oidn-web 0.4.0 → 0.5.0

package/benchmarks/results/profile-fp16-unfused-pool.json

@@ -1,385 +0,0 @@
-{
-  "generatedAt": "2026-08-20T04:06:25.590Z",
-  "currentCommit": "5d9c1c8771e2a03c90bfadafec4b5dd28fada446",
-  "baselineCommit": "136c00c34aa5cf4561b4137034dbd643177f0a5f",
-  "settings": {
-    "width": 512,
-    "height": 512,
-    "tileSize": 512,
-    "warmup": 1,
-    "runs": 3,
-    "model": "rt_hdr_calb_cnrm_large.tza",
-    "gpuQueueCompletionIncluded": true
-  },
-  "browser": "/Applications/Google Chrome.app/Contents/MacOS/Google Chrome",
-  "adapter": {
-    "vendor": "apple",
-    "architecture": "metal-3",
-    "device": "",
-    "description": ""
-  },
-  "results": [
-    {
-      "label": "TFJS (136c00c)",
-      "bundle": "baseline",
-      "baseline": true,
-      "initializationMs": 8.300000004470348,
-      "timesMs": [
-        34.69999999552965,
-        34.600000001490116,
-        34.70000000298023
-      ],
-      "supportsFP16": true,
-      "adapter": {
-        "vendor": "apple",
-        "architecture": "metal-3",
-        "device": "",
-        "description": ""
-      },
-      "summary": {
-        "minMs": 34.600000001490116,
-        "medianMs": 34.69999999552965,
-        "meanMs": 34.666666666666664,
-        "p95Ms": 34.70000000298023,
-        "maxMs": 34.70000000298023
-      },
-      "messages": []
-    },
-    {
-      "label": "WGSL FP32",
-      "bundle": "current",
-      "baseline": false,
-      "precision": "fp32",
-      "initializationMs": 349.90000000596046,
-      "timesMs": [
-        41,
-        41.100000001490116,
-        41.20000000298023
-      ],
-      "executionProfile": {
-        "totalMs": 40.630814,
-        "layers": [
-          {
-            "id": "input-pack",
-            "durationMs": 0.050205
-          },
-          {
-            "id": "enc_conv1a",
-            "durationMs": 0.957234
-          },
-          {
-            "id": "enc_conv1b",
-            "durationMs": 4.513224
-          },
-          {
-            "id": "pool1",
-            "durationMs": 0.408767
-          },
-          {
-            "id": "enc_conv2a",
-            "durationMs": 1.669064
-          },
-          {
-            "id": "enc_conv2b",
-            "durationMs": 2.523179
-          },
-          {
-            "id": "pool2",
-            "durationMs": 0.054997
-          },
-          {
-            "id": "enc_conv3a",
-            "durationMs": 0.87678
-          },
-          {
-            "id": "enc_conv3b",
-            "durationMs": 1.251299
-          },
-          {
-            "id": "pool3",
-            "durationMs": 0.018124
-          },
-          {
-            "id": "enc_conv4a",
-            "durationMs": 0.480971
-          },
-          {
-            "id": "enc_conv4b",
-            "durationMs": 0.716665
-          },
-          {
-            "id": "pool4",
-            "durationMs": 0.008416
-          },
-          {
-            "id": "enc_conv5a",
-            "durationMs": 0.290942
-          },
-          {
-            "id": "enc_conv5b",
-            "durationMs": 0.386601
-          },
-          {
-            "id": "dec_conv4a",
-            "durationMs": 1.461411
-          },
-          {
-            "id": "dec_conv4b",
-            "durationMs": 0.714623
-          },
-          {
-            "id": "dec_conv3a",
-            "durationMs": 2.739875
-          },
-          {
-            "id": "dec_conv3b",
-            "durationMs": 1.162346
-          },
-          {
-            "id": "dec_conv2a",
-            "durationMs": 5.078732
-          },
-          {
-            "id": "dec_conv2b",
-            "durationMs": 2.493431
-          },
-          {
-            "id": "dec_conv1a",
-            "durationMs": 7.843271
-          },
-          {
-            "id": "dec_conv1b",
-            "durationMs": 4.45502
-          },
-          {
-            "id": "dec_conv1c",
-            "durationMs": 0.475637
-          }
-        ]
-      },
-      "runtimeInfo": {
-        "configuredEngine": "wgsl",
-        "gpuEngine": "wgsl",
-        "precision": "fp32",
-        "model": "oidn-unet-large-v1",
-        "modelFamily": "oidn-unet-large",
-        "inputChannels": 9,
-        "dynamicTile": {
-          "enabled": false,
-          "currentTileSize": 512,
-          "minTileSize": 256,
-          "maxTileSize": 512,
-          "targetTileTimeMs": 16
-        },
-        "lastExecution": {
-          "width": 512,
-          "height": 512,
-          "tileWidth": 512,
-          "tileHeight": 512,
-          "tileCount": 1,
-          "durationMs": 42.600000001490116,
-          "tileTimeMs": {
-            "min": 42.600000001490116,
-            "median": 42.600000001490116,
-            "mean": 42.600000001490116,
-            "max": 42.600000001490116
-          }
-        }
-      },
-      "supportsFP16": true,
-      "adapter": {
-        "vendor": "apple",
-        "architecture": "metal-3",
-        "device": "",
-        "description": ""
-      },
-      "summary": {
-        "minMs": 41,
-        "medianMs": 41.100000001490116,
-        "meanMs": 41.100000001490116,
-        "p95Ms": 41.20000000298023,
-        "maxMs": 41.20000000298023
-      },
-      "messages": [],
-      "validation": {
-        "sampleCount": 6144,
-        "meanAbsoluteError": 0.0000011096847932397698,
-        "rootMeanSquaredError": 0.0000018071588418628667,
-        "maxAbsoluteError": 0.000013589859008789062,
-        "passed": true,
-        "thresholds": {
-          "maxMeanError": 0.0001,
-          "maxAbsoluteError": 0.001
-        }
-      }
-    },
-    {
-      "label": "WGSL FP16",
-      "bundle": "current",
-      "baseline": false,
-      "precision": "fp16",
-      "initializationMs": 22.700000002980232,
-      "timesMs": [
-        30.400000005960464,
-        30.299999997019768,
-        30.5
-      ],
-      "executionProfile": {
-        "totalMs": 29.698353,
-        "layers": [
-          {
-            "id": "input-pack",
-            "durationMs": 0.043914
-          },
-          {
-            "id": "enc_conv1a",
-            "durationMs": 0.679791
-          },
-          {
-            "id": "enc_conv1b",
-            "durationMs": 3.891304
-          },
-          {
-            "id": "pool1",
-            "durationMs": 0.33973
-          },
-          {
-            "id": "enc_conv2a",
-            "durationMs": 1.531323
-          },
-          {
-            "id": "enc_conv2b",
-            "durationMs": 2.29611
-          },
-          {
-            "id": "pool2",
-            "durationMs": 0.049831
-          },
-          {
-            "id": "enc_conv3a",
-            "durationMs": 0.738288
-          },
-          {
-            "id": "enc_conv3b",
-            "durationMs": 0.938109
-          },
-          {
-            "id": "pool3",
-            "durationMs": 0.016207
-          },
-          {
-            "id": "enc_conv4a",
-            "durationMs": 0.337562
-          },
-          {
-            "id": "enc_conv4b",
-            "durationMs": 0.531051
-          },
-          {
-            "id": "pool4",
-            "durationMs": 0.010999
-          },
-          {
-            "id": "enc_conv5a",
-            "durationMs": 0.189031
-          },
-          {
-            "id": "enc_conv5b",
-            "durationMs": 0.246777
-          },
-          {
-            "id": "dec_conv4a",
-            "durationMs": 0.948942
-          },
-          {
-            "id": "dec_conv4b",
-            "durationMs": 0.522844
-          },
-          {
-            "id": "dec_conv3a",
-            "durationMs": 1.705478
-          },
-          {
-            "id": "dec_conv3b",
-            "durationMs": 0.932276
-          },
-          {
-            "id": "dec_conv2a",
-            "durationMs": 2.966486
-          },
-          {
-            "id": "dec_conv2b",
-            "durationMs": 2.300068
-          },
-          {
-            "id": "dec_conv1a",
-            "durationMs": 4.382315
-          },
-          {
-            "id": "dec_conv1b",
-            "durationMs": 3.866681
-          },
-          {
-            "id": "dec_conv1c",
-            "durationMs": 0.233236
-          }
-        ]
-      },
-      "runtimeInfo": {
-        "configuredEngine": "wgsl",
-        "gpuEngine": "wgsl",
-        "precision": "fp16",
-        "model": "oidn-unet-large-v1",
-        "modelFamily": "oidn-unet-large",
-        "inputChannels": 9,
-        "dynamicTile": {
-          "enabled": false,
-          "currentTileSize": 512,
-          "minTileSize": 256,
-          "maxTileSize": 512,
-          "targetTileTimeMs": 16
-        },
-        "lastExecution": {
-          "width": 512,
-          "height": 512,
-          "tileWidth": 512,
-          "tileHeight": 512,
-          "tileCount": 1,
-          "durationMs": 31.19999999552965,
-          "tileTimeMs": {
-            "min": 31.19999999552965,
-            "median": 31.19999999552965,
-            "mean": 31.19999999552965,
-            "max": 31.19999999552965
-          }
-        }
-      },
-      "supportsFP16": true,
-      "adapter": {
-        "vendor": "apple",
-        "architecture": "metal-3",
-        "device": "",
-        "description": ""
-      },
-      "summary": {
-        "minMs": 30.299999997019768,
-        "medianMs": 30.400000005960464,
-        "meanMs": 30.400000000993412,
-        "p95Ms": 30.5,
-        "maxMs": 30.5
-      },
-      "messages": [],
-      "validation": {
-        "sampleCount": 6144,
-        "meanAbsoluteError": 0.0010210334021394374,
-        "rootMeanSquaredError": 0.001609680379854656,
-        "maxAbsoluteError": 0.012395620346069336,
-        "passed": true,
-        "thresholds": {
-          "maxMeanError": 0.005,
-          "maxAbsoluteError": 0.05
-        }
-      }
-    }
-  ]
-}

package/benchmarks/results/profile-fp16-unfused-pool.md

@@ -1,47 +0,0 @@
-# oidn-web benchmark
-
-- Current: `5d9c1c8771e2a03c90bfadafec4b5dd28fada446`
-- TFJS baseline: `136c00c34aa5cf4561b4137034dbd643177f0a5f`
-- Input: 512x512, fixed tile 512, 3 runs after 1 warmup(s)
-- Adapter: apple
-
-| Runtime | Init ms | Median ms | P95 ms | Speedup | Precision |
-| --- | ---: | ---: | ---: | ---: | --- |
-| TFJS (136c00c) | 8.30 | 34.70 | 34.70 | 1.00x | fp32 |
-| WGSL FP32 | 349.90 | 41.10 | 41.20 | 0.84x | fp32 |
-| WGSL FP16 | 22.70 | 30.40 | 30.50 | 1.14x | fp16 |
-
-## Output validation
-
-Compared against sampled TFJS FP32 output.
-
-| Runtime | Status | MAE | RMSE | Max error | Samples |
-| --- | --- | ---: | ---: | ---: | ---: |
-| WGSL FP32 | pass | 1.11e-6 | 1.81e-6 | 1.36e-5 | 6144 |
-| WGSL FP16 | pass | 1.02e-3 | 1.61e-3 | 1.24e-2 | 6144 |
-
-## GPU hot layers
-
-### WGSL FP32
-
-Profiled GPU total: 40.63 ms
-
-| Node | GPU ms |
-| --- | ---: |
-| dec_conv1a | 7.84 |
-| dec_conv2a | 5.08 |
-| enc_conv1b | 4.51 |
-| dec_conv1b | 4.46 |
-| dec_conv3a | 2.74 |
-
-### WGSL FP16
-
-Profiled GPU total: 29.70 ms
-
-| Node | GPU ms |
-| --- | ---: |
-| dec_conv1a | 4.38 |
-| enc_conv1b | 3.89 |
-| dec_conv1b | 3.87 |
-| dec_conv2a | 2.97 |
-| dec_conv2b | 2.30 |