WebCal
Ctrl K

가이드

Vulkan

가이드통합FAQ지원
가격블로그
빠른 시작Top-Up and BillingAccelerating Access to Academic ResourcesQuick StartIntroductionMaintenance and TroubleshootingNetwork
플랫폼JetBrain ProjectorTmpWebCal Scholars Program @2026Introduction to the Public Beta/Suqian Zone AAbout UsCopying Data Between InstancesAnalysis of Server Performance MetricsTidal Computing PowerLoad Balancing
중국산 칩Using Huawei MindIEHuawei Ascend NPUMooreThread GPU
환경 설정CUDA/cuDNNMinicondaPython3.XInstalling DependenciesOverviewImages
엔터프라이즈 기능Flexible DeploymentElastic Deployment Release NotesBest Practices for Elastic DeploymentPerformance Metrics Monitoring
컨테이너 인스턴스JupyterLabRemote SSH ConnectionSave the imageScaling ConfigurationMulti-machine, multi-GPU parallel processingDaemonOverviewChange the billing methodMigration Example (Same Region)Migration ExamplesRemote DesktopReset the system
GPU 선택GPU SelectionPerformance Testing
데이터Upload DataDownload DataPublic DataPublic Cloud Storage (Highly Recommended)Compression / DecompressionFile StorageLocal data diskOverview
모범 사례FileZillaGitGromacsHuggingFaceKataGoLinux BasicsMPIOpenCLPyCharm Remote DevelopmentR (RStudio) InstallationSSH TunnelTensorBoardRemote Development with VSCodeVisdomVulkanXShellOpen PortsWeChat MessagesPerformanceExpose multiple servicesMoney-Saving TipsComputation Precision IssuesSoftware Sources

모범 사례

Vulkan

2026. 07. 30.82833회 조회

If you cannot find a GPU device when using Vulkan (or other software that relies on Vulkan, such as Omnigibson), follow these steps to resolve the issue:

Step 1: Write to the ICD file

mkdir -p /etc/vulkan/icd.d
cat >> /etc/vulkan/icd.d/my_nvidia_icd.json <<EOF
{
    "file_format_version" : "1.0.0",
    "ICD": {
        "library_path": "/lib/x86_64-linux-gnu/libEGL_nvidia.so.0",
        "api_version" : "1.3.277"
    }
}
EOF

The library specified in library\_path must be libEGL_nvidia and not libGLX_nvidia, because the server does not have a desktop; the former is required when using headless mode. For details, see the documentation.

Step 2: Verification

Install required dependencies

apt update && apt install -y vulkan-tools libvulkan1 libsm6 libegl1

Finally, run vulkaninfo --summary to verify:

$ export VK_ICD_FILENAMES=/etc/vulkan/icd.d/my_nvidia_icd.json     # 显式指定ICD文件路径
$ vulkaninfo --summary
...
Devices:
========
GPU0:
        apiVersion         = 4206869 (1.3.277)
        driverVersion      = 2308620416 (0x899ac080)
        vendorID           = 0x10de
        deviceID           = 0x2684
        deviceType         = PHYSICAL_DEVICE_TYPE_DISCRETE_GPU  <--- 这里设备类型是GPU,说明正确。如果没有识别到GPU则是: PHYSICAL_DEVICE_TYPE_CPU
        deviceName         = NVIDIA GeForce RTX 4090            <--- 这里可以看出正确识别了GPU型号
        driverID           = DRIVER_ID_NVIDIA_PROPRIETARY
        driverName         = NVIDIA
        driverInfo         = 550.107.02
        conformanceVersion = 1.3.7.2
        deviceUUID         = 75219daa-2fe9-cefe-4994-1a598657b01f
        driverUUID         = 12adfef6-a92a-528b-8610-45df7fa0a5b8
...
다음 글XShell
Compute Rental 문서가이드(으)로 돌아가기