Holoscan OpenEmbedded Layer v0.3.0 Release

nvidia-holoscan · Oct 20, 2022 · b8872fe · b8872fe
1 parent 7eaeb20
commit b8872fe
Show file tree

Hide file tree

Showing 81 changed files with 2,802 additions and 1,185 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -0,0 +1,56 @@
+## [0.3.0] - 2022-10-19
+### Added
+- Support for the NVIDIA IGX Orin Developer Kit (`holoscan-devkit`).
+- Ability to support both iGPU and dGPU configurations via separate includes
+(`holoscan-igpu.conf` and `holoscan-dgpu.conf`, respectively).
+- Mellanox OFED and Rivermax Support.
+
+### Changed
+- dGPU kernel modules are now built from the [NVIDIA Linux Open GPU Kernel
+Module Source](https://github.com/NVIDIA/open-gpu-kernel-modules).
+- CUDA samples for the dGPU are now built from the [CUDA Samples GitHub
+Repo](https://github.com/NVIDIA/cuda-samples).
+
+### Known Issues
+- Holoscan SDK Ultrasound segmentation application is not functional on NVIDIA IGX Orin Developer Kit (holoscan-devkit) with iGPU configuration.
+- Holoscan SDK High-Speed Endoscopy application is not supported in deployment stack.
+
+### Component Versions
+|                 | iGPU and dGPU |
+| --------------- | ------------- |
+| L4T             | 35.1.0        |
+| Holoscan SDK    | 0.3.0         |
+| GXF             | 2.4.3         |
+| AJA NTV2        | 16.2.0        |
+| MLNX OFED       | 5.7-1.0.2.0   |
+| Rivermax        | 1.11.11       |
+
+|                | iGPU     | dGPU      |
+| -------------- | -------- | --------- |
+| GPU Drivers    | 35.1.0   | 515.65.01 |
+| CUDA           | 11.4.239 | 11.7.1    |
+| cuDNN          | 8.4.1.50 | 8.5.0.96  |
+| TensorRT       | 8.4.1    | 8.4.3.1   |
+| nvidia-peermem | N/A      | 1.3.0     |
+
+
+## [0.2.0] - 2022-07-12
+### Added
+- Initial Holoscan MGX release, corresponding with the v0.2.0 release of the
+[Clara Holoscan Embedded SDK](https://github.com/NVIDIA/clara-holoscan-embedded-sdk).
+- Supports dGPU mode on the Clara AGX Developer Kit (`clara-agx-xavier-devkit`).
+
+### Component Versions
+|                | dGPU     |
+| -------------- | -------- |
+| L4T            | 32.7.1   |
+| NVIDIA Drivers | 470.94   |
+| CUDA           | 11.4.3   |
+| cuDNN          | 8.2.4.15 |
+| TensorRT       | 8.2.1.8  |
+| GXF            | 2.4.2    |
+| AJA NTV2       | 16.2.0   |
+| Holoscan SDK   | 0.2.0    |
+
+[0.3.0]: https://github.com/NVIDIA/meta-tegra-clara-holoscan-mgx/compare/v0.2.0...v0.3.0
+[0.2.0]: https://github.com/NVIDIA/meta-tegra-clara-holoscan-mgx/releases/tag/v0.2.0
diff --git a/README.md b/README.md
diff --git a/classes/cuda.bbclass b/classes/cuda.bbclass
diff --git a/conf/holoscan-common.conf b/conf/holoscan-common.conf
@@ -0,0 +1,28 @@
+# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+# DEALINGS IN THE SOFTWARE.
+
+# Enable OpenGL and Vulkan (required for Holoscan SDK)
+DISTRO_FEATURES:append = " opengl vulkan"
+
+# Use systemd as system init manager (for e.g. nvfancontrol service)
+DISTRO_FEATURES:append = " systemd"
+DISTRO_FEATURES_BACKFILL_CONSIDERED = "sysvinit"
+VIRTUAL-RUNTIME_init_manager = "systemd"
+VIRTUAL-RUNTIME_initscripts = ""
diff --git a/conf/holoscan-dgpu.conf b/conf/holoscan-dgpu.conf
@@ -0,0 +1,115 @@
+# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+# DEALINGS IN THE SOFTWARE.
+
+require holoscan-common.conf
+
+TEGRA_DGPU = "1"
+
+# Display Drivers
+IMAGE_INSTALL:append = " \
+    nvidia-open-gpu-kernel-modules \
+    nvidia-dgpu-drivers \
+"
+
+PREFERRED_RPROVIDER_kernel-module-nvidia = "nvidia-open-gpu-kernel-modules"
+PREFERRED_RPROVIDER_kernel-module-nvidia-drm = "nvidia-open-gpu-kernel-modules"
+PREFERRED_RPROVIDER_kernel-module-nvidia-modeset = "nvidia-open-gpu-kernel-modules"
+PREFERRED_RPROVIDER_tegra-libraries-cuda = "nvidia-dgpu-drivers"
+PREFERRED_RPROVIDER_tegra-libraries-eglcore = "nvidia-dgpu-drivers"
+PREFERRED_RPROVIDER_tegra-libraries-glescore = "nvidia-dgpu-drivers"
+PREFERRED_RPROVIDER_tegra-libraries-glxcore = "nvidia-dgpu-drivers"
+PREFERRED_RPROVIDER_tegra-libraries-vulkan = "nvidia-dgpu-drivers"
+
+XSERVER = "xserver-xorg xf86-input-evdev nvidia-dgpu-drivers xserver-xorg-module-libwfb"
+
+# CUDA
+PREFERRED_VERSION_cuda-cccl = "11.7.91-1"
+PREFERRED_VERSION_cuda-command-line-tools = "11.7.1-1"
+PREFERRED_VERSION_cuda-compiler = "11.7.1-1"
+PREFERRED_VERSION_cuda-cudart = "11.7.99-1"
+PREFERRED_VERSION_cuda-cuobjdump = "11.7.91-1"
+PREFERRED_VERSION_cuda-cupti = "11.7.101-1"
+PREFERRED_VERSION_cuda-cuxxfilt = "11.7.91-1"
+PREFERRED_VERSION_cuda-driver = "11.7.99-1"
+PREFERRED_VERSION_cuda-gdb = "11.7.91-1"
+PREFERRED_VERSION_cuda-libraries = "11.7.1-1"
+PREFERRED_VERSION_cuda-nvcc = "11.7.99-1"
+PREFERRED_VERSION_cuda-nvcc-headers = "11.7.99-1"
+PREFERRED_VERSION_cuda-nvdisasm = "11.7.91-1"
+PREFERRED_VERSION_cuda-nvml = "11.7.91-1"
+PREFERRED_VERSION_cuda-nvprof = "11.7.101-1"
+PREFERRED_VERSION_cuda-nvprune = "11.7.91-1"
+PREFERRED_VERSION_cuda-nvrtc = "11.7.99-1"
+PREFERRED_VERSION_cuda-nvtx = "11.7.91-1"
+PREFERRED_VERSION_cuda-profiler-api = "11.7.101-1"
+PREFERRED_VERSION_cuda-sanitizer = "11.7.91-1"
+PREFERRED_VERSION_cuda-toolkit = "11.7.1-1"
+PREFERRED_VERSION_libcublas = "11.10.3.66-1"
+PREFERRED_VERSION_libcufft = "10.7.2.91-1"
+PREFERRED_VERSION_libcurand = "10.2.10.91-1"
+PREFERRED_VERSION_libcusolver = "11.4.0.1-1"
+PREFERRED_VERSION_libcusparse = "11.7.4.91-1"
+PREFERRED_VERSION_libnpp = "11.7.4.75-1"
+PREFERRED_VERSION_cuda-cccl-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-command-line-tools-native = "11.7.1-1"
+PREFERRED_VERSION_cuda-compiler-native = "11.7.1-1"
+PREFERRED_VERSION_cuda-cudart-native = "11.7.99-1"
+PREFERRED_VERSION_cuda-cuobjdump-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-cupti-native = "11.7.101-1"
+PREFERRED_VERSION_cuda-cuxxfilt-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-driver-native = "11.7.99-1"
+PREFERRED_VERSION_cuda-gdb-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-libraries-native = "11.7.1-1"
+PREFERRED_VERSION_cuda-nvcc-native = "11.7.99-1"
+PREFERRED_VERSION_cuda-nvcc-headers-native = "11.7.99-1"
+PREFERRED_VERSION_cuda-nvdisasm-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-nvml-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-nvprof-native = "11.7.101-1"
+PREFERRED_VERSION_cuda-nvprune-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-nvrtc-native = "11.7.99-1"
+PREFERRED_VERSION_cuda-nvtx-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-profiler-api-native = "11.7.101-1"
+PREFERRED_VERSION_cuda-sanitizer-native = "11.7.91-1"
+PREFERRED_VERSION_cuda-toolkit-native = "11.7.1-1"
+PREFERRED_VERSION_libcublas-native = "11.10.3.66-1"
+PREFERRED_VERSION_libcufft-native = "10.7.2.91-1"
+PREFERRED_VERSION_libcurand-native = "10.2.10.91-1"
+PREFERRED_VERSION_libcusolver-native = "11.4.0.1-1"
+PREFERRED_VERSION_libcusparse-native = "11.7.4.91-1"
+PREFERRED_VERSION_libnpp-native = "11.7.4.75-1"
+
+# CUDA Samples
+PREFERRED_VERSION_cuda-samples = "11.6"
+
+# cuDNN
+PREFERRED_VERSION_cudnn = "8.5.0.96"
+
+# TensorRT
+PREFERRED_VERSION_tensorrt-samples = "8.4.3.1"
+PREFERRED_PROVIDER_tensorrt-core = "tensorrt"
+PREFERRED_PROVIDER_tensorrt-plugins = "tensorrt"
+PREFERRED_PROVIDER_tensorrt-trtexec = "tensorrt"
+
+# Override the CUDA version (11.7) and GPU arch to:
+#    7.5 (Turing) for the RTX 6000 in the Clara AGX Devkit.
+#    8.6 (Ampere) for the RTX A6000 in the Holoscan Devkit.
+CUDA_VERSION = "11.7"
+TEGRA_CUDA_ARCHITECTURE:clara-agx-xavier-devkit ?= "75"
+TEGRA_CUDA_ARCHITECTURE:holoscan-devkit ?= "86"