Merge branch 'main' of github.com:pytorch/torchcodec into audio_support

Author: NicolasHug

.github/workflows/cpp_tests.yaml

@@ -19,7 +19,9 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        ffmpeg-version-for-tests: ['4.4.2', '5.1.2', '6.1.1', '7.0.1']
+        # TODO put back 6.1.1. See
+        # https://github.com/pytorch/torchcodec/issues/518
+        ffmpeg-version-for-tests: ['4.4.2', '5.1.2', '7.0.1']
     steps:
       - name: Check out repo
         uses: actions/checkout@v3

.github/workflows/docs.yaml

@@ -75,9 +75,6 @@ jobs:
         uses: pytorch/test-infra/.github/actions/setup-miniconda@main
         with:
           python-version: ${{ matrix.python-version }}
-          #
-          # For some reason nvidia::libnpp=12.4 doesn't install but nvidia/label/cuda-12.4.0::libnpp does.
-          # So we use the latter convention for libnpp.
           # We install conda packages at the start because otherwise conda may have conflicts with dependencies.
           default-packages: "nvidia/label/cuda-${{ matrix.cuda-version }}.0::libnpp nvidia::cuda-nvrtc=${{ matrix.cuda-version }} nvidia::cuda-toolkit=${{ matrix.cuda-version }} nvidia::cuda-cudart=${{ matrix.cuda-version }} nvidia::cuda-driver-dev=${{ matrix.cuda-version }} conda-forge::ffmpeg=${{ matrix.ffmpeg-version-for-tests }}"
       - name: Check env

.github/workflows/linux_cuda_wheel.yaml

@@ -66,9 +66,10 @@ jobs:
           # For the actual release we should add that label and change this to
           # include more python versions.
         python-version: ['3.9']
-        cuda-version: ['11.8', '12.4', '12.6']
+        cuda-version: ['11.8', '12.6']
         # TODO: put back ffmpeg 5 https://github.com/pytorch/torchcodec/issues/325
-        ffmpeg-version-for-tests: ['6', '7']
+        ffmpeg-version-for-tests: ['4.4.2', '6', '7']
+
     container:
       image: "pytorch/manylinux2_28-builder:cuda${{ matrix.cuda-version }}"
       options: "--gpus all -e NVIDIA_DRIVER_CAPABILITIES=video,compute,utility"
@@ -88,9 +89,6 @@ jobs:
         uses: pytorch/test-infra/.github/actions/setup-miniconda@main
         with:
           python-version: ${{ matrix.python-version }}
-          #
-          # For some reason nvidia::libnpp=12.4 doesn't install but nvidia/label/cuda-12.4.0::libnpp does.
-          # So we use the latter convention for libnpp.
           # We install conda packages at the start because otherwise conda may have conflicts with dependencies.
           default-packages: "nvidia/label/cuda-${{ matrix.cuda-version }}.0::libnpp nvidia::cuda-nvrtc=${{ matrix.cuda-version }} nvidia::cuda-toolkit=${{ matrix.cuda-version }} nvidia::cuda-cudart=${{ matrix.cuda-version }} nvidia::cuda-driver-dev=${{ matrix.cuda-version }} conda-forge::ffmpeg=${{ matrix.ffmpeg-version-for-tests }}"
       - name: Check env
@@ -141,7 +139,7 @@ jobs:
           ${CONDA_RUN} python test/decoders/manual_smoke_test.py
       - name: Run Python tests
         run: |
-          ${CONDA_RUN} FAIL_WITHOUT_CUDA=1 pytest test -vvv
+          ${CONDA_RUN} FAIL_WITHOUT_CUDA=1 pytest test -v --tb=short
       - name: Run Python benchmark
         run: |
           ${CONDA_RUN} time python benchmarks/decoders/gpu_benchmark.py --devices=cuda:0,cpu --resize_devices=none

README.md

@@ -146,14 +146,12 @@ First, make sure you have a GPU that has NVDEC hardware that can decode the
 format you want. Refer to Nvidia's GPU support matrix for more details
 [here](https://developer.nvidia.com/video-encode-and-decode-gpu-support-matrix-new).
 
-1. Install CUDA Toolkit. Pytorch and TorchCodec supports CUDA Toolkit
-   versions 11.8, 12.1 or 12.4. In particular TorchCodec depends on
-   CUDA libraries libnpp and libnvrtc (which are part of CUDA Toolkit).
+1. Install Pytorch corresponding to your CUDA Toolkit using the
+   [official instructions](https://pytorch.org/get-started/locally/). You'll
+   need the `libnpp` and `libnvrtc` CUDA libraries, which are usually part of
+   the CUDA Toolkit.
 
-2. Install Pytorch that corresponds to your CUDA Toolkit version using the
-   [official instructions](https://pytorch.org/get-started/locally/).
-
-3. Install or compile FFmpeg with NVDEC support.
+2. Install or compile FFmpeg with NVDEC support.
    TorchCodec with CUDA should work with FFmpeg versions in [5, 7].
 
    If FFmpeg is not already installed, or you need a more recent version, an
@@ -184,13 +182,13 @@ format you want. Refer to Nvidia's GPU support matrix for more details
    ffmpeg -hwaccel cuda -hwaccel_output_format cuda -i test/resources/nasa_13013.mp4 -f null -
    ```
 
-4. Install TorchCodec by passing in an `--index-url` parameter that corresponds
+3. Install TorchCodec by passing in an `--index-url` parameter that corresponds
    to your CUDA Toolkit version, example:
 
    ```bash
-   # This corresponds to CUDA Toolkit version 12.4. It should be the same one
+   # This corresponds to CUDA Toolkit version 12.6. It should be the same one
    # you used when you installed PyTorch (If you installed PyTorch with pip).
-   pip install torchcodec --index-url=https://download.pytorch.org/whl/cu124
+   pip install torchcodec --index-url=https://download.pytorch.org/whl/cu126
    ```
 
    Note that without passing in the `--index-url` parameter, `pip` installs
@@ -207,7 +205,10 @@ The top row is a [Mandelbrot](https://ffmpeg.org/ffmpeg-filters.html#mandelbrot)
 generated from FFmpeg that has a resolution of 1280x720 at 60 fps and is 120 seconds long.
 The bottom row is [promotional video from NASA](https://download.pytorch.org/torchaudio/tutorial-assets/stream-api/NASAs_Most_Scientifically_Complex_Space_Observatory_Requires_Precision-MP4_small.mp4)
 that has a resolution of 960x540 at 29.7 fps and is 206 seconds long. Both videos were
-encoded with libx264 and yuv420p pixel format.
+encoded with libx264 and yuv420p pixel format. All decoders, except for TorchVision, used FFmpeg 6.1.2. TorchVision used FFmpeg 4.2.2.
+
+For TorchCodec, the "approx" label means that it was using [approximate mode](https://pytorch.org/torchcodec/stable/generated_examples/approximate_mode.html)
+for seeking.
 
 ## Planned future work