)]}'
{
  "commit": "a082d8e7cb9118e9df5965e63c4ff64e62e5fd14",
  "tree": "a08df2bffc23a81a2036517114a7262ff447b272",
  "parents": [
    "2e756bef75e52b2e880917e5e0ff1f60ad5fcd4d"
  ],
  "author": {
    "name": "Lokbondo Kung",
    "email": "lokokung@google.com",
    "time": "Thu Jan 29 17:07:24 2026 -0800"
  },
  "committer": {
    "name": "Dawn LUCI CQ",
    "email": "dawn-scoped@luci-project-accounts.iam.gserviceaccount.com",
    "time": "Thu Jan 29 17:07:24 2026 -0800"
  },
  "message": "[dawn] Add additional tier2 for maxSampledTexturesPerShaderStage.\n\n- Adds a new tier2 for some limits and increase\n  maxSampledTexturesPerShaderStage to 48 for the new tier. Note\n  that we need to add a new tier and cannot just increase tier1\n  because OpenGL basically never allows more than 16 sampled\n  textures, and upgrading tier1 to 48 would result in a lot of\n  OpenGL backends to regress to using tier0 and cause some other\n  limits to drop as well which may be a breaking change.\n- From investigating the different devices per platform, chose\n  48 for the next tier value mainly because of Vulkan and Metal.\n  Metal\u0027s limits at Apple Gen 4 and beyond (96) which is split\n  (unequally) between sampled and storage textures would\n  comfortably fit 48 because its currently split 54/42. For\n  Vulkan, using 48 would only additionally lose the devices\n  where it\u0027s limit is 31, which according to the database is\n  only around 100 devices. For D3D11 and D3D12, the value\n  doesn\u0027t matter unless we want to increase it past 128.\n  Finally, for OpenGL, increasing the value past 16 will already\n  lose a huge majority of devices (60%+) so I figured it doesn\u0027t\n  make sense to try to optimize for that backend\u0027s compatibility\n  for tier1 at all.\n\n- Detailed breakdown per backend:\n  - Vulkan:\n    - Relvant limit is maxPerStageDescriptorSampledImages.\n    - From the link below, it seems like the increased tier\n      will contain a vast majority of devices.\n    - https://vulkan.gpuinfo.org/displaydevicelimit.php?name\u003dmaxPerStageDescriptorSampledImages\u0026platform\u003dall\n\n  - Metal:\n    - Relevant limit is maxTextureArgumentEntriesPerFunc.\n    - From our own tables:\n      https://source.chromium.org/chromium/chromium/src/+/main:third_party/dawn/src/dawn/native/metal/PhysicalDeviceMTL.mm;l\u003d816;drc\u003dbf712ec1a13783224debb691ba88ad5c15b93194\n    - This means we will drop support for tier1 for Apple\n      families 1, 2, and 3. This is up to around the iPhone 7.\n\n  - D3D12:\n    - Relevant limit is maxSRVsPerStage.\n    - From our own tables:\n      https://source.chromium.org/chromium/chromium/src/+/main:third_party/dawn/src/dawn/native/d3d12/PhysicalDeviceD3D12.cpp;l\u003d316?q\u003dmaxSampledTexturesPerShaderStage\u0026ss\u003dchromium%2Fchromium%2Fsrc:third_party%2Fdawn%2Fsrc%2Fdawn%2Fnative%2F\n    - Minimum seems like it was already 128 for all D3D12, so increasing\n      to 48 should be no-op.\n\n  - D3D11:\n    - Relevant limit is D3D11_COMMONSHADER_INPUT_RESOURCE_SLOT_COUNT.\n    - Documentation of it:\n      https://microsoft.github.io/windows-docs-rs/doc/windows/Win32/Graphics/Direct3D11/constant.D3D11_COMMONSHADER_INPUT_RESOURCE_SLOT_COUNT.html\n    - The documentation also implies the min was already 128, so no-op.\n\n  - OpenGL:\n    - Relevant limits are GL_MAX_TEXTURE_IMAGE_UNITS and\n      GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS.\n    - From gpuinfo:\n      https://opengles.gpuinfo.org/displaycapability.php?name\u003dGL_MAX_TEXTURE_IMAGE_UNITS\u0026esversion\u003d2\n      https://opengles.gpuinfo.org/displaycapability.php?name\u003dGL_MAX_TEXTURE_IMAGE_UNITS\u0026esversion\u003d3\n      https://opengles.gpuinfo.org/displaycapability.php?name\u003dGL_MAX_VERTEX_TEXTURE_IMAGE_UNITS\u0026esversion\u003d2\n      https://opengles.gpuinfo.org/displaycapability.php?name\u003dGL_MAX_VERTEX_TEXTURE_IMAGE_UNITS\u0026esversion\u003d3\n    - For what it\u0027s worth, increasing this limit for this backend past\n      16, necessarily loses a majority of devices. As a result, there\u0027s\n      really no point in considering the results for this class.\n\nBug: 475255737\nChange-Id: I079140e7fbe2823b4d6e9b4c212e56a2abda6128\nReviewed-on: https://dawn-review.googlesource.com/c/dawn/+/287198\nReviewed-by: Kai Ninomiya \u003ckainino@chromium.org\u003e\nCommit-Queue: Kai Ninomiya \u003ckainino@chromium.org\u003e\nAuto-Submit: Loko Kung \u003clokokung@google.com\u003e\nCommit-Queue: Loko Kung \u003clokokung@google.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "6401e185a2e9d68fafab6ab26419158c4c1061ee",
      "old_mode": 33188,
      "old_path": "src/dawn/common/Constants.h",
      "new_id": "1d8c325e22ae8ea75001b61939c9ee07ed0dd25e",
      "new_mode": 33188,
      "new_path": "src/dawn/common/Constants.h"
    },
    {
      "type": "modify",
      "old_id": "32b732348443d55fe863fa25290f1a54be026694",
      "old_mode": 33188,
      "old_path": "src/dawn/native/Limits.cpp",
      "new_id": "9c1bb038bed3c5ed780cf62774ee1cd27fdcddc2",
      "new_mode": 33188,
      "new_path": "src/dawn/native/Limits.cpp"
    },
    {
      "type": "modify",
      "old_id": "8d1b77ae6a0473bee2e4347d8fc99c4cd086f121",
      "old_mode": 33188,
      "old_path": "src/dawn/tests/end2end/ArchTierLimitsExhaustive.cpp",
      "new_id": "5bc2f8a3722a6b30dea40b4ea882e7fd89eea059",
      "new_mode": 33188,
      "new_path": "src/dawn/tests/end2end/ArchTierLimitsExhaustive.cpp"
    },
    {
      "type": "modify",
      "old_id": "b9d3b67f63710b8614c1ace83063c6ca3228b08a",
      "old_mode": 33188,
      "old_path": "webgpu-cts/expectations.txt",
      "new_id": "3e34896d5f4befee510f8ca5ba564f83b44807ea",
      "new_mode": 33188,
      "new_path": "webgpu-cts/expectations.txt"
    }
  ]
}