
v.sort_unstable();
v.dedup();
let wsize = ((1u64 << ACCEPT_DATASET_LOG2) >> (i.win as u64).min(2)) as f64;
distinct.push(v.len() as u32);
ratio.push(v.len() as f64 / (n - n * n / (2.0 * wsize)));
k += 1;
}
let half = (ACCEPT_UNITS * LANES / 2) as u32;
let bias_max_c = ones_c.iter().map(|&o| o.abs_diff(half)).max().unwrap_or(0);
Side { name, distinct, ratio, distinct_mean_c: ld_c as f64 / (ACCEPT_UNITS * LANES) as f64, saturated_c: sat_c, bias_max_c }
}
pub fn run(p: &Program, day: u64, threads: usize, zero_plant: bool) {
let t0 = std::time::Instant::now();
let live_ds = Epoch::chain_dataset_day(&day_bytes(day), ProgramClass::V4, 0, ACCEPT_DATASET_LOG2);
let closed_ds = DatasetSource::from_key(p.seed, DatasetMode::ClosedForm, ACCEPT_DATASET_LOG2);
let layout = p.class.layout();
let live = Words { ds: &live_ds, layout, zero: zero_plant };
let closed = Words { ds: &closed_ds, layout, zero: false };
println!("gap: program id {:016x} attempt {} class {} day {day} (cache filled in {:.1} s) plant zero-dataset {zero_plant}", p.program_id(), p.attempt, p.class.name(), t0.elapsed().as_secs_f64());
// validation: the mirror's hashes equal the library's on both datasets over 8 units
let bases = accept_base_nonces_n(&p.seed, 8);
let mut scratch: Vec<Vec<u32>> = vec![Vec::new(); 16];
let (mut ld, mut sat, mut ones) = (0u64, 0u32, [0u32; 64]);
let mut mism = 0;
for &b in &bases {
let hc = warp(p, &closed, b, &mut scratch, &mut ld, &mut sat, &mut ones);
if hc != hash_warp(p, b, &closed_ds) { mism += 1; }
if !zero_plant {
let hl = warp(p, &live, b, &mut scratch, &mut ld, &mut sat, &mut ones);
if hl != hash_warp(p, b, &live_ds) { mism += 1; }
}
}
assert_eq!(mism, 0, "the mirror disagrees with the library's hash_warp; nothing below is trusted");
println!("gap: mirror validated against hash_warp on 8 units (closed form{})", if zero_plant { "; live skipped under the plant" } else { " and live" });
let units = ACCEPT_UNITS_DISTINCT_V4;
let c = side(p, &closed, "closed", units, threads);
let lib = distinct_indices_v4(p, units).expect("library distinct pass");
assert_eq!(c.distinct, lib, "the mirror's closed-form distinct counts differ from the library's distinct_indices_v4");
println!("gap: closed-form per-site distinct counts equal the library's distinct_indices_v4 on {units} units");
let l = side(p, &live, "live", units, threads);
println!("gap: (c) metrics on {} units: closed distinct_mean {:.3} saturated {} bias_max {}; live distinct_mean {:.3} saturated {} bias_max {}", ACCEPT_UNITS, c.distinct_mean_c, c.saturated_c, c.bias_max_c, l.distinct_mean_c, l.saturated_c, l.bias_max_c);
let mut worst = (0f64, 0usize);
for s in 0..c.ratio.len() {
let d = c.ratio[s] - l.ratio[s];
if d.abs() > worst.0.abs() { worst = (d, s); }
println!("gap: site {s:2} closed ratio {:.4} ({}) live ratio {:.4} ({}) diff {:+.4}", c.ratio[s], c.distinct[s], l.ratio[s], l.distinct[s], d);
}
let cmin = c.ratio.iter().cloned().fold(9.0, f64::min);
let lmin = l.ratio.iter().cloned().fold(9.0, f64::min);
println!("gap: min ratio closed {cmin:.4} live {lmin:.4}; largest site gap {:+.4} at site {}; verdict under the 0.98 floor: closed {} live {}; {:.1} s", worst.0, worst.1, if cmin >= 0.98 { "accept" } else { "REJECT" }, if lmin >= 0.98 { "accept" } else { "REJECT" }, t0.elapsed().as_secs_f64());
println!("gap DONE {:016x}", p.program_id());
}
}
fn gap_cmd(args: &[String]) {
let get = |k: &str, d: &str| -> String { args.windows(2).find(|w| w[0] == k).map(|w| w[1].clone()).unwrap_or_else(|| d.to_string()) };
let seeds: Vec<u64> = get("--seeds", "2").split(',').map(|x| x.parse().unwrap()).collect();
let day: u64 = get("--day", "20730").parse().unwrap();
let threads: usize = get("--threads", "32").parse().unwrap();
let zero = args.iter().any(|a| a == "--plant") && get("--plant", "") == "zero-dataset";
for k in seeds {
let p = if k == 1 {
let g = unhex(GENESIS_HEX).unwrap();
Epoch::chain_program(&g, Some(&g), ProgramClass::V4, "p1-devnet-epoch0")
} else {
chain_v4(&format!("igneum-attack-f8/{k}"), &seed_bytes("epoch", k), &seed_bytes("era", k))
};
println!("gap: seed {k}");
gap::run(&p, day, threads, zero);
}
}