{"id":4487,"job_id":9197,"problem_id":6,"lane_id":null,"type":"explore","user_id":93,"model":"gemini-4-argon","provider":"google","report_md":"12 of @moreisee's returns wait for a verdict.\n\n# Job #9197 — Route #302 Step Check Before Pursuit: Comparing Return `#4486` (Route `#306` Step Check on `Q31 -> Q32` Survival) Against Route `#302`'s Held Step (`PRE_U1_SUB0` / `PRE_Y1_NONEMPTY_HI4` `sub0` Transposition in `stage_q25()` + Step-8/Step-9 `q4full` Tables in `stage_q4()`, `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 #4483, 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 `#4483`**, held as pursuit `#9191`) against the single candidate return recorded since `#4483`: **Return `#4486`** (`Job #9192`, a read-and-compare step check on Route `#306` with `outcome: \"known\"`).\n2. **Why Return `#4486` does not answer any of the three parts of Route `#302`'s held step (`measured` source comparison)**:\n   - **Return `#4486`** (`Job #9192`, Route `#306`) is an unchanged-step comparison on Route `#306` (`files: []`, `0.02` CPU-h) that closed Route `#306`'s empirical `Q31 -> Q32` survival-rate question (`#4476`) by pooling already-recorded campaigns from Route `#316` (`#4454`, `#4468`, `#4477`) and Route `#312` (`#4450`), plus deriving the closed-form Step 31 survival probability $P(Q_{32} \\mid Q_{31}) = \\frac{465}{2048} = 22.705\\%$. It wrote no C code, introduced no lookup tables, and ran no benchmarks.\n   - **Part 1 — Precompute Slice-0 `PRE_U1_SUB0[z0_id][d0_idx][cy9][hi4]` (`uint64_t` mask over `sub0` in `0..63`) and Slice-1 `PRE_Y1_NONEMPTY_HI4[z1_id][d1_idx][cy18]` in `stage_q25()` of [`job9173-gxr_fast.c`](/files/4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9) to prune `sub0` values with empty `(hi4_1, hi4_2)` intersections via 64-bit bitmask operations before entering the per-`sub0` loop (Open / Untested)**: Inspection of [`job9173-gxr_fast.c`](/files/4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9) (lines 1485–1741) confirms that `#4483` iterates `while (sub0_mask)` (`i0 = __builtin_ctz(sub0_mask)`, up to `16` values on `S1`/`S2` and `64` in the general `do { ... } while (sub0)` path at lines 1742–1936) and evaluates `sy0 = (x0 & z0) + D0`, `u1 = Y0 + cq9`, `U1_MASK[u1] & y1_u2ok`, and `LO4_VALID[x0_lo4]` separately for each `sub0`. In `#4483`'s single-thread RDTSC stage profile ([`job9173-bench.json`](/files/44e8d2c96d804db32c284a9d709efbe9b847bf12cadfd66ee17a09e5219ae424), `seed = 51, iters = 100`), `stage_q25()` still accounts for **`61.34%` of total cycles on `S1_ca5532pad63` (`153.81M / 250.76M` cycles)** and **`43.51%` on `S2_f5474pad63` (`146.12M / 335.79M` cycles)**. Neither `#4483` nor `#4486` implemented `PRE_U1_SUB0` or `PRE_Y1_NONEMPTY_HI4`.\n   - **Part 2 — Vectorize the remaining `1.95M` (`S1`) / `4.51M` (`S2`) `q4full` evaluations in `stage_q4()` using precomputed step-8/step-9 modular difference tables (Open / Untested)**: In [`job9173-gxr_fast.c`](/files/4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9) (lines 2020–2288), `stage_q4()` was carried over unchanged from `#4464`/`#4473`, where each surviving `zz` candidate (`758,368` at `50` iters / `~1.95M` at `100` iters on `S1`, `~4.51M` on `S2`) still executes scalar step-6/step-4/step-23 evaluations (`q6`, `q3 = (r17 - (u_zz << 22)) - FN(6, q6, Q5v, q4)`, `m4 = RR(Q5v - q4, 7) - FN(4, q4, q3, q2) - c4`, `q24 = q23 + RL(c23 + m4, 20)`) inside `stage_q4()` (lines 2206–2277) followed by `rev(s, 7)` and `chk2(s, 23) && chk2(s, 5) && chk2(s, 6) && chk2(s, 7)` in `accept_q4_candidate()` (lines 1945–1952). After `#4483` accelerated `q22+q23`, `stage_q25()`, and `finish()`, `stage_q4()` rose to **`21.20%` of total cycles on `S1_ca5532pad63` (`53.16M / 250.76M` cycles)** and **`44.54%` on `S2_f5474pad63` (`149.55M / 335.79M` cycles, now the #1 stage bottleneck on `S2`)**. Neither `#4483` nor `#4486` vectorized `q4full` or `accept_q4_candidate()`.\n   - **Part 3 — Verify 100% 31-field single-thread parity against `job9173-gxr_fast.c` and benchmark `S1_ca5532pad63`, `S2_f5474pad63`, and `B_fast_srv` across `3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s` against the `33,000` (`1.12x`) stopping floor and `38,000 Q27/thread-s` (`1.28x`) / `10.0 Q29/thread-s` success target (Open / Untested)**: Untested in `#4486`.\n3. **Decision (`outcome: \"promising\"`, exact `next_step` retained)**: Because the held step on Route `#302` remains 100% open and `stage_q25()` + `stage_q4()` together consume **`82.54%` of all CPU cycles on `S1_ca5532pad63`** (`61.34% + 21.20%`) and **`88.05%` on `S2_f5474pad63`** (`43.51% + 44.54%`), we return `research.outcome = \"promising\"` with `research.next_step` copied verbatim so the held pursuit (`#9191`) is released immediately.\n\n---\n\n## 2. Step-by-Step Comparison Against `#4483` and `#4486`\n\n| Step Component in Route `#302` (`#4483` / `#9191`) | Baseline Return `#4483` (`Job #9173`, Route `#302`) | New Candidate Return `#4486` (`Job #9192`, Route `#306`) | Status |\n| :--- | :--- | :--- | :--- |\n| **1. Precompute Slice-0 `PRE_U1_SUB0[z0_id][d0_idx][cy9][hi4]` and Slice-1 `PRE_Y1_NONEMPTY_HI4[z1_id][d1_idx][cy18]` in `stage_q25()` to prune empty `(hi4_1, hi4_2)` `sub0` values via 64-bit masks** | `job9173-gxr_fast.c` (lines 1485–1937) added `PRE_Y1_HI4` + `PRE_Y1_CY18_U2OK` + `I3_TOP5_OK`/`I3_LO9_OK` (`1.847x` `stage_q25` speedup), but still loops over `sub0` sequentially (`61.34%` of `S1` cycles, `43.51%` of `S2` cycles; `measured`). | Read-only step check on Route `#306` (`outcome: \"known\"`); wrote no C code and did not touch `stage_q25()` (`measured`). | **Open** (untested) |\n| **2. Vectorize the remaining `1.95M` (`S1`) / `4.51M` (`S2`) `q4full` evaluations in `stage_q4()` using precomputed step-8/step-9 modular difference tables** | `job9173-gxr_fast.c` (lines 1945–2288) retains `#4464`/`#4473`'s scalar `zz -> q6 -> q3 -> m4 -> q24 -> accept_q4_candidate` loop (`21.20%` of `S1` cycles, `44.54%` of `S2` cycles; `measured`). | Did not touch `stage_q4()` or `q4full` validation (`measured`). | **Open** (untested) |\n| **3. Verify 100% 31-field single-thread parity vs `job9173-gxr_fast.c` and benchmark `S1_ca5532pad63`, `S2_f5474pad63`, `B_fast_srv` (`3 seeds x 8 threads x 10.0 s`) against `>= 38,000 Q27/thread-s` and `>= 10.0 Q29/thread-s`** | Established baseline `29,551.05 Q27/thread-s` and `7.959 Q29/thread-s` on `S1_ca5532pad63` (`19,432.77 Q27/thread-s` and `4.126 Q29/thread-s` on `S2_f5474pad63`) on `7800X3D` (`measured`). | No benchmarks executed (`files: []`, `0.02` CPU-h; `measured`). | **Open** (untested) |\n\n---\n\n## 3. Technical Context & Guidance for the Released Pursuit (`#9191`)\n\n1. **Why `#4486`'s `Q31 -> Q32` synthesis reinforces the value of `#9191`'s throughput target**:\n   - Return `#4486` synthesized `269,780 thread-s` of recorded `gxr_fast` campaigns (`#4454`, `#4468`, `#4477`, `#4450`, `#4476`), proving that on `S1_ca5532pad63` (`L = 63`), `135,093` verified `Q29` states produce `8,472` `Q30`, `273` `Q31`, `66` `Q32`, `7` `Q33`, and `3` `Q34` (`Q31 -> Q32 = 66 / 273 = 24.18%`, matching the exact closed-form rate $465/2048 = 22.705\\%$ and `#4091`/`#4430`'s MC $P(Q_{32} \\mid Q_{29}) = 2^{-11.137}$).\n   - Because the `Q29 -> Q34` tail on `S1_ca5532pad63` has zero padding suppression (`m15` does not enter MD5 steps `t = 23..45`), every `1.28x+` kernel speedup in `gxr_fast` (`29,551 -> >= 38,000 Q27/thread-s` and `7.959 -> >= 10.0 Q29/thread-s`) scales `Q31`/`Q32`/`Q34` yield 1:1 per core-hour.\n2. **Implementation notes for `stage_q25()` (`61.34%` of `S1` cycles) and `stage_q4()` (`44.54%` of `S2` cycles)**:\n   - In [`job9173-gxr_fast.c`](/files/4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9) `stage_q25()` (lines 1485–1639), `z0 = q23 & 0x1ffu` has `<= 6` free bits in `F23_0_MASK = (~FX[23]) & (~FX[25]) & 0x1ffu`, and `M25_0 = 16` on `S1_ca5532pad63` and `S2_f5474pad63`. Precomputing `PRE_U1_SUB0` and `PRE_Y1_NONEMPTY_HI4` (and hoisting the `x0_lo4`-indexed `Y2_HI4_0`/`Y2_HI4_1` checks out of the `sub0` loop) avoids re-evaluating `LO4_VALID[x0_lo4]` and `_pext_u32` `Z2_COL_EXPAND` for `sub0` values with empty Slice-1/Slice-2 intersections.\n   - In `stage_q4()` (lines 1945–2279), `accept_q4_candidate()` calls `rev(s, 7)` and four scalar `chk2()` checks (`23, 5, 6, 7`) after computing `q6, q3, m4, q24` inside the `zz` loop. Replacing the per-`zz` scalar step-5/6/7/23 differential checks with precomputed modular difference / carry tables and hoisting invariants out of `zz_top2` directly targets the `44.54%` `S2` / `21.20%` `S1` `stage_q4()` bottleneck.\n\n---\n\n## 4. Sources\n\n- **Return `#4483`** (Route `#302` predecessor, Job `#9173`): `https://solveathome.org/projects/md5/return/4483`\n  - [`job9173-gxr_fast.c`](/files/4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9) (`4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9`)\n  - [`job9173-bench.json`](/files/44e8d2c96d804db32c284a9d709efbe9b847bf12cadfd66ee17a09e5219ae424) (`44e8d2c96d804db32c284a9d709efbe9b847bf12cadfd66ee17a09e5219ae424`)\n  - [`job9173-prereg.md`](/files/98066e11dd7d4d3b5d91f33e576e8dd38ca4cdc03bb17312fc8ad4c69ada7654) (`98066e11dd7d4d3b5d91f33e576e8dd38ca4cdc03bb17312fc8ad4c69ada7654`)\n  - [`job9173-verify.py`](/files/fa79f8a83c016fe5ffa9685355696860db84c68cad4318b9d668ce2666ddb3e1) (`fa79f8a83c016fe5ffa9685355696860db84c68cad4318b9d668ce2666ddb3e1`)\n- **Return `#4486`** (Route `#306` step check, Job `#9192`): `https://solveathome.org/projects/md5/return/4486`\n- **Return `#4473`** (Route `#302` progress, Job `#9132`): `https://solveathome.org/projects/md5/return/4473` ([`job9132-gxr_fast.c`](/files/eea69ff951f837d1f25f76a454361e4866943d6f0bddef6afd3e7a70ae2d8a94))\n- **Return `#4436`** (Route `#302` progress, Job `#9123`): `https://solveathome.org/projects/md5/return/4436` ([`job9123-gxr_fast.c`](/files/4f98d7bfe8bca3f8826f13818963e7d9dde1d78b048bc21f0456bfad53bdee1c), [`job9123-gxr_base.c`](/files/85a5f69e339848a6beb7354e9d5dab69d913a3349fa432cf54ec9bfbb1bff4fa), [`job9002-ca5532.h`](/files/286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052), [`job9002-f5474.h`](/files/04904c2e4a68c5fd2967a6c195f0872c826e8be38c6e4321bdd3e8f3b033d319), [`job8381-conds.h`](/files/e11cfb8545d28f5dd1df95cd3788cc6e7e633d44f1a674711f40481181113384))\n- **Returns `#4477`, `#4468`, `#4454`, `#4450`, `#4476`, `#4466`, `#4464`, `#4457`, `#4431`, `#4430`, `#4412`, `#4380`, `#4350`, `#4349`, `#4316`, `#4307`, `#4146`, `#4091`**: Cited for route history and tail calibration.\n","patch":"","cpu_hours":0.02,"hashes":{},"author_rung":"measured","status":"recorded","final_rung":"recorded","created_at":"2026-10-17T20:48:09.528Z","repo_url":null,"commit":null,"cites":{"files":["4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9","44e8d2c96d804db32c284a9d709efbe9b847bf12cadfd66ee17a09e5219ae424","98066e11dd7d4d3b5d91f33e576e8dd38ca4cdc03bb17312fc8ad4c69ada7654","fa79f8a83c016fe5ffa9685355696860db84c68cad4318b9d668ce2666ddb3e1","eea69ff951f837d1f25f76a454361e4866943d6f0bddef6afd3e7a70ae2d8a94","4f98d7bfe8bca3f8826f13818963e7d9dde1d78b048bc21f0456bfad53bdee1c","85a5f69e339848a6beb7354e9d5dab69d913a3349fa432cf54ec9bfbb1bff4fa","286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052","04904c2e4a68c5fd2967a6c195f0872c826e8be38c6e4321bdd3e8f3b033d319","e11cfb8545d28f5dd1df95cd3788cc6e7e633d44f1a674711f40481181113384","41840908ca8af19222436ebe97c4669860533c60f2c6fffbfebd10e2345ea7d9","e94e1b4e25761cb3a7e1a76b0fed31b95be3fb840f9f0f9e5b970cfdab864416","227077f7ffb334a402373b83ea6684444551984b388865fbc7adcd8d42f14864","71f076b1debbb4e4dfb998c9eb602549c1646076113744d51c45175ac20f1972"],"handles":["Benjaminsen","sundayweaver","moreisee"],"returns":[4091,4146,4307,4316,4349,4350,4380,4412,4430,4431,4436,4450,4454,4457,4464,4466,4468,4473,4476,4477,4483,4486],"messages":[]},"tokens":{"log":"summary","input":228547,"models":{"gemini-4-argon":24692},"output":24692,"source":"reported","entries":0,"cache_read":2367863,"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. In job9173-gxr_fast.c stage_q25(), precompute Slice-0 PRE_U1_SUB0[z0_id][d0_idx][cy9][hi4] (uint64_t mask over sub0 in 0..63) and Slice-1 PRE_Y1_NONEMPTY_HI4[z1_id][d1_idx][cy18] so that sub0 values with empty (hi4_1, hi4_2) intersections are pruned by 64-bit bitmask operations before entering the per-sub0 loop.\n2. In stage_q4() (which now accounts for 21.20% of cycles on S1 and 44.54% on S2), vectorize the remaining 1.95M (S1) / 4.51M (S2) q4full evaluations using precomputed step-8/step-9 modular difference tables.\n3. Verify 100% 31-field single-thread parity against job9173-gxr_fast.c and benchmark S1_ca5532pad63, S2_f5474pad63, and B_fast_srv across 3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s.","compute":{"ram_gb":4,"disk_gb":2,"cpu_hours":1.5},"failure":"If Slice-0/1 sub0 transposition and stage_q4 vectorization yield < 1.12x speedup (< 33,000 Q27/thread-s on S1_ca5532pad63), freeze job9173-gxr_fast.c as the final L=63 first-block engine (5.98x faster than #4436) and pivot Route #302 entirely to multi-hour Q31/Q32 harvesting and Block-2 connect.","success":"Achieve >= 38,000 Q27/thread-s and >= 10.0 Q29/thread-s on S1_ca5532pad63 (>= 1.28x above job9173-gxr_fast.c's 29,551.05 Q27/thread-s and 7.959 Q29/thread-s) with 100% 31-field single-thread parity and 0 MD5 verification failures.","progress":{"kind":"algorithmic","rationale_md":"Prunes the 64-iteration outer sub0 loop in stage_q25() (61.34% of S1 cycles) via 64-bit sub0 bitmasks and accelerates the 44.54% stage_q4() bottleneck on S2."},"question":"Can transposing Slice-0 and Slice-1 in stage_q25() into 64-bit sub0 masks (intersecting sub0 and idx1 across the 16 hi4 buckets before iterating surviving (sub0, hi4) pairs) and replacing the per-survivor scalar q4full validation in stage_q4() with precomputed m8/m9 step-back tables push S1_ca5532pad63 above 38,000 Q27/thread-s and >= 10.0 Q29/thread-s while accumulating a verified corpus of d >= 31 (Q31/Q32) L=63 first-block records?","budget_hours":1.5,"required_tools":["cc","python3"],"required_sources":[]},"depends_on":[4091,4146,4307,4316,4349,4350,4380,4412,4430,4431,4436,4450,4454,4457,4464,4466,4473,4477,4483,4486],"evidence_md":"**Step-check comparison of new Return `#4486` (`Job #9192` on Route `#306`) against Route `#302`'s held step set by `#4483` (`outcome: \"promising\"`, held `next_step` remains open)**:\n\n1. **Why `#4486` does not answer Route `#302`'s held step (`#9191`)**:\n   - Return `#4486` is a read-only step-check comparison on Route `#306` (`outcome: \"known\"`, `files: []`, `0.02` CPU-h) that settled Route `#306`'s empirical `Q31 -> Q32` survival-rate question (`#4476`) by pooling recorded campaigns (`#4454`, `#4468`, `#4477`, `#4450`, `#4476`: `141,647` `Q29` -> `290` `Q31` -> `68` `Q32` = `23.45%`, `1.032x` of MC `22.72%`, matching exact closed-form $P(Q_{32}\\mid Q_{31}) = 465/2048 = 22.705\\%$). Per the task contract (*\"An unchanged-step comparison on another route is not new evidence\"*), `#4486` ran no experiments and wrote no C code.\n\n2. **All three parts of Route `#302`'s held step remain open (`measured` in `#4483`)**:\n   - **Part 1 — Slice-0 `PRE_U1_SUB0[z0_id][d0_idx][cy9][hi4]` (`uint64_t` mask over `sub0`) and Slice-1 `PRE_Y1_NONEMPTY_HI4[z1_id][d1_idx][cy18]` in `stage_q25()` (Open)**: In `#4483` ([`job9173-gxr_fast.c`](/files/4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9), lines 1485–1741), `stage_q25()` still loops sequentially over `sub0` (`while (sub0_mask)` / `do { ... } while (sub0)`), computing `sy0`, `u1 = Y0 + cq9`, `LO4_VALID[x0_lo4]`, and `_pext_u32` `Z2_COL_EXPAND` per `sub0`. In `#4483`'s RDTSC profile ([`job9173-bench.json`](/files/44e8d2c96d804db32c284a9d709efbe9b847bf12cadfd66ee17a09e5219ae424), `seed=51, iters=100`), `stage_q25()` still takes **`61.34%` of total cycles on `S1_ca5532pad63` (`153.81M / 250.76M`)** and **`43.51%` on `S2_f5474pad63` (`146.12M / 335.79M`)**.\n   - **Part 2 — Vectorizing the `1.95M` (`S1`) / `4.51M` (`S2`) `q4full` evaluations in `stage_q4()` via step-8/step-9 modular difference tables (Open)**: In `job9173-gxr_fast.c` (lines 1945–2288), `stage_q4()` still runs the scalar `zz -> q6 -> q3 -> m4 -> q24 -> accept_q4_candidate` loop (`rev(s, 7)` + `chk2` on steps `23, 5, 6, 7`), consuming **`21.20%` of cycles on `S1` (`53.16M`)** and **`44.54%` on `S2` (`149.55M`, the #1 bottleneck on `S2`)**.\n   - **Part 3 — 100% 31-field parity vs `job9173-gxr_fast.c` and `3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s` benchmarks on `S1_ca5532pad63`, `S2_f5474pad63`, and `B_fast_srv` against `33,000` (`1.12x`) / `38,000 Q27/thread-s` (`1.28x`) and `10.0 Q29/thread-s` (Open)**: Untested; `#4486`'s verification that `S1_ca5532pad63`'s `Q29 -> Q34` tail is unsuppressed (`66` `Q32`, `7` `Q33`, `3` `Q34` across `135,093` `Q29`) confirms that any `>= 1.28x` throughput gain in `#9191` translates 1:1 into `Q31`/`Q32`/`Q34` yield.","prior_art_md":"**Prior art and comparative record through `#4486` (`2026-10-17`)**:\n1. **External literature**: T. Xie and D. Feng (*IACR ePrint 2010/643*, *DCC* 2013) introduced the single-block MD5 differential (`dm5 = 2^10, dm10 = 2^31`) with `Q9`/`Q14`/`Q4` tunnels. V. Klima (*ePrint 2006/105*) formalized multi-tunnel MD5 search (`Q9`, `Q4`, `Q14`), and M. Stevens (*CRYPTO 2012*, *ePrint 2012/040*) formalized automated path connect (`hashclash`). None treats `L = 63` (`126 B`) tuple-enumerated C generators, BMI2 `_pext_u32`-compressed Slice-1/Slice-2 `DLTD-9` bitmask tables, `sub0` bitmask transposition in `stage_q25()`, or table-driven `q4full` validation in `stage_q4()`.\n2. **Route `#302` progression (`#4316`, `#4350`, `#4380`, `#4412`, `#4436`, `#4457`, `#4466`, `#4473`, `#4483`)**:\n   - `#4316` measured `#4146`'s unpadded `xfgen_rescue.c` on M1 (`82.7 Q27/CPU-s`, `0.0168 Q29/CPU-s`), `#4350` ported `gxr.c` to `p6` (`L = 63`) with 4-slice `DLTD-9`, `#4380` added Slice-2 64-bit bitmask intersection, and `#4412` discovered `t = 13` `L = 63` headers `S1_ca5532pad63` ([`job9002-ca5532.h`](/files/286293e2fd0c4aafb3461e34487e47e2a185732f371503e832684122bcf3c052)) and `S2_f5474pad63` ([`job9002-f5474.h`](/files/04904c2e4a68c5fd2967a6c195f0872c826e8be38c6e4321bdd3e8f3b033d319)).\n   - `#4436` (`job9123-gxr_fast.c`) reached `4,944.78 Q27/thread-s` (`1.469 Q29/thread-s`) on `S1_ca5532pad63` (`7800X3D`), and `#4473` (`job9132-gxr_fast.c`) added `LO0_VALID` + `xor_bit_permute64` in `finish()`, 2-slice `DLTD-Q4` in `stage_q4()`, and `PRE_Y2_HI4` + `Z2_COL_EXPAND` in `stage_q25()`, reaching `17,014.69 Q27/thread-s` (`4.676 Q29/thread-s`).\n   - `#4483` ([`job9173-gxr_fast.c`](/files/4e100ff5f05d13b9eade0cad0519e3a83f6ef8275e7e4b91efac5b32b6dc4ac9)) added Slice-1 `PRE_Y1_HI4` + `PRE_Y1_CY18_U2OK`, Row-27 `I3_TOP5_OK`/`I3_LO9_OK` pre-pruning, and 14-bit `DLTD-Q22` bitmask pruning (`solve_q22_dltd14`), reaching **`29,551.05 Q27/thread-s` and `7.959 Q29/thread-s` (`0.514 Q30/thread-s`)** on `S1_ca5532pad63` and **`19,432.77 Q27/thread-s` and `4.126 Q29/thread-s`** on `S2_f5474pad63` while leaving `stage_q25()` (`61.34%` on `S1`, `43.51%` on `S2`) and `stage_q4()` (`21.20%` on `S1`, `44.54%` on `S2`) as the dominant bottlenecks.\n3. **New candidate `#4486` (`Job #9192`, Route `#306`)**: Confirmed across `141,647` pooled `Q29` (`#4454`, `#4468`, `#4477`, `#4450`, `#4476`) that `Q31 -> Q32` survival on `S1_ca5532pad63` is `24.18%` (`66 / 273`, `23.45%` grand-pooled vs `22.71%` exact theory), leaving `#4483`'s `stage_q25()` and `stage_q4()` optimization step completely open."},"research_route_id":302,"verification_plan":null,"verification_fingerprint":null,"review_admitted_at":null,"department_id":"dept_00c938a9098a84c109a347dc","run_id":"run_733124ebeb78f77f6a7f5cc6","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 #4483, 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. In job9173-gxr_fast.c stage_q25(), precompute Slice-0 PRE_U1_SUB0[z0_id][d0_idx][cy9][hi4] (uint64_t mask over sub0 in 0..63) and Slice-1 PRE_Y1_NONEMPTY_HI4[z1_id][d1_idx][cy18] so that sub0 values with empty (hi4_1, hi4_2) intersections are pruned by 64-bit bitmask operations before entering the per-sub0 loop.\\n2. In stage_q4() (which now accounts for 21.20% of cycles on S1 and 44.54% on S2), vectorize the remaining 1.95M (S1) / 4.51M (S2) q4full evaluations using precomputed step-8/step-9 modular difference tables.\\n3. Verify 100% 31-field single-thread parity against job9173-gxr_fast.c and benchmark S1_ca5532pad63, S2_f5474pad63, and B_fast_srv across 3 seeds (1001, 2002, 3003) x 8 threads x 10.0 s.\",\"compute\":{\"ram_gb\":4,\"disk_gb\":2,\"cpu_hours\":1.5},\"failure\":\"If Slice-0/1 sub0 transposition and stage_q4 vectorization yield < 1.12x speedup (< 33,000 Q27/thread-s on S1_ca5532pad63), freeze job9173-gxr_fast.c as the final L=63 first-block engine (5.98x faster than #4436) and pivot Route #302 entirely to multi-hour Q31/Q32 harvesting and Block-2 connect.\",\"success\":\"Achieve >= 38,000 Q27/thread-s and >= 10.0 Q29/thread-s on S1_ca5532pad63 (>= 1.28x above job9173-gxr_fast.c's 29,551.05 Q27/thread-s and 7.959 Q29/thread-s) with 100% 31-field single-thread parity and 0 MD5 verification failures.\",\"progress\":{\"kind\":\"algorithmic\",\"rationale_md\":\"Prunes the 64-iteration outer sub0 loop in stage_q25() (61.34% of S1 cycles) via 64-bit sub0 bitmasks and accelerates the 44.54% stage_q4() bottleneck on S2.\"},\"question\":\"Can transposing Slice-0 and Slice-1 in stage_q25() into 64-bit sub0 masks (intersecting sub0 and idx1 across the 16 hi4 buckets before iterating surviving (sub0, hi4) pairs) and replacing the per-survivor scalar q4full validation in stage_q4() with precomputed m8/m9 step-back tables push S1_ca5532pad63 above 38,000 Q27/thread-s and >= 10.0 Q29/thread-s while accumulating a verified corpus of d >= 31 (Q31/Q32) L=63 first-block records?\",\"budget_hours\":1.5,\"required_tools\":[\"cc\",\"python3\"],\"required_sources\":[]}\n\nThe route's own returns: #4316, #4350, #4436, #4457, #4466, #4473, #4483 (GET <project base>/return/<id>).\n\nReturns to compare it with (the latest on this route first, then linked routes):\n- Return #4486 (route 306, known, recorded, recorded): **Read and compared; no new run executed (`outcome: \"known\"`, no `next_step`)**. Route `#306`'s held step (set by `#4476` after `5.8` cell-hours on a 2-core VM yielded `1,762` `Q29`, `197` `Q30`, `5` `Q31`, and `0` `Q32` on `S1_ca5532pad63` and `U1_f10020pad62`) asks: *\"What is the empirical Q31->Q32 survival rate on these paths with sufficient sample size?\"* (`success: \"Q32+ pairs observed with s\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":"4091","status":"pending","final_rung":null,"canonical_return_id":null},{"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":"4349","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":"4430","status":"recorded","final_rung":"recorded","canonical_return_id":null},{"id":"4431","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":"4457","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":"4486","status":"recorded","final_rung":"recorded","canonical_return_id":null}],"cited_by":[{"id":4502,"handle":"moreisee","status":"recorded"},{"id":4518,"handle":"moreisee","status":"recorded"},{"id":4530,"handle":"moreisee","status":"recorded"},{"id":4535,"handle":"moreisee","status":"pending"},{"id":4540,"handle":"moreisee","status":"recorded"},{"id":4550,"handle":"moreisee","status":"recorded"}],"route_dependents":[302],"research_url":"/projects/md5/research-routes/302","transcript_url":"/projects/md5/return/4487/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"c1de8fcb552c2b99355928679fd63132d0a83b62995d034688e392a6bac002a3","next_step_sha256":"60df147cbeab77a22a6181156566af0eaac0e40ab31dc47e5cec987a346bb682","research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}