JetPack 3.3 — L4T R28.2.1 Release for Jetson TX1/TX2

Autonomous Machines Jetson & Embedded Systems Jetson TX2
1

We’re pleased to announce that JetPack 3.3 is now available for Jetson TX2/TX2i and Jetson TX1!

The highlight of this release is TensorRT 4.0, enabling support for TensorFlow’s TensorRT integration feature. Also, cuDNN has a small update to support the new TensorRT version, and all other JetPack components remain unchanged from JetPack 3.2.1. Please view the Release Notes.

JetPack 3.3 components:

  • L4T R28.2.1 (TX2/TX2i)
  • L4T R28.2    (TX1)
  • CUDA 9.0
  • cuDNN 7.1.5
  • TensorRT 4.0
  • VisionWorks 1.6

Download JetPack…https://developer.nvidia.com/embedded/jetpack
Release Notes…https://developer.nvidia.com/embedded/jetpack-notes
L4T Release Notes…https://developer.nvidia.com/embedded/dlc/release-notes-r2821
L4T R28.2.1 Page…https://developer.nvidia.com/embedded/linux-tegra
L4T Archive Page…https://developer.nvidia.com/embedded/linux-tegra-archive

2

I have download JetPack3.3 and find that the TensorRT version is still 3.0 and cuDNN version is still 7.0.5. What wrong?

3

Did you download JetPack 3.3 to a fresh directory? Which region are you in?

Can you post the .json files from under jetpack_downloads/ that JetPack created? (you can use [ code ] [ /code ] tags (no spaces) to post the contents)

4

I am in China.
The manifest.json contains:

{
    "manifest": {
        "name": "JetPack L4T", 
        "portfolio": [
            {
                "name": "JetPack L4T", 
                "version": "3.2.1", 
                "url": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/3.2.1/m8u2ki//JetPackL4T_321_b23/jetpack.json", 
                "installerMin": "3.2.1"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "3.1", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/013/json/jetpack-l4t-devzone.json", 
                "installerMin": "3.1"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "3.0", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/011/json/jetpack-l4t-devzone.json", 
                "installerMin": "3.0"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "2.3.1", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/010/json/jetpack-l4t-devzone.json", 
                "installerMin": "2.3"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "2.3", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/008/json/jetpack-l4t-devzone.json", 
                "installerMin": "2.3"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "2.2.1", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/005/json/jetpack-l4t-devzone.json", 
                "installerMin": "2.2.1"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "2.2", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/004/json/jetpack-l4t-devzone.json", 
                "installerMin": "2.2"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "2.1", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/002/json/jetpack-l4t-devzone.json", 
                "installerMin": "2.1"
            }, 
            {
                "name": "JetPack L4T", 
                "version": "2.0", 
                "url": "http://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/001/json/jetpack-l4t-devzone.json", 
                "installerMin": "1.0"
            }
        ]
    }
}
5 
{
    "package": {
        "version": "3.2.1", 
        "compDirectory": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/3.2.1/m8u2ki/JetPackL4T_321_b23/", 
        "division": [
            {
                "alias": "tx2_64bit", 
                "selected": "1", 
                "name": "Jetson TX2"
            }, 
            {
                "alias": "tx2i_64bit", 
                "selected": "1", 
                "name": "Jetson TX2i - r28.2"
            }, 
            {
                "alias": "tx1_64bit", 
                "selected": "0", 
                "name": "Jetson TX1"
            }
        ], 
        "category": [
            {
                "alias": "common", 
                "name": "Host - Ubuntu", 
                "description": ""
            }, 
            {
                "category": "common", 
                "division": "tx2i_64bit tx2_64bit tx1_64bit", 
                "alias": "visionworksgroup_host", 
                "name": "VisionWorks Pack", 
                "description": "NVIDIA VisionWorks for Host Ubuntu 16.04 OS"
            }, 
            {
                "alias": "jetson_64_tx1", 
                "name": "Target - Jetson TX1", 
                "division": "tx1_64bit", 
                "description": ""
            }, 
            {
                "category": "jetson_64_tx1", 
                "alias": "l4t_64_tx1", 
                "name": "Linux for Tegra Host Side Image Setup", 
                "division": "tx1_64bit", 
                "description": "Sample root file system derived from Ubuntu 14.04 and NVIDIA Tegra Linux Driver Package to support development for Jetson TX1 Developer Kit"
            }, 
            {
                "category": "jetson_64_tx1", 
                "alias": "postinstall_tx1_64bit", 
                "name": "Install on Target", 
                "division": "tx1_64bit", 
                "description": "Post installation, including flashing OS to device and installing components on device."
            }, 
            {
                "category": "postinstall_tx2_64bit", 
                "alias": "vt_tx2_64", 
                "name": "VisionWorks Pack", 
                "division": "tx2_64bit tx2i_64bit", 
                "description": "NVIDIA VisionWorks Toolkit for Jetson TX2/TX2i Developer Kit"
            }, 
            {
                "category": "postinstall_tx1_64bit", 
                "alias": "vt_tx1_64", 
                "name": "VisionWorks Pack", 
                "division": "tx1_64bit", 
                "description": "NVIDIA VisionWorks Toolkit for Jetson TX1 Developer Kit"
            }, 
            {
                "alias": "jetson_64_tx2", 
                "name": "Target - Jetson TX2/TX2i", 
                "division": "tx2_64bit tx2i_64bit", 
                "description": ""
            }, 
            {
                "category": "jetson_64_tx2", 
                "alias": "l4t_64_tx2", 
                "name": "Linux for Tegra Host Side Image Setup", 
                "division": "tx2_64bit tx2i_64bit", 
                "description": "Sample root file system derived from Ubuntu 14.04 and NVIDIA Tegra Linux Driver Package to support development for Jetson TX2/TX2i Developer Kit"
            }, 
            {
                "category": "jetson_64_tx2", 
                "alias": "postinstall_tx2_64bit", 
                "name": "Install on Target", 
                "division": "tx2_64bit tx2i_64bit", 
                "description": "Post installation, including flashing OS to device and installing components on device."
            }
        ], 
        "component": [
            {
                "category": "common", 
                "reside": "host", 
                "name": "Tegra Graphics Debugger", 
                "license": "license_internal", 
                "space": [
                    {
                        "size": "181399142"
                    }
                ], 
                "description": "Tegra Graphics Debugger is a console-grade tool that allows developers to debug and profile OpenGL/GL ES applications.", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.nvtools.battle", 
                "version": "2.5", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/2017_11_28_1956-23204764-NVIDIA_Tegra_Graphics_Debugger_2.5.17332.1956_Release_External_tgd-l4t_linux-l4t.run", 
                        "checksum": "d2184d6d6c4206544ff28aa4db64fc8a", 
                        "cache": "./2017_11_28_1956-23204764-NVIDIA_Tegra_Graphics_Debugger_2.5.17332.1956_Release_External_tgd-l4t_linux-l4t.run", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/tegra-graphics-debugger.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "common", 
                "reside": "host", 
                "name": "NVIDIA System Profiler", 
                "license": "license_internal", 
                "space": [
                    {
                        "size": "194864947"
                    }
                ], 
                "description": "NVIDIA System Profiler is a tool for diagnosing and optimizing applications' CPU and GPU workload and interactions.", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.nvtools.quadd", 
                "version": "4.0", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/SystemProfiler-linux-public-4.0.387-eecbec0.tar.gz", 
                        "checksum": "cea7ab031e5d762ff98aa29545c7b01a", 
                        "cache": "./SystemProfiler-linux-public-4.0.387-eecbec0.tar.gz", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/tegra-profiler.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "common", 
                "reside": "host", 
                "name": "JetPack Documentation", 
                "license": "license_internal", 
                "space": [
                    {
                        "size": "21000806"
                    }
                ], 
                "description": "Documentation including user guides, best practices, and manuals for various tools and middleware.", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.docs.jetpack", 
                "version": "3.2", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/jetpack_docs.zip", 
                        "checksum": "089a159b95988dc6cb84de85645ff50c", 
                        "cache": "./jetpack_docs.zip", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/NVIDIA_Tegra_Linux_Driver_Package.tar", 
                        "checksum": "e27f9307c7eecd04c305ae7b799b5800", 
                        "cache": "./NVIDIA_Tegra_Linux_Driver_Package.tar", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/Tegra_Linux_Driver_Package_Release_Notes_R28.2.1.pdf", 
                        "checksum": "28f8c81bb66b0f8bf97c83076b255da2", 
                        "cache": "./Tegra_Linux_Driver_Package_Release_Notes_R28.2.1.pdf", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "common", 
                "reside": "host", 
                "name": "DevTools Documentation", 
                "license": "license_internal", 
                "space": [
                    {
                        "size": "1000806"
                    }
                ], 
                "description": "Documentation for Tegra Graphics Debugger and NVIDIA System Profiler", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.docs.devtools", 
                "version": "3.2", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/devtools_docs.zip", 
                        "checksum": "496c7379255253c1fcaf80f047aeda9e", 
                        "cache": "./devtools_docs.zip", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "common", 
                "reside": "host", 
                "division": "tx1_64bit tx2_64bit tx2i_64bit", 
                "name": "OpenCV", 
                "license": "license_ocv", 
                "platform": "16.04", 
                "description": "OpenCV for Host 16.04", 
                "express": "1", 
                "space": [
                    {
                        "size": "3412306330"
                    }
                ], 
                "reuse": "1", 
                "alias": "org.opencv.middleware.ocv_64_host", 
                "version": "3.3.1", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.host'=='9.0'", 
                "download": [
                    {
                        "url": "/libopencv_3.3.1_amd64.deb", 
                        "checksum": "2c06efa9272327a18f7e6b92af82ce52", 
                        "cache": "./libopencv_3.3.1_amd64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libopencv-dev_3.3.1_amd64.deb", 
                        "checksum": "00e624174cf1065897cc208c8e71d4f9", 
                        "cache": "./libopencv-dev_3.3.1_amd64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libopencv-python_3.3.1_amd64.deb", 
                        "checksum": "a7653cba47413e5cab72688754fb4f2b", 
                        "cache": "./libopencv-python_3.3.1_amd64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libopencv-samples_3.3.1_amd64.deb", 
                        "checksum": "2ead46d3c1a83b78c2b51b02a6cc223b", 
                        "cache": "./libopencv-samples_3.3.1_amd64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/opencv.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "visionworksgroup_host", 
                "reside": "host", 
                "name": "VisionWorks", 
                "license": "license_visionworks", 
                "space": [
                    {
                        "size": "233253888"
                    }
                ], 
                "description": "NVIDIA VisionWorks Toolkit is a CUDA accelerated software development package for computer vision (CV) and image processing. The toolkit is designed to provide a production-quality, standard interface that implements and extends the OpenVX standard, enabling high performance computer vision applications on NVIDIA Tegra devices.", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.middleware.visionworks_host", 
                "version": "1.6", 
                "platform": "16.04", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.host'>='9.0' and 'org.opencv.middleware.ocv_64_host'=='3.3.1'", 
                "download": [
                    {
                        "url": "/libvisionworks-repo_1.6.0.500n_amd64.deb", 
                        "checksum": "f9b584d298d07da030b9c6a29365e8c2", 
                        "cache": "./libvisionworks-repo_1.6.0.500n_amd64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "visionworksgroup_host", 
                "reside": "host", 
                "name": "VisionWorks Plus (SFM)", 
                "license": "license_visionworks", 
                "space": [
                    {
                        "size": "64567910"
                    }
                ], 
                "description": "NVIDA VisionWorks SFM is a library of computer vision primitives and algorithms with framework optimized for NVIDIA platforms built on top of NVIDIA(R) VisionWorks and extends its API. VisionWorks SFM implements the simplest example of the Structure from Motion (SFM) pipeline. SFM itself refers to the process of estimating three-dimensional structures from two-dimensional image sequences which may be coupled with local motion signals.", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.middleware.visionworks_host_sfm", 
                "version": "0.90", 
                "platform": "16.04", 
                "location": "", 
                "dependencies": "'com.nvidia.middleware.visionworks_host' == '1.6'", 
                "download": [
                    {
                        "url": "/libvisionworks-sfm-repo_0.90.3_amd64.deb", 
                        "checksum": "1354fc439c573c6135dd21b5d4715c58", 
                        "cache": "./libvisionworks-sfm-repo_0.90.3_amd64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "visionworksgroup_host", 
                "reside": "host", 
                "name": "VisionWorks Object Tracker", 
                "license": "license_visionworks", 
                "space": [
                    {
                        "size": "18725683"
                    }
                ], 
                "description": "NVIDA VisionWorks Object Tracker is a library of computer vision primitives with framework optimized for NVIDIA platforms built on top of NVIDIA(R) VisionWorks(TM) and extends its API. VisionWorks Object Tracker provides general Object Tracker interface and implements Keypoint Object Tracker algorithm.", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.middleware.visionworks_host_tracking", 
                "version": "0.88", 
                "platform": "16.04", 
                "location": "", 
                "dependencies": "'com.nvidia.middleware.visionworks_host' == '1.6'", 
                "download": [
                    {
                        "url": "/libvisionworks-tracking-repo_0.88.2_amd64.deb", 
                        "checksum": "8c1bb675422c24280e879bba29eba103", 
                        "cache": "./libvisionworks-tracking-repo_0.88.2_amd64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "visionworksgroup_host", 
                "reside": "host", 
                "name": "VisionWorks References", 
                "license": "license_visionworks", 
                "space": [
                    {
                        "size": "7160832"
                    }
                ], 
                "description": "NVIDIA VisionWorks Toolkit Documentation", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.middleware.visionworks_docs", 
                "version": "1.6", 
                "platform": "16.04", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/NVIDIA_VisionWorks_References.zip", 
                        "checksum": "fe8197f368f7edf37cd3f35b75a92584", 
                        "cache": "./NVIDIA_VisionWorks_References.zip", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "common", 
                "reside": "host", 
                "division": "tx1_64bit tx2_64bit tx2i_64bit", 
                "name": "CUDA Toolkit", 
                "license": "license_cuda", 
                "platform": "16.04", 
                "description": "CUDA Toolkit on Host for Cross Compiling. the NVIDIA CUDA Toolkit provides a comprehensive development environment for C and C++ developers building GPU-accelerated applications.", 
                "express": "1", 
                "space": [
                    {
                        "size": "3412306330"
                    }
                ], 
                "reuse": "1", 
                "alias": "com.nvidia.cuda.host", 
                "version": "9.0", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/cuda-repo-ubuntu1604-9-0-local_9.0.252-1_amd64.deb", 
                        "checksum": "d30e035f20d83b377147dbf8d795e5b7", 
                        "cache": "./cuda-repo-ubuntu1604-9-0-local_9.0.252-1_amd64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/cuda.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "common", 
                "reside": "host", 
                "division": "tx1_64bit tx2_64bit tx2i_64bit", 
                "name": "CUDA Toolkit", 
                "license": "license_cuda", 
                "platform": "14.04", 
                "description": "CUDA Toolkit on Host for Cross Compiling. the NVIDIA CUDA Toolkit provides a comprehensive development environment for C and C++ developers building GPU-accelerated applications.", 
                "express": "1", 
                "space": [
                    {
                        "size": "3412306330"
                    }
                ], 
                "reuse": "1", 
                "alias": "com.nvidia.cuda.host", 
                "version": "9.0", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/cuda-repo-ubuntu1404-9-0-local_9.0.252-1_amd64.deb", 
                        "checksum": "f0c326f089cd130bc47920dcdd450799", 
                        "cache": "./cuda-repo-ubuntu1404-9-0-local_9.0.252-1_amd64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/cuda.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "l4t_64_tx1", 
                "division": "tx1_64bit", 
                "reside": "host", 
                "name": "File System and OS", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "2582806732"
                    }
                ], 
                "description": "File System (64-Bit)", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.l4t.filesystem_64_tx1", 
                "version": "28.2", 
                "location": "", 
                "dependencies": "'com.nvidia.l4t.driver4os_64_tx1'=='24.2.1'", 
                "download": [
                    {
                        "url": "/Tegra_Linux_Sample-Root-Filesystem_R28.2.1_aarch64.tbz2", 
                        "checksum": "93b204a44c5516edf4a098714d007a5c", 
                        "cache": "./Tegra_Linux_Sample-Root-Filesystem_R28.2.1_aarch64.tbz2", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "0"
            }, 
            {
                "category": "l4t_64_tx1", 
                "division": "tx1_64bit", 
                "reside": "host", 
                "name": "Drivers", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "137922355"
                    }
                ], 
                "description": "NVIDIA Tegra Linux Driver Package (64-Bit)", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.l4t.driver4os_64_tx1", 
                "version": "28.2", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/Tegra210_Linux_R28.2.0_aarch64.tbz2", 
                        "checksum": "0b78d620654add9f9147f99b1bb317de", 
                        "cache": "./Tegra210_Linux_R28.2.0_aarch64.tbz2", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "jetson_64_tx1", 
                "division": "tx1_64bit", 
                "reside": "tx1_64bit", 
                "name": "Flash OS Image to Target", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "5054295859"
                    }
                ], 
                "description": "Flash 64 Bit OS to device", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.l4t.flash_64_tx1", 
                "version": "28.2", 
                "location": "", 
                "dependencies": "'com.nvidia.l4t.driver4os_64_tx1'>='24.0' and 'com.nvidia.l4t.filesystem_64_tx1'>='24.0'", 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "postinstall_tx1_64bit", 
                "division": "tx1_64bit", 
                "reside": "host", 
                "name": "CUDA Toolkit", 
                "license": "license_cuda", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "CUDA Toolkit for L4T. The NVIDIA CUDA Toolkit provides a comprehensive development environment for C and C++ developers building GPU-accelerated applications.", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.cuda.l4t_64_tx1", 
                "version": "9.0", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/cuda-repo-l4t-9-0-local_9.0.252-1_arm64.deb", 
                        "checksum": "9385806be429ad92d49f0750f748e2b2", 
                        "cache": "./cuda-repo-l4t-9-0-local_9.0.252-1_arm64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/cuda.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "postinstall_tx1_64bit", 
                "division": "tx1_64bit", 
                "reside": "tx1_64bit", 
                "name": "Compile CUDA Samples", 
                "license": "license_cuda", 
                "space": [
                    {
                        "size": "606954701"
                    }
                ], 
                "description": "Compile CUDA Samples", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.cuda.compile_cuda_64_tx1", 
                "version": "9.0", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.l4t_64_tx1'>='9.0' and 'com.nvidia.cuda.host'>='9.0'", 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/cuda.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "postinstall_tx1_64bit", 
                "division": "tx1_64bit", 
                "reside": "tx1_64bit", 
                "name": "cuDNN Package", 
                "license": "license_nv", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "The NVIDIA CUDA Deep Neural Network library (cuDNN) is a GPU-accelerated library of primitives for deep neural networks. Deep learning developers and researchers worldwide rely on the highly optimized routines in cuDNN which allow them to focus on designing and training neural network models rather than spending time on low-level performance tuning.", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.middleware.cudnn_64_tx1", 
                "version": "7.0", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.l4t_64_tx1'=='9.0'", 
                "download": [
                    {
                        "url": "/libcudnn7_7.0.5.15-1+cuda9.0_arm64.deb", 
                        "checksum": "6aab1d77a60871a0ee0d306aee6f9a0e", 
                        "cache": "./libcudnn7_7.0.5.15-1+cuda9.0_arm64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libcudnn7-dev_7.0.5.15-1+cuda9.0_arm64.deb", 
                        "checksum": "919feccd183273cd3b1efab1aca39f04", 
                        "cache": "./libcudnn7-dev_7.0.5.15-1+cuda9.0_arm64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libcudnn7-doc_7.0.5.15-1+cuda9.0_arm64.deb", 
                        "checksum": "f1746b71c84cf42542a42e9b52e03f2d", 
                        "cache": "./libcudnn7-doc_7.0.5.15-1+cuda9.0_arm64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "postinstall_tx1_64bit", 
                "division": "tx1_64bit", 
                "reside": "tx1_64bit", 
                "name": "TensorRT", 
                "license": "license_gie", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "NVIDIA TensorRT\u2122 is a high performance neural network inference engine for production deployment of deep learning applications. TensorRT can be used to rapidly optimize, validate and deploy trained neural network for inference to hyperscale data centers, embedded, or automotive product platforms.", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.gie.l4t_64_tx1", 
                "version": "3.0", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.l4t_64_tx1'>='9.0' and 'com.nvidia.middleware.cudnn_64_tx1'=='7.0'", 
                "download": [
                    {
                        "url": "/nv-tensorrt-repo-ubuntu1604-ga-cuda9.0-trt3.0.4-20180208_1-1_arm64.deb", 
                        "checksum": "5d625140e1a7c92706191b1f4cbc4cce", 
                        "cache": "./nv-tensorrt-repo-ubuntu1604-ga-cuda9.0-trt3.0.4-20180208_1-1_arm64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "postinstall_tx1_64bit", 
                "division": "tx1_64bit", 
                "reside": "tx1_64bit", 
                "name": "Multimedia API package", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "Multimedia API offers developers to use the lower level APIs for flexible application development. This is in addition to the Gstreamer framework available from the previous releases.\nThe APIs include Argus API for imaging applications and V4L2 for encode, decode, scaling and other media functions. These lower level APIs provide a better control over the underlying hardware blocks. ", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.mmapi.l4t_64_tx1", 
                "version": "28.2", 
                "location": "", 
                "dependencies": "'com.nvidia.gie.l4t_64_tx1'>='3.0' and 'org.opencv.middleware.ocv_l4t_64_tx1'>='2.4'", 
                "download": [
                    {
                        "url": "/Tegra_Multimedia_API_R28.2.1_aarch64.tbz2", 
                        "checksum": "ffc676c2bfae02c74c7520e7b72ab757", 
                        "cache": "./Tegra_Multimedia_API_R28.2.1_aarch64.tbz2", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "postinstall_tx1_64bit", 
                "division": "tx1_64bit", 
                "reside": "tx1", 
                "name": "OpenCV", 
                "license": "license_ocv", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "OpenCV for Jetson TX1 Developer Kit.", 
                "express": "1", 
                "reuse": "0", 
                "alias": "org.opencv.middleware.ocv_l4t_64_tx1", 
                "version": "3.3.1", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.l4t_64_tx1'>='9.0'", 
                "download": [
                    {
                        "url": "/libopencv_3.3.1_t210_arm64.deb", 
                        "checksum": "919aba869414845a1e36ba5e6d414ce4", 
                        "cache": "./libopencv_3.3.1_t210_arm64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libopencv-dev_3.3.1_t210_arm64.deb", 
                        "checksum": "1cd568ace874939c2e83a42c3de0ae40", 
                        "cache": "./libopencv-dev_3.3.1_t210_arm64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libopencv-python_3.3.1_t210_arm64.deb", 
                        "checksum": "7ad4ca02aaddcdd7c1ce3105d32bbd24", 
                        "cache": "./libopencv-python_3.3.1_t210_arm64.deb", 
                        "checksumType": "md5"
                    }, 
                    {
                        "url": "/libopencv-samples_3.3.1_t210_arm64.deb", 
                        "checksum": "0165fad7b53f08fc5db735e7027c648f", 
                        "cache": "./libopencv-samples_3.3.1_t210_arm64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/opencv.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "vt_tx1_64", 
                "division": "tx1_64bit", 
                "reside": "tx1", 
                "name": "VisionWorks", 
                "license": "license_visionworks", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "NVIDIA VisionWorks Toolkit is a CUDA accelerated software development package for computer vision (CV) and image processing. The toolkit is designed to provide a production-quality, standard interface that implements and extends the OpenVX standard, enabling high performance computer vision applications on NVIDIA Tegra devices.", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.middleware.visionworks_target_64_tx1", 
                "version": "1.6", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.l4t_64_tx1'=='9.0' and 'org.opencv.middleware.ocv_l4t_64_tx1'=='3.3' ", 
                "download": [
                    {
                        "url": "/libvisionworks-repo_1.6.0.500n_arm64.deb", 
                        "checksum": "066cb90ca605f805aa616c5be6b06926", 
                        "cache": "./libvisionworks-repo_1.6.0.500n_arm64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "l4t_64_tx2", 
                "division": "tx2_64bit tx2i_64bit", 
                "reside": "host", 
                "name": "File System and OS", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "File System (64-Bit)", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.l4t.filesystem_64_tx2", 
                "version": "28.2.1", 
                "location": "", 
                "dependencies": "'com.nvidia.l4t.driver4os_64_tx2'=='28.2'", 
                "download": [
                    {
                        "url": "/Tegra_Linux_Sample-Root-Filesystem_R28.2.1_aarch64.tbz2", 
                        "checksum": "93b204a44c5516edf4a098714d007a5c", 
                        "cache": "./Tegra_Linux_Sample-Root-Filesystem_R28.2.1_aarch64.tbz2", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "0"
            }, 
            {
                "category": "l4t_64_tx2", 
                "division": "tx2_64bit tx2i_64bit", 
                "reside": "host", 
                "name": "Drivers", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "NVIDIA Tegra Linux Driver Package (64-Bit)", 
                "express": "1", 
                "reuse": "1", 
                "alias": "com.nvidia.l4t.driver4os_64_tx2", 
                "version": "28.2.1", 
                "location": "", 
                "dependencies": "", 
                "download": [
                    {
                        "url": "/Tegra186_Linux_R28.2.1_aarch64.tbz2", 
                        "checksum": "0a00f9b4c6c7fcd73d301082019d09ea", 
                        "cache": "./Tegra186_Linux_R28.2.1_aarch64.tbz2", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "jetson_64_tx2", 
                "division": "tx2_64bit", 
                "reside": "tx2_64bit", 
                "name": "Flash OS Image to Target", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "Flash 64 Bit OS to device", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.l4t.flash_64_tx2", 
                "version": "28.2.1", 
                "location": "", 
                "dependencies": "'com.nvidia.l4t.driver4os_64_tx2'>='28.2' and 'com.nvidia.l4t.filesystem_64_tx2'>='28.2'", 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "jetson_64_tx2", 
                "division": "tx2i_64bit", 
                "reside": "tx2_64bit", 
                "name": "Flash OS Image to Target", 
                "license": "license_bsp", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "Flash 64 Bit OS to device", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.l4t.flash_64_tx2i", 
                "version": "28.2.1", 
                "location": "", 
                "dependencies": "'com.nvidia.l4t.driver4os_64_tx2'>='28.2' and 'com.nvidia.l4t.filesystem_64_tx2'>='28.2'", 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "vt_tx2_64", 
                "division": "tx2_64bit tx2i_64bit", 
                "reside": "tx2", 
                "name": "VisionWorks", 
                "license": "license_visionworks", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
                "description": "NVIDIA VisionWorks Toolkit is a CUDA accelerated software development package for computer vision (CV) and image processing. The toolkit is designed to provide a production-quality, standard interface that implements and extends the OpenVX standard, enabling high performance computer vision applications on NVIDIA Tegra devices.", 
                "express": "1", 
                "reuse": "0", 
                "alias": "com.nvidia.middleware.visionworks_target_64_tx2", 
                "version": "1.6", 
                "location": "", 
                "dependencies": "'com.nvidia.cuda.l4t_64_tx2'=='9.0' and 'org.opencv.middleware.ocv_l4t_64_tx2'=='3.3'", 
                "download": [
                    {
                        "url": "/libvisionworks-repo_1.6.0.500n_arm64.deb", 
                        "checksum": "066cb90ca605f805aa616c5be6b06926", 
                        "cache": "./libvisionworks-repo_1.6.0.500n_arm64.deb", 
                        "checksumType": "md5"
                    }
                ], 
                "icon": "https://developer.download.nvidia.com/devzone/devcenter/mobile/jetpack_l4t/icons/nvidia.png", 
                "autoDetect": "0", 
                "interactive": "1"
            }, 
            {
                "category": "vt_tx1_64", 
                "division": "tx1_64bit", 
                "reside": "tx1", 
                "name": "VisionWorks Plus (SFM)", 
                "license": "license_visionworks", 
                "space": [
                    {
                        "size": "0"
                    }
                ], 
            
        ]
    }
}
6

Why all the posts from users got deleted? Or I just don’t see it. There was a good question about tensorRT and Python.

BTW, good job Dustin :)

7

Are these the posts you were looking for?

[url]https://devtalk.nvidia.com/default/topic/1037813/announcements/jetpack-3-3-mdash-l4t-r28-2-1-release-for-jetson-tx1-tx2/[/url]

I think this forum software might not have support for linking threads, so there’re multiple duplicated announcement threads are being posted in different sub-forums.

-albertr

8

Yes Exactly. Thank you very much! very appreciate.

9

My teammate have downloaded all the packages from American server, the TensorRT version is 4.0.
So I guess the server in China hasn’t been updated yet.

10

What happend with Russian server? I woudn’t upload new and old Jetpack versions on my dev boards for 2 weeks… always see problem with transfering of CUDA OpenCV and TensorRT pockets!
Massage from terminal:

dpkg-query: package 'cuda-toolkit-9-0' is not installed and no information is available
dpkg-query: package 'libfreeimage-dev' is not installed and no information is available
dpkg-query: package 'libopenmpi-dev' is not installed and no information is available
dpkg-query: package 'openmpi-bin' is not installed and no information is available
Use dpkg --info (= dpkg-deb --info) to examine archive files,
and dpkg --contents (= dpkg-deb --contents) to list their contents.
1
Error: CUDA cannot be installed on device. This may be caused by other apt-get command running on device when installing CUDA. Please use apt-get command in a terminal to make sure following packages are installed correctly on device before continuing:
cuda-toolkit-9-0 libgomp1 libfreeimage-dev libopenmpi-dev openmpi-bin
After these packages are installed on device, press Enter key to continue

I neet install 40 pockets manual and after Jetpack ended the uploading i don’t fimd much of librares. What happend?

11

It doesn’t look like your manifest contains references to JetPack 3.3. We will check if the geo cache needs flushed in your region.

12

Does the JetPack 3.3 have Nvidia’s part number

13

The NVIDIA part numbers for the Jetsons? You can find those here:

https://developer.nvidia.com/embedded/faq#jetson-part-numbers

14

Hi there is no part number for the JETPACK 3.3 in that link

15

Software releases do not get part numbers.

16

There was a CDN caching issue on the mainland China servers which has been resolved now. Please try again and you should be able to access Jetpack 3.3 and it’s contents.

17

Appreciate for you guys output, it seems like tensorflow for jetpack 4.4/tensorrt 4.0 has not released, when this version will come?

18

Hi adonishong, we are working on the TF 1.9 pip whl for JetPack 3.3, we will update this thread with it when it’s ready.

19

Looking forward to it ^_^

Met a new issue last night, we are converting a network(which has PRELU layer included, abs opt is required for PRELU) from uff with giexec, this issues stop us:

UFFParser: Parser error: resnet_v1_50/prelu0/Abs: Unary not supported for other non-constant node

Also tried a script (scripts/convert_plan.py) from official tensorflow to tensorrt python project:

Still got similar error:

UFFParser: Parser error: resnet_v1_50/prelu0/Abs: Unary not supported for other non-constant node

From tensorrt 4.0 document, abs operation is support for TensorFlow, my question is with this coming TF 1.9 pip whl for JetPack 3.3, will this issue be solved? Seems like this issue comes from “/usr/lib/aarch64-linux-gnu/libnvparsers.so.4.1.3”

20

OK, please post this issue as a new topic on the TX2 forum so it can be referred to properly. Thanks.