Remove normalization buffers that amplify multicore image preparation cost

benchmark/results/normalization_scaling_receiving_20261008/README.md

@@ -0,0 +1,16 @@
+# Matched WoundHealing normalization receiving
+
+Source: `6cbf26d62db0385e88a2138eaac6809d68dc1898`, clean and frozen through receiving. ArrayBridge: merged `31632d8ce31016f0fe6ef28df61f49da397bf3e6`, version 0.3.8. The existing materialization worktree was reused; native extension source matches the sealed qualified baseline.
+
+Actual fixed twelve-assignment workload, one and four OpenHCS workers on the same owned endpoint, one warm-up plus three measured repetitions per mode. All 96 assignment comparisons passed existing CellProfiler tolerances, with zero database, CSV, image, or declared-output differences. Native CellProfiler was not re-executed: comparison used genuine retained eight-assignment observations at their original cardinality. No native timing projection or native speedup claim is made here.
+
+| Workers | Full output-complete execution median | Axis execution span median | Compilation median | Complete client operation median |
+| --- | ---: | ---: | ---: | ---: |
+| 1 | 7.908462 s | 7.839103 s | 0.179042 s | 8.149202 s |
+| 4 | 3.199810 s | 3.060829 s | 0.183937 s | 3.443651 s |
+
+Actual full execution scaling is **2.471542x** using the newly measured one-worker denominator. Full execution includes worker setup, processing, saving, publication, exports and finalization; external server startup and scientific qualification are excluded. Compilation is reported separately. Four-worker full execution decreased from the published prior cohort's 5.207406 s; the new one-worker measurement is 7.908462 s versus the prior 7.754487 s. The minimum 3x thirty-workflow goal remains unfinished.
+
+The recipe records endpoint incarnation, all server thread affinities, workload and retained-reference arguments. Parent control uses CPUs 2–5; all server threads use CPU 5 for the one-worker mode and CPUs 2–5 for four workers. Numerical thread limits are one. Qualification uses CPUs 0–1 outside the measured processing clock.
+
+Diagnostics are explicitly separate from production timings. Source reconstruction proves original loaded pixels are reused; source guards were not loosened. A four-fork saved-source replay priced normalization allocation changes, with exact numerical, metadata and producer-immutability controls. The archived P0 phases were collected before the patch and are diagnostic only. Raw reports, all eight receipts and progress streams, and a SHA-256 inventory support the production result above. This bounded receiving does not replace the published full thirty-workflow manuscript figures.

benchmark/results/normalization_scaling_receiving_20261008/diagnostics/concurrent-normalization-proof.json

@@ -0,0 +1,316 @@
+{
+  "scope": "Saved exact original JPEG uint8 pixels and reconstructed STACK/source scales; preparation proved source reuse; numeric-only four-fork replay, not production pipeline",
+  "exact_array_metadata_and_source_immutability": true,
+  "lanes": [
+    [
+      {
+        "round": 0,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.0899204779416323,
+        "user_seconds": 0.098233,
+        "sys_seconds": 0.981675,
+        "minor_faults": 149231,
+        "major_faults": 0
+      },
+      {
+        "round": 0,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.328625600785017,
+        "user_seconds": 0.06587100000000001,
+        "sys_seconds": 0.25915,
+        "minor_faults": 39739,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3164014583453536,
+        "user_seconds": 0.056548000000000015,
+        "sys_seconds": 0.2569279999999998,
+        "minor_faults": 45868,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.08542849868536,
+        "user_seconds": 0.08447999999999997,
+        "sys_seconds": 0.990583,
+        "minor_faults": 150369,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.1104760123416781,
+        "user_seconds": 0.08516299999999999,
+        "sys_seconds": 1.0153660000000002,
+        "minor_faults": 155985,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3982424968853593,
+        "user_seconds": 0.06040799999999996,
+        "sys_seconds": 0.332249,
+        "minor_faults": 38705,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.39833038207143545,
+        "user_seconds": 0.09090300000000007,
+        "sys_seconds": 0.30278300000000025,
+        "minor_faults": 39728,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.1877849753946066,
+        "user_seconds": 0.10045400000000004,
+        "sys_seconds": 1.075697,
+        "minor_faults": 145763,
+        "major_faults": 0
+      }
+    ],
+    [
+      {
+        "round": 0,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.0945017263293266,
+        "user_seconds": 0.07950299999999999,
+        "sys_seconds": 1.005688,
+        "minor_faults": 152756,
+        "major_faults": 0
+      },
+      {
+        "round": 0,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3466633642092347,
+        "user_seconds": 0.051539,
+        "sys_seconds": 0.292489,
+        "minor_faults": 41263,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3103548102080822,
+        "user_seconds": 0.04754899999999998,
+        "sys_seconds": 0.25983599999999996,
+        "minor_faults": 44848,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.0335356537252665,
+        "user_seconds": 0.07837499999999997,
+        "sys_seconds": 0.946164,
+        "minor_faults": 146796,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.244527623988688,
+        "user_seconds": 0.09650000000000003,
+        "sys_seconds": 1.1376689999999998,
+        "minor_faults": 149341,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.42102981451898813,
+        "user_seconds": 0.06712099999999999,
+        "sys_seconds": 0.34824299999999964,
+        "minor_faults": 39729,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.4077610643580556,
+        "user_seconds": 0.07613199999999998,
+        "sys_seconds": 0.32737799999999995,
+        "minor_faults": 36661,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.19007140211761,
+        "user_seconds": 0.08646799999999993,
+        "sys_seconds": 1.0927549999999995,
+        "minor_faults": 139120,
+        "major_faults": 0
+      }
+    ],
+    [
+      {
+        "round": 0,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.0589780705049634,
+        "user_seconds": 0.082421,
+        "sys_seconds": 0.966828,
+        "minor_faults": 150712,
+        "major_faults": 0
+      },
+      {
+        "round": 0,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3321199119091034,
+        "user_seconds": 0.070289,
+        "sys_seconds": 0.25756,
+        "minor_faults": 47915,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.26224829722195864,
+        "user_seconds": 0.06523000000000001,
+        "sys_seconds": 0.194129,
+        "minor_faults": 46890,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.1582563808187842,
+        "user_seconds": 0.06754600000000002,
+        "sys_seconds": 1.0804869999999998,
+        "minor_faults": 153436,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.253949728794396,
+        "user_seconds": 0.09754999999999997,
+        "sys_seconds": 1.1453460000000004,
+        "minor_faults": 145762,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.42218572180718184,
+        "user_seconds": 0.07216800000000001,
+        "sys_seconds": 0.3455499999999998,
+        "minor_faults": 36150,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3298925580456853,
+        "user_seconds": 0.060631999999999964,
+        "sys_seconds": 0.26204499999999964,
+        "minor_faults": 34112,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.1089950324967504,
+        "user_seconds": 0.08534799999999998,
+        "sys_seconds": 1.009709,
+        "minor_faults": 150369,
+        "major_faults": 0
+      }
+    ],
+    [
+      {
+        "round": 0,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.0951432837173343,
+        "user_seconds": 0.073476,
+        "sys_seconds": 1.0123090000000001,
+        "minor_faults": 143060,
+        "major_faults": 0
+      },
+      {
+        "round": 0,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.34646662324666977,
+        "user_seconds": 0.06039299999999999,
+        "sys_seconds": 0.2832649999999999,
+        "minor_faults": 45863,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3215368306264281,
+        "user_seconds": 0.05558199999999999,
+        "sys_seconds": 0.26227699999999987,
+        "minor_faults": 30529,
+        "major_faults": 0
+      },
+      {
+        "round": 1,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.1571019664406776,
+        "user_seconds": 0.085475,
+        "sys_seconds": 1.0614290000000004,
+        "minor_faults": 164671,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.1902522621676326,
+        "user_seconds": 0.085511,
+        "sys_seconds": 1.0954099999999998,
+        "minor_faults": 147294,
+        "major_faults": 0
+      },
+      {
+        "round": 2,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.40609566774219275,
+        "user_seconds": 0.06027199999999999,
+        "sys_seconds": 0.34204500000000015,
+        "minor_faults": 39727,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "proposed",
+        "three_assignments_seconds": 0.3732900321483612,
+        "user_seconds": 0.067212,
+        "sys_seconds": 0.3018590000000003,
+        "minor_faults": 37173,
+        "major_faults": 0
+      },
+      {
+        "round": 3,
+        "algorithm": "original",
+        "three_assignments_seconds": 1.128940388560295,
+        "user_seconds": 0.10124300000000003,
+        "sys_seconds": 1.0163060000000002,
+        "minor_faults": 142699,
+        "major_faults": 0
+      }
+    ]
+  ],
+  "critical_three_assignment_seconds": {
+    "original": [
+      1.0951432837173343,
+      1.1582563808187842,
+      1.253949728794396,
+      1.19007140211761
+    ],
+    "proposed": [
+      0.3466633642092347,
+      0.3215368306264281,
+      0.42218572180718184,
+      0.4077610643580556
+    ]
+  }
+}

benchmark/results/normalization_scaling_receiving_20261008/diagnostics/cupy-normalization-controls.json

@@ -0,0 +1,5 @@
+{
+  "exact_cupy_source_immutability_dtype_device_controls": 3,
+  "status": "PASS",
+  "timing_claim": null
+}

benchmark/results/normalization_scaling_receiving_20261008/diagnostics/final-qualification.json

@@ -0,0 +1,32 @@
+{
+  "scope": "Final source qualification; saved-source replay timing, not whole pipeline gain",
+  "file_sha256": {
+    "/home/ts/code/projects/openhcs-cohort-qualification-main429-20261002/openhcs/core/runtime_image_values.py": "6735e80510f5b46f7b32f287d26e4a02bb5fdc13191b8e8e75531c36337ac96a",
+    "/home/ts/code/projects/openhcs/external/arraybridge/src/arraybridge/array_operations.py": "02379a3aff684f17c3b1c5026a2ba0f3e3f6fcd2f5e4535de987c74204a08d38",
+    "/home/ts/code/projects/openhcs/external/arraybridge/src/arraybridge/types.py": "8a6f831b39ccedf65ce31aa3e8b324ef553d8dcca62d8ae20529583a6ff3cfea",
+    "/home/ts/code/projects/openhcs/external/arraybridge/tests/test_native_geometry.py": "dc4c65195b4da1e6930b4fbc6b37be049aac7e8df2a8ad9c25ead8b09899f4da"
+  },
+  "original_owner_controls": 252,
+  "genuine_wound_pixels_and_metadata_exact": true,
+  "cupy_controls": 3,
+  "external_release_checks": {
+    "ruff": "PASS",
+    "black": "PASS",
+    "mypy": "PASS",
+    "pytest_passed": 242,
+    "pytest_skipped": 17
+  },
+  "existing_openhcs_normalization_tests": 2,
+  "receipts": [
+    "phase-analysis.json",
+    "guard-proof.json",
+    "image-request-profile.txt",
+    "concurrent-normalization-proof.json",
+    "normalization-controls.json",
+    "cupy-normalization-controls.json"
+  ],
+  "scratch_critical_three_assignment_seconds": {
+    "original": 1.174163891468197,
+    "one_owned_buffer": 0.37721221428364515
+  }
+}

benchmark/results/normalization_scaling_receiving_20261008/diagnostics/guard-proof.json

@@ -0,0 +1,14 @@
+{
+  "input_shape": [
+    2,
+    1536,
+    2304,
+    3
+  ],
+  "input_dtype": "uint8",
+  "input_plane_axis": "RuntimePlaneAxis.RUNTIME_SLICE",
+  "edge_main_flow_projection": "MainFlowInputProjection.COMPLETE_PAYLOAD",
+  "invocation_position": 0,
+  "dependency": "StepInputDependencyKind.PIPELINE_START",
+  "loaded_source_reused": true
+}

benchmark/results/normalization_scaling_receiving_20261008/diagnostics/normalization-controls.json

@@ -0,0 +1,4 @@
+{
+  "exact_original_owner_controls": 252,
+  "dtype_shape_nan_inf_strides_source_immutability_and_metadata": "PASS"
+}