Revert: Nvidia OpenCL

Dockerfile.opencl.nvidia

@@ -1,36 +1,30 @@
 FROM python:3.12.7
 
-ARG PYTORCH_OCL_VERSION=0.2.0
-ARG TORCH_VERSION=2.5
-ARG PYTHON_VERSION=cp312
-ARG PLATFORM=linux_x86_64
-
-ENV WHL_FILE=pytorch_ocl-${PYTORCH_OCL_VERSION}+torch${TORCH_VERSION}-${PYTHON_VERSION}-none-${PLATFORM}.whl
-ENV WHL_URL=https://github.com/artyom-beilis/pytorch_dlprim/releases/download/${PYTORCH_OCL_VERSION}/${WHL_FILE}
-
-ENV NVIDIA_VISIBLE_DEVICES all
-ENV NVIDIA_DRIVER_CAPABILITIES compute,utility
+ENV DEBIAN_FRONTEND="noninteractive"
 
 WORKDIR /app
 
 RUN apt update && \
     apt full-upgrade -y && \
-    apt install python3 python3-full python3-pip python3-venv git wget ocl-icd-opencl-dev opencl-clhpp-headers opencl-c-headers opencl-headers ocl-icd-libopencl1 clinfo -y && \
-    python3 -m venv /app/venv && \
-    /app/venv/bin/pip install --upgrade pip
-
-# Configure OpenCL ICD loaders
-RUN mkdir -p /etc/OpenCL/vendors && \
-    echo "libnvidia-opencl.so.1" > /etc/OpenCL/vendors/nvidia.icd
+    apt install python3 python3-dev git ocl-icd-opencl-dev opencl-clhpp-headers opencl-c-headers ocl-icd-libopencl1 clinfo -y && \
+    pip3 install --upgrade pip
 
 COPY requirements.txt /app/requirements.txt
-RUN /app/venv/bin/pip install --no-cache-dir -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cpu
+RUN pip3 install --no-cache-dir -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cpu
 
 # Install OpenCL backend for PyTorch
-RUN wget ${WHL_URL} && \
-    /app/venv/bin/pip install ${WHL_FILE} && \
-    rm ${WHL_FILE}
+RUN wget https://github.com/artyom-beilis/pytorch_dlprim/releases/download/0.2.0/pytorch_ocl-0.2.0+torch2.5-cp312-none-linux_x86_64.whl && \
+    pip3 install pytorch_ocl-0.2.0+torch2.5-cp312-none-linux_x86_64.whl && \
+    rm pytorch_ocl-0.2.0+torch2.5-cp312-none-linux_x86_64.whl
+
+# Install Nvidia OpenCL drivers
+RUN mkdir -p /etc/OpenCL/vendors && \
+    echo "libnvidia-opencl.so.1" > /etc/OpenCL/vendors/nvidia.icd
+ENV NVIDIA_VISIBLE_DEVICES all
+ENV NVIDIA_DRIVER_CAPABILITIES compute,utility,display
+
+RUN rm -f /etc/OpenCL/vendors/mesa.icd
 
 COPY . /app
 
-CMD ["/app/venv/bin/python", "-m", "fastapi", "run", "main.py", "--proxy-headers", "--host", "0.0.0.0", "--port", "8000"]
+CMD ["fastapi", "run", "main.py", "--proxy-headers", "--host", "0.0.0.0", "--port", "8000"]

README.md

@@ -9,15 +9,16 @@
 
 - For ease of use it's recommended to use the provided [docker-compose.yml](https://github.com/doppeltilde/image_video_classification/blob/main/docker-compose.yml).
 
-**CPU Support:** Use the `latest` tag.
+### **CPU Support**
+Use the `latest` tag.
 ```yml
 services:
   image_video_classification:
     image: ghcr.io/doppeltilde/image_video_classification:latest
     ports:
       - "8000:8000"
     volumes:
-      - models:/root/.cache/huggingface/hub:rw
+      - ./models:/root/.cache/huggingface/hub:rw
     environment:
       - DEFAULT_MODEL_NAME
       - BATCH_SIZE
@@ -26,20 +27,18 @@ services:
       - USE_API_KEYS
       - API_KEYS
     restart: unless-stopped
-
-volumes:
-  models:
 ```
 
-**NVIDIA GPU Support:** Use the `latest-cuda` tag.
+### **NVIDIA GPU Support**
+**CUDA:**
 ```yml
 services:
   image_video_classification_cuda:
     image: ghcr.io/doppeltilde/image_video_classification:latest-cuda
     ports:
       - "8000:8000"
     volumes:
-      - models:/root/.cache/huggingface/hub:rw
+      - ./models:/root/.cache/huggingface/hub:rw
     environment:
       - DEFAULT_MODEL_NAME
       - BATCH_SIZE
@@ -55,9 +54,58 @@ services:
             - driver: nvidia
               count: all
               capabilities: [ gpu ]
+```
+**OpenCL:**
+```yml
+services:
+  image_video_classification_opencl:
+    image: ghcr.io/doppeltilde/image_video_classification:latest-opencl-nvidia
+    ports:
+      - "8000:8000"
+    volumes:
+      - ./models:/root/.cache/huggingface/hub:rw
+    environment:
+      - DEFAULT_MODEL_NAME
+      - BATCH_SIZE
+      - ACCESS_TOKEN
+      - DEFAULT_SCORE
+      - USE_API_KEYS
+      - API_KEYS
+    restart: unless-stopped
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [ gpu ]
+```
 
-volumes:
-  models:
+### **AMD GPU Support**
+**OpenCL:**
+```yml
+services:
+  image_video_classification_cuda:
+    image: ghcr.io/doppeltilde/image_video_classification:latest-cuda
+    ports:
+      - "8000:8000"
+    volumes:
+      - ./models:/root/.cache/huggingface/hub:rw
+    environment:
+      - DEFAULT_MODEL_NAME
+      - BATCH_SIZE
+      - ACCESS_TOKEN
+      - DEFAULT_SCORE
+      - USE_API_KEYS
+      - API_KEYS
+    restart: unless-stopped
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [ gpu ]
 ```
 
 - Create a `.env` file and set the preferred values.