diff --git a/.github/workflows/backend.yml b/.github/workflows/backend.yml index 70adcafcfb18..8fde80212fab 100644 --- a/.github/workflows/backend.yml +++ b/.github/workflows/backend.yml @@ -197,6 +197,18 @@ jobs: backend: "rerankers" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-rerankers' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "rerankers" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -209,6 +221,18 @@ jobs: backend: "llama-cpp" dockerfile: "./backend/Dockerfile.llama-cpp" context: "./" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-llama-cpp' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "llama-cpp" + dockerfile: "./backend/Dockerfile.llama-cpp" + context: "./" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -221,6 +245,18 @@ jobs: backend: "vllm" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-vllm' + runs-on: 'arc-runner-set' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "vllm" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -233,6 +269,18 @@ jobs: backend: "transformers" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-transformers' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "transformers" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -245,7 +293,19 @@ jobs: backend: "diffusers" dockerfile: "./backend/Dockerfile.python" context: "./backend" - # CUDA 12 additional backends + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-diffusers' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "diffusers" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" + # CUDA additional backends - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -258,6 +318,18 @@ jobs: backend: "kokoro" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-kokoro' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "kokoro" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -270,6 +342,18 @@ jobs: backend: "faster-whisper" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-faster-whisper' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "faster-whisper" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -282,6 +366,18 @@ jobs: backend: "coqui" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-coqui' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "coqui" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -294,6 +390,18 @@ jobs: backend: "bark" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-bark' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "bark" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" @@ -306,6 +414,18 @@ jobs: backend: "chatterbox" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-chatterbox' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "chatterbox" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" # hipblas builds - build-type: 'hipblas' cuda-major-version: "" @@ -625,6 +745,18 @@ jobs: backend: "stablediffusion-ggml" dockerfile: "./backend/Dockerfile.golang" context: "./" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13-stablediffusion-ggml' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "stablediffusion-ggml" + dockerfile: "./backend/Dockerfile.golang" + context: "./" - build-type: 'cublas' cuda-major-version: "11" cuda-minor-version: "7" @@ -710,6 +842,18 @@ jobs: backend: "whisper" dockerfile: "./backend/Dockerfile.golang" context: "./" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-12-whisper' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "whisper" + dockerfile: "./backend/Dockerfile.golang" + context: "./" - build-type: 'cublas' cuda-major-version: "11" cuda-minor-version: "7" @@ -846,6 +990,18 @@ jobs: backend: "rfdetr" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-12-rfdetr' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "rfdetr" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "11" cuda-minor-version: "7" @@ -907,6 +1063,18 @@ jobs: backend: "exllama2" dockerfile: "./backend/Dockerfile.python" context: "./backend" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-12-exllama2' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + backend: "exllama2" + dockerfile: "./backend/Dockerfile.python" + context: "./backend" - build-type: 'cublas' cuda-major-version: "11" cuda-minor-version: "7" diff --git a/.github/workflows/image-pr.yml b/.github/workflows/image-pr.yml index a2410b22827e..4322a20c3009 100644 --- a/.github/workflows/image-pr.yml +++ b/.github/workflows/image-pr.yml @@ -34,6 +34,15 @@ jobs: fail-fast: false matrix: include: + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'false' + tag-suffix: '-gpu-nvidia-cuda-13' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + makeflags: "--jobs=3 --output-sync=target" - build-type: 'cublas' cuda-major-version: "12" cuda-minor-version: "0" diff --git a/.github/workflows/image.yml b/.github/workflows/image.yml index 3864930d03ed..c7a7d503fd87 100644 --- a/.github/workflows/image.yml +++ b/.github/workflows/image.yml @@ -100,6 +100,17 @@ jobs: skip-drivers: 'false' makeflags: "--jobs=4 --output-sync=target" aio: "-aio-gpu-nvidia-cuda-12" + - build-type: 'cublas' + cuda-major-version: "13" + cuda-minor-version: "0" + platforms: 'linux/amd64' + tag-latest: 'auto' + tag-suffix: '-gpu-nvidia-cuda-13' + runs-on: 'ubuntu-latest' + base-image: "ubuntu:22.04" + skip-drivers: 'false' + makeflags: "--jobs=4 --output-sync=target" + aio: "-aio-gpu-nvidia-cuda-13" - build-type: 'vulkan' platforms: 'linux/amd64' tag-latest: 'auto'