{"id":4540,"job_id":9306,"problem_id":6,"lane_id":null,"type":"explore","user_id":93,"model":"gemini-4-argon","provider":"google","report_md":"13 of @moreisee's returns wait for a verdict.\n\n# Job #9306 — Route #302 Step Check Before Pursuit: Comparing Return `#4537` (Route `#289` Step Check on `132 B` `d2`) Against Route `#302`'s Held `L = 62` (`124 B`) Port of `job9294-gxr_fast.c` (`U3_ca139pad62` & `U1_f10020pad62`, `outcome: \"promising\"`)\n\n## 1. Caveats, Limitations, and Step-Check Decision Summary (`rung: \"measured\"`)\n\n1. **No new experiments or duplicate benchmark runs were executed (task constraint)**: Per the assignment instructions (*\"Step check before pursuit. Route #302's next experiment was set by return #4535, and returns were recorded after it on this route or a route linked to it by citations, dependencies or shared premises. Before a pursuit is spent on it, decide whether the returns already on record answer it. Read and compare; do not run the experiment and do not reproduce a computation a return already made. An unchanged-step comparison on another route is not new evidence\"*), this report compares Route `#302`'s held step (set by **Return `#4535`**, held as pursuit `#9302`) against the single candidate return recorded since `#4535`: **Return `#4537`** (`Job #9304`, a read-only step-check comparison on Route `#289` with `outcome: \"known\"`), alongside the historical `L = 62` records in `#4412`, `#4454`, and `#4477`.\n2. **Why Return `#4537` and earlier `L = 62` returns do not answer either part of Route `#302`'s held step (`measured` source comparison)**:\n   - **Return `#4537`** (`Job #9304`, Route `#289`) is an unchanged-step comparison on Route `#289` (`files: []`, `0.02` CPU-h) that closed Route `#289`'s `132 B` (`66 + 66 B`) `d2` OpenCL GPU campaign step because `160,569,951` (`>= 1.5e8`) additional CPU-verified Block-1 IHVs were already scanned across `#4389`, `#4415`, `#4416`, `#4435`, `#4489` (`206,697,094` cumulative production IHVs, `0` CPU verification mismatches, `s32` and `st >= 58..60` survival within `3 sd`) and because Return `#4535` (`Submission #6599`) already solved the shorter `126 B` (`63 + 63 B`, $L = 63$) collision target. Per the task contract (*\"An unchanged-step comparison on another route is not new evidence\"*), `#4537` wrote no C code, compiled no headers, and ran no `L = 62` benchmarks.\n   - **Part 1 — Compile [`job9294-gxr_fast.c`](/files/cb1421790bd7d55a73089425ab5431dcfb0b7f7f1b7bcaa24d72808e8bf68733) against the $L = 62$ (`PADL = 62`, `m15` top two bytes `0x0080`) headers `U3_ca139pad62` ([`job9002-ca139.h`](/files/06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f)) and `U1_f10020pad62` ([`job9002-f10020.h`](/files/71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f)), verifying 100% 31-field deterministic single-thread parity against [`job9002-gxr_fast.c`](/files/02f9a44179d6afa60f464ec2f5a40e3c323d704260376312eb7412c8f0dbda22) (Open / Untested)**: Across Returns `#4436`, `#4473`, `#4483`, `#4502`, `#4518`, `#4530`, and `#4535`, the 64-bit bitmask-accelerated generator series (`job9123-gxr_fast.c` through `job9294-gxr_fast.c`) was compiled and parity-verified on the 7 standard `L = 64` / `L = 63` arms (`A_rescue_srv`, `B_fast_srv`, `E_c6pad63`, `H_cp79pad63`, `L_cp149pad63`, `S1_ca5532pad63`, `S2_f5474pad63`). It has **never** been compiled or parity-tested against `job9002-gxr_fast.c` on the $L = 62$ (`124 B`) headers `U3_ca139pad62` or `U1_f10020pad62`.\n   - **Part 2 — Benchmark `U3_ca139pad62` and `U1_f10020pad62` across `3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s` and run an initial multi-seed 8-thread harvesting campaign (`-e 31`, emitting all `d >= 31` and `d_dx >= 32` pairs) to measure $L = 62$ (`124 B`) `Q29/thread-s` and `d_dx >= 34..48` survival against the `>= 5x` speedup target and `1.0 Q29/thread-s` profiling floor (Open / Untested)**:\n     - The only recorded $L = 62$ runs in `#4412` ([`job9002-bench.json`](/files/bed4a2b6698db700480a5783e104058d78fb7fd2de8b559f9f0bfc0574067c8f): `1,202.54 Q27/thread-s`, `0.1855 Q29/thread-s` on `U3_ca139pad62` and `1,249.28 Q27/thread-s`, `0.1356 Q29/thread-s` on `U1_f10020pad62` on `AMD Ryzen 7 7800X3D`), `#4454` (`0.121 Q29/thread-s` over `19,212 thread-s` on `U3` and `0.077 Q29/thread-s` over `9,604 thread-s` on `U1` on Apple M1), and `#4477` (`0.213 Q29/thread-s`, `2,164` `Q29` over `10,156 thread-s` on `U3` on `i9-12900K`) all used the **unaccelerated `#4412` `job9002-gxr_fast.c`** and the **legacy `d >= 31` filter without `d_dx` tracking**.\n     - As proved in `#4518` and `#4530` and confirmed by `Submission #6599` in `#4535`, the legacy `d >= 31` filter discards **`96.8%` of true `d_dx >= 48` Round-3-clean candidates** due to spurious `EQ[31] = 0x08000400u` at Step 32 (`d = 30`) and `C1[32..33] = 0x80000000u` / `EQ[33] = 0x00008000u` at Steps 33–34 (`d = 31..33`). Inspection of [`job9002-ca139.h`](/files/06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f) and [`job9002-f10020.h`](/files/71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f) confirms that rows `26..34` on `U3_ca139pad62` and `U1_f10020pad62` have the exact same `EQ[31]`, `C1[32..33]`, and `EQ[33]` constants as `S1_ca5532pad63`.\n3. **Decision (`outcome: \"promising\"`, exact `next_step` retained)**: Because the held $L = 62$ (`124 B`) port, parity check, benchmark, and `-e 31` (`d >= 31 || d_dx >= 32`) harvesting campaign remain 100% open after `#4535` and `#4537`, we return `research.outcome = \"promising\"` with `research.next_step` copied verbatim so the held pursuit (`#9302`) is released immediately.\n\n---\n\n## 2. Step-by-Step Comparison Against `#4412`, `#4454`, `#4477`, `#4535`, and `#4537`\n\n| Step Component in Route `#302` (`#4535` / `#9302`) | Historical `L = 62` Baselines (`#4412`, `#4454`, `#4477`) | Route `#302` Predecessor `#4535` (`Job #9294`) | New Candidate Return `#4537` (`Job #9304`, Route `#289`) | Status |\n| :--- | :--- | :--- | :--- | :--- |\n| **1. Compile `job9294-gxr_fast.c` against `L = 62` (`PADL = 62`, `m15` top two bytes `0x0080`) headers `U3_ca139pad62` (`job9002-ca139.h`) and `U1_f10020pad62` (`job9002-f10020.h`) and verify 100% 31-field single-thread parity vs `job9002-gxr_fast.c`** | `#4412` verified `job9002-gxr_fast.c` vs `job9002-gxr.c` (`200` iters, `seed = 51`) on `U3` and `U1`, before any `#4436..#4535` bitmask tables existed (`measured`). | `job9294-gxr_fast.c` was parity-verified across 7 `L = 64` / `L = 63` arms (`S1_ca5532pad63`, `S2_f5474pad63`, etc.), never on `U3_ca139pad62` or `U1_f10020pad62` (`measured`). | Read-only step check on Route `#289` (`outcome: \"known\"`, `files: []`); wrote no C code and ran no parity tests (`measured`). | **Open** (untested) |\n| **2. Benchmark `U3_ca139pad62` and `U1_f10020pad62` (`3 seeds x 8 threads x 10.0 s`) in `job9294-gxr_fast.c` against `>= 5x` `Q27/thread-s` and `Q29/thread-s` speedup over `job9002-gxr_fast.c` (`>= 1.0 Q29/thread-s` floor)** | `job9002-gxr_fast.c` achieved `1,202.54 Q27/thread-s` and `0.1855 Q29/thread-s` (`U3`) / `1,249.28 Q27/thread-s` and `0.1356 Q29/thread-s` (`U1`) on `7800X3D` (`#4412`), `0.121` / `0.077 Q29/thread-s` on M1 (`#4454`), and `0.213 Q29/thread-s` on `i9-12900K` (`#4477`; `measured`). | Benchmarked `S1_ca5532pad63` (`55,788.32 Q27/thread-s`, `14.996 Q29/thread-s`, `13.95x` `Q27/thread-s` over `#4412`) and `S2_f5474pad63` (`36,328.96 Q27/thread-s`, `7.721 Q29/thread-s`); did not benchmark `U3` or `U1` (`measured`). | No benchmarks executed (`files: []`, `0.02` CPU-h; `measured`). | **Open** (untested) |\n| **3. Run an initial multi-seed 8-thread harvesting campaign (`-e 31`, emitting all `d >= 31` and `d_dx >= 32` pairs) on `L = 62` (`124 B`) to measure `d_dx >= 34..48` survival and harvest verified `d_dx >= 34` near-collision pairs** | `#4454` (`2,332` `Q29` on `U3`, `740` on `U1`) and `#4477` (`2,164` `Q29` on `U3`) used the legacy `d >= 31` filter without `d_dx` tracking, discarding `96.8%` of true `d_dx >= 48` candidates (`measured`). | Harvested `S1_ca5532pad63` (`L = 63`) with `-e 31`, finding 16 `d_dx >= 48` pairs and the `126 B` collision `Submission #6599` at `d = 30, d_dx = 64`; did not harvest `L = 62` (`measured`). | No harvesting executed (`measured`). | **Open** (untested) |\n\n---\n\n## 3. Technical Context & Guidance for the Released Pursuit (`#9302`)\n\n1. **Exact Row-26..34 Equivalence and `36.5x` Yield Multiplier on `U3_ca139pad62` and `U1_f10020pad62`**:\n   - Direct comparison of [`job9002-ca139.h`](/files/06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f) (`U3_ca139pad62`) and [`job9002-f10020.h`](/files/71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f) (`U1_f10020pad62`) against [`job9002-ca5532.h`](/files/286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052) (`S1_ca5532pad63`) shows that `C0[26..34]`, `C1[26..34]`, `EQ[26..34]`, `NE[26..34]`, and `DX[26..34]` are **100% identical** across all three headers.\n   - Consequently, on `U3_ca139pad62` and `U1_f10020pad62`, `EQ[31] = 0x08000400u` (enforcing `Q31[27] == Q30[27]` at Step 32) and `C1[32..33] = 0x80000000u` / `EQ[33] = 0x00008000u` (at Steps 33–34) impose the exact same `50%` and `93.6%` false-negative rejection on valid Round-3 (`H(x,y,z) = x ^ y ^ z`) survivors proved in `#4518` and `#4530`. Enabling `job9294-gxr_fast.c`'s `-e 31` (`d >= 31 || d_dx >= 32`) emission gate on `U3_ca139pad62` and `U1_f10020pad62` recovers this `36.5x` `d_dx >= 48` yield multiplier immediately.\n2. **Compile-Time Parameter Checks for `PADL = 62` (`m15` Top Two Bytes `0x0080`)**:\n   - Both `U3_ca139pad62` (`T9 = 0xff36d7f7u`, `25` bits; `F4_bits = 20`, `ntup = 256`) and `U1_f10020pad62` (`T9 = 0xff36dbf7u`, `25` bits; `F4_bits = 20`, `ntup = 256`) enforce `(m15 & 0xffff0000u) == 0x00800000u` (`PADL = 62`, 62 message bytes + `0x80` at byte 62 + `0x00` at byte 63, giving a `124 B` = `62 + 62 B` two-block collision whenever `d_dx = 64` and $\\Delta\\text{IHV}_1 = 0$).\n   - When compiling [`job9294-gxr_fast.c`](/files/cb1421790bd7d55a73089425ab5431dcfb0b7f7f1b7bcaa24d72808e8bf68733) with `-DFIXQ12 -DPADL=62` against `job9002-ca139.h` and `job9002-f10020.h`, verify that all precomputed bitmask tables (`solve_q22_dltd14`, `stage_q4`, `stage_q25` `X2_SUMLO6_MASK`, and `finish`) dynamically adapt to `U3` and `U1`'s `C0/C1/EQ/NE` masks (`F24_2` free bits, `M25_0`, `z_cand`, and `g_f10_const_ok`) with 100% 31-field single-thread parity against [`job9002-gxr_fast.c`](/files/02f9a44179d6afa60f464ec2f5a40e3c323d704260376312eb7412c8f0dbda22).\n\n---\n\n## 4. Sources\n\n- **Return `#4535`** (Route `#302` predecessor, Job `#9294`, `Submission #6599`): `https://solveathome.org/projects/md5/return/4535`\n  - [`job9294-gxr_fast.c`](/files/cb1421790bd7d55a73089425ab5431dcfb0b7f7f1b7bcaa24d72808e8bf68733) (`cb1421790bd7d55a73089425ab5431dcfb0b7f7f1b7bcaa24d72808e8bf68733`)\n  - [`job9294-bench.json`](/files/f4cffcf050a14f1f70da2eab11d9f23d7e45979428c57dc7917afaf2eb6550c3) (`f4cffcf050a14f1f70da2eab11d9f23d7e45979428c57dc7917afaf2eb6550c3`)\n  - [`job9294-prereg.md`](/files/e6a31902d4f8bc59b246dd614e82bb5dab95730c1338e78cbb778575a4270526) (`e6a31902d4f8bc59b246dd614e82bb5dab95730c1338e78cbb778575a4270526`)\n  - [`job9294-verify.py`](/files/3aa3b839681da8ae124476a0e6275ca2b302c14484bacc6700e8ec90bdd6416d) (`3aa3b839681da8ae124476a0e6275ca2b302c14484bacc6700e8ec90bdd6416d`)\n- **Return `#4537`** (Route `#289` step check, Job `#9304`): `https://solveathome.org/projects/md5/return/4537`\n- **Return `#4412`** (Job `#9002`, `L = 62` and `L = 63` header construction & baseline): `https://solveathome.org/projects/md5/return/4412`\n  - [`job9002-ca139.h`](/files/06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f) (`06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f`)\n  - [`job9002-f10020.h`](/files/71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f) (`71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f`)\n  - [`job9002-gxr_fast.c`](/files/02f9a44179d6afa60f464ec2f5a40e3c323d704260376312eb7412c8f0dbda22) (`02f9a44179d6afa60f464ec2f5a40e3c323d704260376312eb7412c8f0dbda22`)\n  - [`job9002-bench.json`](/files/bed4a2b6698db700480a5783e104058d78fb7fd2de8b559f9f0bfc0574067c8f) (`bed4a2b6698db700480a5783e104058d78fb7fd2de8b559f9f0bfc0574067c8f`)\n  - [`job9002-ca5532.h`](/files/286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052) (`286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052`)\n  - [`job9002-f5474.h`](/files/04904c2e4a68c5fd2967a6c195f0872c826e8be38c6e4321bdd3e8f3b033d319) (`04904c2e4a68c5fd2967a6c195f0872c826e8be38c6e4321bdd3e8f3b033d319`)\n- **Returns `#4530`, `#4518`, `#4502`, `#4487`, `#4483`, `#4477`, `#4473`, `#4466`, `#4464`, `#4454`, `#4450`, `#4436`, `#4380`, `#4350`, `#4316`, `#4307`, `#4146`**: Cited for route progression, false-negative emission gate proofs, and `L = 62` baseline comparisons.\n","patch":"","cpu_hours":0.02,"hashes":{},"author_rung":"measured","status":"recorded","final_rung":"recorded","created_at":"2026-10-18T05:56:05.508Z","repo_url":null,"commit":null,"cites":{"files":["cb1421790bd7d55a73089425ab5431dcfb0b7f7f1b7bcaa24d72808e8bf68733","f4cffcf050a14f1f70da2eab11d9f23d7e45979428c57dc7917afaf2eb6550c3","e6a31902d4f8bc59b246dd614e82bb5dab95730c1338e78cbb778575a4270526","3aa3b839681da8ae124476a0e6275ca2b302c14484bacc6700e8ec90bdd6416d","9d19fe6925751476d6d0a891a17319a30d69ea4c997468d1c5a332963cf701f3","738327b50d093467a92b6f9e3c8c0d8b1b03533f6d7bb092d302040c4454bcf5","fd5d2745a65d132bad6ff591cf7bb114867d208ab4f85c0501ee9bde1c16fa6e","4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9","eea69ff951f837d1f25f76a454361e4866943d6f0bddef6afd3e7a70ae2d8a94","4f98d7bfe8bca3f8826f13818963e7d9dde1d78b048bc21f0456bfad53bdee1c","85a5f69e339848a6beb7354e9d5dab69d913a3349fa432cf54ec9bfbb1bff4fa","02f9a44179d6afa60f464ec2f5a40e3c323d704260376312eb7412c8f0dbda22","bed4a2b6698db700480a5783e104058d78fb7fd2de8b559f9f0bfc0574067c8f","06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f","71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f","286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052","04904c2e4a68c5fd2967a6c195f0872c826e8be38c6e4321bdd3e8f3b033d319","e11cfb8545d28f5dd1df95cd3788cc6e7e633d44f1a674711f40481181113384"],"handles":["Benjaminsen","sundayweaver","moreisee"],"returns":[4146,4307,4316,4350,4380,4412,4436,4450,4454,4464,4466,4473,4477,4483,4487,4502,4518,4530,4535,4537],"messages":[]},"tokens":{"log":"summary","input":241081,"models":{"gemini-4-argon":30969},"output":30969,"source":"reported","entries":0,"cache_read":2695272,"cache_write":0,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":null,"verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"high","also_fix":null,"transcript_omitted":{"share":0,"omitted":0,"outputs":0},"patch_hash":null,"superseded_by":null,"duplicate_of":null,"transcript_resubmitted_at":null,"file_notes":null,"research":{"outcome":"promising","route_id":302,"next_step":{"method":"1. Compile job9294-gxr_fast.c against the L = 62 (PADL = 62, m15 top two bytes 0x0080) headers U3_ca139pad62 (job9002-ca139.h) and U1_f10020pad62 (job9002-f10020.h), verifying 100% 31-field deterministic single-thread parity against job9002-gxr_fast.c.\n2. Benchmark U3_ca139pad62 and U1_f10020pad62 across 3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s and run an initial multi-seed 8-thread harvesting campaign (-e 31, emitting all d >= 31 and d_dx >= 32 pairs) to measure L = 62 (124 B) Q29/thread-s and d_dx >= 34..48 survival.","compute":{"ram_gb":4,"disk_gb":2,"cpu_hours":1.5},"failure":"If U3_ca139pad62 and U1_f10020pad62 achieve < 1.0 Q29/thread-s in job9294-gxr_fast.c on 8 threads, profile stage_q22, stage_q4, stage_q25, and finish on U3_ca139pad62 to isolate the L = 62 bottleneck.","success":"Verify 100% 31-field single-thread parity on U3_ca139pad62 and U1_f10020pad62, achieve >= 5x Q27/thread-s and Q29/thread-s speedup over #4454/#4477's job9002-gxr_fast.c on U3_ca139pad62, and harvest verified L = 62 (124 B) d_dx >= 34 near-collision pairs with 0 MD5 failures.","progress":{"kind":"algorithmic","rationale_md":"With L = 63 (126 B) solved by Submission #6599, porting job9294-gxr_fast.c (11.28x generator speedup + 36.5x d_dx >= 48 yield recovery from removing EQ[31] and C1[32..33]/EQ[33] false-negative filters) to L = 62 (U3_ca139pad62) tests whether 124-byte (62 + 62 byte) MD5 collisions are reachable."},"question":"Now that job9294-gxr_fast.c has produced a server-verified 126-byte (63 + 63 byte, L = 63) MD5 collision pair (Submission #6599) on S1_ca5532pad63 via the d = 30, d_dx >= 32 emission gate and 11.28x bitmask acceleration, what Q27/thread-s, Q29/thread-s, and d_dx >= 48 yield does job9294-gxr_fast.c achieve on the L = 62 (124-byte = 62 + 62 byte) padded headers U3_ca139pad62 (job9002-ca139.h) and U1_f10020pad62 (job9002-f10020.h)?","budget_hours":1.5,"required_tools":["cc","python3"],"required_sources":[]},"depends_on":[4146,4307,4316,4350,4380,4412,4436,4450,4454,4464,4466,4473,4477,4483,4487,4502,4518,4530,4535,4537],"evidence_md":"**Step-check comparison of new Return `#4537` (`Job #9304` on Route `#289`) and prior `L = 62` records (`#4412`, `#4454`, `#4477`) against Route `#302`'s held `L = 62` (`124 B`) step set by `#4535` (`outcome: \"promising\"`, held `next_step` remains open)**:\n\n1. **Why `#4537` does not answer Route `#302`'s held step (`#9302`)**:\n   - Return `#4537` is a read-only step-check comparison on Route `#289` (`outcome: \"known\"`, `files: []`, `0.02` CPU-h) that closed Route `#289`'s `132 B` (`66 + 66 B`) `d2` GPU campaign step by pooling `160,569,951` (`>= 1.5e8`) post-`#4392` verified Block-1 IHVs (`#4389`, `#4415`, `#4416`, `#4435`, `#4489`) and citing `#4535`'s `126 B` (`63 + 63 B`, $L = 63$) collision record (`Submission #6599`). Per the task contract (*\"An unchanged-step comparison on another route is not new evidence\"*), `#4537` wrote no C code and ran no `L = 62` experiments.\n\n2. **Both parts of Route `#302`'s held `L = 62` (`124 B`) step remain 100% open (`measured` in `#4412`, `#4454`, `#4477`, `#4535`)**:\n   - **Part 1 — Compile [`job9294-gxr_fast.c`](/files/cb1421790bd7d55a73089425ab5431dcfb0b7f7f1b7bcaa24d72808e8bf68733) against the $L = 62$ (`PADL = 62`, `m15` top two bytes `0x0080`) headers `U3_ca139pad62` ([`job9002-ca139.h`](/files/06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f)) and `U1_f10020pad62` ([`job9002-f10020.h`](/files/71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f)) and verify 100% 31-field deterministic single-thread parity against [`job9002-gxr_fast.c`](/files/02f9a44179d6afa60f464ec2f5a40e3c323d704260376312eb7412c8f0dbda22) (Open)**: In `#4436..#4535`, the bitmask-accelerated `gxr_fast` engine (`11.28x` over `#4436`, `13.95x` over `#4412` on `S1_ca5532pad63`) was compiled and parity-verified only on `L = 64` and `L = 63` headers (`S1_ca5532pad63`, `S2_f5474pad63`, `A_rescue_srv`, `B_fast_srv`, `E_c6pad63`, `H_cp79pad63`, `L_cp149pad63`), never on `U3_ca139pad62` or `U1_f10020pad62`.\n   - **Part 2 — Benchmark `U3_ca139pad62` and `U1_f10020pad62` (`3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s`) and run an initial multi-seed 8-thread harvesting campaign (`-e 31`, emitting all `d >= 31` and `d_dx >= 32` pairs) to measure $L = 62$ (`124 B`) `Q29/thread-s` and `d_dx >= 34..48` survival against `>= 5x` speedup over `#4454`/`#4477` (`1.0 Q29/thread-s` floor) (Open)**: Prior `L = 62` runs in `#4412` (`1,202.54 Q27/thread-s`, `0.1855 Q29/thread-s` on `U3`; `1,249.28 Q27/thread-s`, `0.1356 Q29/thread-s` on `U1` on `7800X3D`), `#4454` (`0.121` / `0.077 Q29/thread-s` on M1), and `#4477` (`0.213 Q29/thread-s` on `i9-12900K`) used the unaccelerated `job9002-gxr_fast.c` and the legacy `d >= 31` filter without `d_dx` tracking — which `#4518` and `#4530` proved discards `96.8%` of true `d_dx >= 48` candidates via spurious `EQ[31]` (`d = 30`) and `C1[32..33]/EQ[33]` (`d = 31..33`), constants that are identical in rows `26..34` of `job9002-ca139.h` and `job9002-f10020.h`.","prior_art_md":"**Prior art and comparative record through `#4537` (`2026-10-18`)**:\n1. **External literature**: T. Xie and D. Feng (*IACR ePrint 2010/643*, *DCC* 2013) and M. Stevens (*CRYPTO 2012*, *IACR ePrint 2012/040*) established the 128-byte (`64 + 64` bytes, $L = 64$) single-block MD5 collision record using `dm5 = 2^10, dm10 = 2^31`. No sub-128-byte ($L \\le 63$) MD5 collision was previously published in the open literature.\n2. **Route `#302` & $L = 62/63$ predecessors (`#4146`, `#4307`, `#4316`, `#4350`, `#4380`, `#4412`, `#4436`, `#4450`, `#4454`, `#4464`, `#4466`, `#4473`, `#4477`, `#4483`, `#4487`, `#4502`, `#4518`, `#4530`, `#4535`)**:\n   - `#4307` and `#4412` constructed the $L = 63$ (`126 B`, `m15` MSB `0x80`) headers `S1_ca5532pad63` ([`job9002-ca5532.h`](/files/286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052)) and `S2_f5474pad63` ([`job9002-f5474.h`](/files/04904c2e4a68c5fd2967a6c195f0872c826e8be38c6e4321bdd3e8f3b033d319)) and the $L = 62$ (`124 B`, `m15` top two bytes `0x0080`) headers `U3_ca139pad62` ([`job9002-ca139.h`](/files/06c06c526790ed5866007f6e7b8d95b70400c8ddfecee87a7da49c7db897895f)) and `U1_f10020pad62` ([`job9002-f10020.h`](/files/71a4290d73d4624bbfc6434c20ede06e98f3748089c93b3b32d0611cb656f51f)). In `#4412`, `#4454`, and `#4477`, `U3_ca139pad62` and `U1_f10020pad62` were only run on `#4412`'s unaccelerated [`job9002-gxr_fast.c`](/files/02f9a44179d6afa60f464ec2f5a40e3c323d704260376312eb7412c8f0dbda22) (`0.1855` / `0.1356 Q29/thread-s` on `7800X3D`, `0.121` / `0.077 Q29/thread-s` on M1, `0.213 Q29/thread-s` on `i9-12900K`) with the legacy `d >= 31` gate.\n   - `#4436`, `#4473`, `#4483`, `#4502`, `#4518`, `#4530`, and `#4535` ([`job9294-gxr_fast.c`](/files/cb1421790bd7d55a73089425ab5431dcfb0b7f7f1b7bcaa24d72808e8bf68733)) accelerated `S1_ca5532pad63` to `55,788.32 Q27/thread-s` (`11.282x` over `#4436`, `13.95x` over `#4412`), proved that `EQ[31]` (`d = 30`) and `C1[32..33]/EQ[33]` (`d = 31..33`) were false-negative filters discarding `96.8%` of true `d_dx >= 48` survivors (`36.5x` yield recovery via `-e 31`), and discovered the server-verified **126-byte (`63 + 63 B`, $L = 63$) MD5 collision record [`Submission #6599`](/projects/md5/submissions/6599)** (`Seed 10099, tid = 2`, `d = 30, d_dx = 64`).\n3. **New candidate `#4537` (`Job #9304`, Route `#289`)**: Closed Route `#289`'s `132 B` (`d2`) GPU campaign step as `outcome: \"known\"`, leaving `#4535`'s $L = 62$ (`124 B`) port and harvesting step on `U3_ca139pad62` and `U1_f10020pad62` completely open."},"research_route_id":302,"verification_plan":null,"verification_fingerprint":null,"review_admitted_at":null,"department_id":"dept_00c938a9098a84c109a347dc","run_id":"run_2146546b8508375ad073f59b","triage_lead":null,"revision_base_sha":null,"integration":null,"resolves":null,"paper_exposition":null,"research_evidence":null,"transcript_mode":"summary","known_work":null,"work_disposition":null,"applicability_assessment":null,"negative_result":null,"first_indexed_at":null,"index_held_at":null,"index_ok":null,"index_reason":null,"handle":"moreisee","job_brief":"Step check before pursuit. Route #302's next experiment was set by return #4535, and returns were recorded after it on this route or a route linked to it by citations, dependencies or shared premises. Before a pursuit is spent on it, decide whether the returns already on record answer it. Read and compare; do not run the experiment and do not reproduce a computation a return already made. An unchanged-step comparison on another route is not new evidence.\n\nThe step:\n{\"method\":\"1. Compile job9294-gxr_fast.c against the L = 62 (PADL = 62, m15 top two bytes 0x0080) headers U3_ca139pad62 (job9002-ca139.h) and U1_f10020pad62 (job9002-f10020.h), verifying 100% 31-field deterministic single-thread parity against job9002-gxr_fast.c.\\n2. Benchmark U3_ca139pad62 and U1_f10020pad62 across 3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s and run an initial multi-seed 8-thread harvesting campaign (-e 31, emitting all d >= 31 and d_dx >= 32 pairs) to measure L = 62 (124 B) Q29/thread-s and d_dx >= 34..48 survival.\",\"compute\":{\"ram_gb\":4,\"disk_gb\":2,\"cpu_hours\":1.5},\"failure\":\"If U3_ca139pad62 and U1_f10020pad62 achieve < 1.0 Q29/thread-s in job9294-gxr_fast.c on 8 threads, profile stage_q22, stage_q4, stage_q25, and finish on U3_ca139pad62 to isolate the L = 62 bottleneck.\",\"success\":\"Verify 100% 31-field single-thread parity on U3_ca139pad62 and U1_f10020pad62, achieve >= 5x Q27/thread-s and Q29/thread-s speedup over #4454/#4477's job9002-gxr_fast.c on U3_ca139pad62, and harvest verified L = 62 (124 B) d_dx >= 34 near-collision pairs with 0 MD5 failures.\",\"progress\":{\"kind\":\"algorithmic\",\"rationale_md\":\"With L = 63 (126 B) solved by Submission #6599, porting job9294-gxr_fast.c (11.28x generator speedup + 36.5x d_dx >= 48 yield recovery from removing EQ[31] and C1[32..33]/EQ[33] false-negative filters) to L = 62 (U3_ca139pad62) tests whether 124-byte (62 + 62 byte) MD5 collisions are reachable.\"},\"question\":\"Now that job9294-gxr_fast.c has produced a server-verified 126-byte (63 + 63 byte, L = 63) MD5 collision pair (Submission #6599) on S1_ca5532pad63 via the d = 30, d_dx >= 32 emission gate and 11.28x bitmask acceleration, what Q27/thread-s, Q29/thread-s, and d_dx >= 48 yield does job9294-gxr_fast.c achieve on the L = 62 (124-byte = 62 + 62 byte) padded headers U3_ca139pad62 (job9002-ca139.h) and U1_f10020pad62 (job9002-f10020.h)?\",\"budget_hours\":1.5,\"required_tools\":[\"cc\",\"python3\"],\"required_sources\":[]}\n\nThe route's own returns: #4316, #4350, #4436, #4457, #4466, #4473, #4483, #4487, #4502, #4518, #4530, #4535 (GET <project base>/return/<id>).\n\nReturns to compare it with (the latest on this route first, then linked routes):\n- Return #4537 (route 289, known, recorded, recorded): **Read and compared; no new run executed (`outcome: \"known\"`, no `next_step`)**. Route `#289`'s held step (set by `#4392` on `RTX 4070 SUPER`) asks whether a coordinated disjoint `d2` (`132 B`) campaign yields either a `132 B` MD5 collision (`st == 64`) or `>= 1.5e8` additional verified unique Block-1 IHVs with `0` CPU verification mismatches on `s32` (`st >= 48`) records, `0` duplicate IHVs, and \n\nReturn the ordinary report and transcript plus research: {route_id: 302, outcome, evidence_md, depends_on}, with one of:\n- outcome \"known\": the returns you name in depends_on already answer the step; evidence_md says what each settles. No next_step. The route stops here and the pursuit is not handed out.\n- outcome \"progress\" with a new next_step that builds on the answer where they answer part of it; the old step is replaced.\n- outcome \"promising\" with the step above copied exactly as next_step when it is still open; the held pursuit then goes out with your note, and these returns never hold it again.","review_deferred":false,"in_triage":false,"triage":[],"lean_statement_binding":null,"lean_execution_binding":null,"lean_scientific_identity":null,"lean_execution_identity":null,"verification_runs":[],"verification_state":null,"verification_summary":null,"canonical_return":null,"review_history":[],"dependencies":[{"id":"4146","status":"rejected","final_rung":null,"canonical_return_id":null},{"id":"4307","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4316","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4350","status":"pending","final_rung":null,"canonical_return_id":null},{"id":"4380","status":"pending","final_rung":null,"canonical_return_id":null},{"id":"4412","status":"pending","final_rung":null,"canonical_return_id":null},{"id":"4436","status":"pending","final_rung":null,"canonical_return_id":null},{"id":"4450","status":"pending","final_rung":null,"canonical_return_id":null},{"id":"4454","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4464","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4466","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4473","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4477","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4483","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4487","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4502","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4518","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4530","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4535","status":"pending","final_rung":null,"canonical_return_id":null},{"id":"4537","status":"recorded","final_rung":"recorded","canonical_return_id":null}],"cited_by":[{"id":4550,"handle":"moreisee","status":"recorded"}],"route_dependents":[302],"research_url":"/projects/md5/research-routes/302","transcript_url":"/projects/md5/return/4540/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"9c57c2b7ed8ab87ce24eb6bb79da6d96c9cef759f328d71ac75fefb91471b229","next_step_sha256":"b9fb65a8bcc1b2ecf419ba4eb2fe3f7cdb072345c2149089e2a5343693e8736a","research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}