Autogen webgpu dawn, removing wgpu-py dependency (f16 support part 1) (tinygrad#8646)

* Switch to dawn, all tests passing locally

* Use dawn-python

* Skip failing test

* Skip midcast and fix timestamp on metal ci

* Autogen webgpu

* Try fetch dawn lib again

* /usr/lib

* Without lib prefix

* Test autogen diff

* Delete webgpu support, move everything to ops_webgpu

* mypy fix

* Simplify, refactor

* Line savings

* No ResultContainer

* Type annotation for result

* Some more simplifications

* Why was this explicit sync used at all?

* Refactor: delete functions that are only used once

* Create shader module inline

* Clear unit tests cache, maybe that solves it

* That wasn't it

* Try deleting cache to pass failing weight compare

* weights_only=False for pytorch 2.6

* Simplify ctype array creation

* Remove nanosecond precision timestamps

* Simplify error handling

* Refactor, add back type annotations

* Deleted custom submit function, refactor

* read_buffer simplify

* Fix use after free, refactor

* Simplify supported_features

* Runtime docs

---------

Co-authored-by: George Hotz <72895+geohot@users.noreply.github.com>

Author: wpmed92

.github/workflows/test.yml

@@ -72,6 +72,9 @@ jobs:
         sudo apt update -y || true
         sudo apt install -y --no-install-recommends git g++ cmake ninja-build llvm-15-dev zlib1g-dev libglew-dev \
           flex bison libfl-dev libboost-thread-dev libboost-filesystem-dev nvidia-cuda-toolkit-gcc libzstd-dev
+    - name: Install packages (webgpu)
+      run: |
+        sudo curl -L https://github.com/wpmed92/pydawn/releases/download/v0.1.6/libwebgpu_dawn.so -o /usr/local/lib/libwebgpu_dawn.so
     - name: Install packages (amd)
       run: |
         echo 'Acquire::http::Pipeline-Depth "5";' | sudo tee -a /etc/apt/apt.conf.d/99parallel
@@ -132,6 +135,11 @@ jobs:
         ./autogen_stubs.sh io_uring
         diff /tmp/libc.py.bak tinygrad/runtime/autogen/libc.py
         diff /tmp/io_uring.py.bak tinygrad/runtime/autogen/io_uring.py
+    - name: Verify WebGPU autogen
+      run: |
+        cp tinygrad/runtime/autogen/webgpu.py /tmp/webgpu.py.bak
+        ./autogen_stubs.sh webgpu
+        diff /tmp/webgpu.py.bak tinygrad/runtime/autogen/webgpu.py
     - name: Verify LLVM autogen
       run: |
         cp tinygrad/runtime/autogen/llvm.py /tmp/llvm.py.bak
@@ -401,7 +409,10 @@ jobs:
         path: ~/.local/lib/python3.11/site-packages
         key: webgpu-testing-user3-packages-${{ hashFiles('**/setup.py') }}
     - name: Install Dependencies
-      run: pip install --user -e '.[webgpu,testing]' --extra-index-url https://download.pytorch.org/whl/cpu
+      run: pip install --user -e '.[testing]' --extra-index-url https://download.pytorch.org/whl/cpu
+    - name: Install dawn (WebGPU)
+      run: |
+        sudo curl -L https://github.com/wpmed92/pydawn/releases/download/v0.1.6/libwebgpu_dawn.so -o /usr/lib/libwebgpu_dawn.so
     - name: Install dependencies for software-based vulkan
       run: |
         sudo apt update -y || true
@@ -417,7 +428,7 @@ jobs:
         WEBGPU=1 DEBUG=4 FORWARD_ONLY=1 python3 test/test_ops.py TestOps.test_add
     - name: Run selected webgpu tests
       run: |
-          WEBGPU=1 WGPU_BACKEND_TYPE=Vulkan python3 -m pytest -n=auto test/ --ignore=test/external --ignore=test/models --ignore=test/unit \
+          WEBGPU=1 python3 -m pytest -n=auto test/ --ignore=test/external --ignore=test/models --ignore=test/unit \
           --ignore=test/test_copy_speed.py --ignore=test/test_rearrange_einops.py --ignore=test/test_speed_v_torch.py --ignore=test/test_transcendental.py \
           --ignore=test/test_fuzz_shape_ops.py --ignore=test/test_linearizer_failures.py --durations=20
     - name: Run process replay tests
@@ -447,6 +458,10 @@ jobs:
         key: metal-m1-testing-user3-packages-${{ hashFiles('**/setup.py') }}
     - name: Install Dependencies
       run: pip install --user -e '.[webgpu,testing]' --extra-index-url https://download.pytorch.org/whl/cpu
+    - name: Install dawn (WebGPU)
+      run: |
+        sudo mkdir -p /usr/local/lib
+        sudo curl -L https://github.com/wpmed92/pydawn/releases/download/v0.1.6/libwebgpu_dawn.dylib -o /usr/local/lib/libwebgpu_dawn.dylib
     - name: Cache downloads
       uses: actions/cache@v4
       with:
@@ -478,7 +493,7 @@ jobs:
       run: TRANSCENDENTAL=2 python -m pytest -n=auto test/test_ops.py::TestOps::test_sin test/test_ops.py::TestOps::test_cos test/test_ops.py::TestOps::test_tan test/test_ops.py::TestOps::test_exp test/test_ops.py::TestOps::test_log --durations=20
     # WebGPU e2e tests
     - name: Build WEBGPU Efficientnet
-      run: WEBGPU=1 WGPU_BACKEND_TYPE=Metal python3 -m examples.compile_efficientnet
+      run: WEBGPU=1 python3 -m examples.compile_efficientnet
     - name: Clean npm cache
       run: npm cache clean --force
     - name: Install Puppeteer

autogen_stubs.sh

@@ -357,6 +357,14 @@ generate_am() {
   fixup $BASE/am/smu_v13_0_0.py
 }
 
+generate_webgpu() {
+  clang2py -l /usr/local/lib/libwebgpu_dawn.so extra/webgpu/webgpu.h -o $BASE/webgpu.py
+  fixup $BASE/webgpu.py
+  sed -i 's/import ctypes/import ctypes, ctypes.util/g' $BASE/webgpu.py
+  sed -i "s|ctypes.CDLL('/usr/local/lib/libwebgpu_dawn.so')|ctypes.CDLL(ctypes.util.find_library('webgpu_dawn'))|g" $BASE/webgpu.py
+  python3 -c "import tinygrad.runtime.autogen.webgpu"
+}
+
 if [ "$1" == "opencl" ]; then generate_opencl
 elif [ "$1" == "hip" ]; then generate_hip
 elif [ "$1" == "comgr" ]; then generate_comgr
@@ -375,6 +383,7 @@ elif [ "$1" == "kgsl" ]; then generate_kgsl
 elif [ "$1" == "adreno" ]; then generate_adreno
 elif [ "$1" == "pci" ]; then generate_pciaccess
 elif [ "$1" == "vfio" ]; then generate_vfio
-elif [ "$1" == "all" ]; then generate_opencl; generate_hip; generate_comgr; generate_cuda; generate_nvrtc; generate_hsa; generate_kfd; generate_nv; generate_amd; generate_io_uring; generate_libc; generate_am
+elif [ "$1" == "webgpu" ]; then generate_webgpu
+elif [ "$1" == "all" ]; then generate_opencl; generate_hip; generate_comgr; generate_cuda; generate_nvrtc; generate_hsa; generate_kfd; generate_nv; generate_amd; generate_io_uring; generate_libc; generate_am; generate_webgpu
 else echo "usage: $0 <type>"
 fi

docs/runtime.md

@@ -12,3 +12,4 @@ tinygrad supports various runtimes, enabling your code to scale across a wide ra
 | [GPU (OpenCL)](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_gpu.py) | Accelerates computations using OpenCL on GPUs | OpenCL 2.0 compatible device |
 | [CLANG (C Code)](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_clang.py) | Runs on CPU using the clang compiler | `clang` compiler in system `PATH` |
 | [LLVM (LLVM IR)](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_llvm.py) | Runs on CPU using the LLVM compiler infrastructure | llvm libraries installed and findable |
+| [WEBGPU](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/ops_webgpu.py) | Runs on GPU using the Dawn WebGPU engine (used in Google Chrome) | Dawn library installed and findable. Download binaries [here](https://github.com/wpmed92/pydawn/releases/tag/v0.1.6). |