您现在访问的是微软AZURE全球版技术文档网站,若需要访问由世纪互联运营的MICROSOFT AZURE中国区技术文档网站,请访问 https://docs.azure.cn.

GPU 优化虚拟机大小GPU optimized virtual machine sizes

GPU 优化 VM 大小是具有单个或多个 NVIDIA GPU 的专用虚拟机。GPU optimized VM sizes are specialized virtual machines available with single or multiple NVIDIA GPUs. 这些大小是针对计算密集型、图形密集型和可视化工作负荷设计的。These sizes are designed for compute-intensive, graphics-intensive, and visualization workloads. 本文介绍了此分组中各个大小的 GPU、vCPU、数据磁盘和 NIC 的数目及类型以及存储吞吐量及网络带宽的相关信息。This article provides information about the number and type of GPUs, vCPUs, data disks, and NICs as well as storage throughput and network bandwidth for each size in this grouping.

  • NC、NCv2、NCv3 和 ND 大小针对计算密集型和网络密集型应用程序和算法进行了优化,包括基于 CUDA 和 OpenCL 的应用程序和模拟、AI 以及深度学习。NC, NCv2, NCv3, and ND sizes are optimized for compute-intensive and network-intensive applications and algorithms, including CUDA- and OpenCL-based applications and simulations, AI, and Deep Learning.
  • NV 大小已针对远程可视化效果、流式处理、游戏、编码和 VDI 方案进行了优化和设计,使用 OpenGL 和 DirectX 之类的框架。NV sizes are optimized and designed for remote visualization, streaming, gaming, encoding, and VDI scenarios utilizing frameworks such as OpenGL and DirectX.

NC 系列NC-series

NC 系列 VM 采用 NVIDIA Tesla K80 卡。NC-series VMs are powered by the NVIDIA Tesla K80 card. 通过将 CUDA 用于能源勘探应用、碰撞模拟、光纤跟踪渲染、深度学习等领域,用户可以更快地分析数据。Users can crunch through data faster by leveraging CUDA for energy exploration applications, crash simulations, ray traced rendering, deep learning and more. NC24r 配置提供了针对紧密耦合的并行计算工作负荷优化的低延迟、高吞吐量网络接口。The NC24r configuration provides a low latency, high-throughput network interface optimized for tightly coupled parallel computing workloads.

大小Size vCPUvCPU 内存:GiBMemory: GiB 临时存储 (SSD) GiBTemp storage (SSD) GiB GPUGPU 最大数据磁盘数Max data disks 最大 NIC 数Max NICs
Standard_NC6Standard_NC6 66 5656 340340 11 2424 11
Standard_NC12Standard_NC12 1212 112112 680680 22 4848 22
Standard_NC24Standard_NC24 2424 224224 14401440 44 6464 44
Standard_NC24r*Standard_NC24r* 2424 224224 14401440 44 6464 44

1 GPU = 半块 K80 卡。1 GPU = one-half K80 card.

*支持 RDMA*RDMA capable

NCv2 系列NCv2-series

NCv2 系列 VM 采用 NVIDIA Tesla P100 GPU。NCv2-series VMs are powered by NVIDIA Tesla P100 GPUs. 这些 GPU 可提供比 NC 系列高 2 倍以上的计算性能。These GPUs can provide more than 2x the computational performance of the NC-series. 客户可将这些更新的 GPU 用于传统的 HPC 工作负荷,例如油藏模拟、DNA 测序、蛋白质分析、Monte Carlo 模拟和其他工作负荷。Customers can take advantage of these updated GPUs for traditional HPC workloads such as reservoir modeling, DNA sequencing, protein analysis, Monte Carlo simulations, and others. NC24rs v2 配置提供了针对紧密耦合的并行计算工作负荷优化的低延迟、高吞吐量网络接口。The NC24rs v2 configuration provides a low latency, high-throughput network interface optimized for tightly coupled parallel computing workloads.

重要

对于此大小系列,订阅中的 vCPU(核心)配额在每个区域中最初都设置为 0。For this size family, the vCPU (core) quota in your subscription is initially set to 0 in each region. 可以请求在某个可用区域提高此系列的 vCPU 配额Request a vCPU quota increase for this family in an available region.

大小Size vCPUvCPU 内存:GiBMemory: GiB 临时存储 (SSD) GiBTemp storage (SSD) GiB GPUGPU 最大数据磁盘数Max data disks 最大 NIC 数Max NICs
Standard_NC6s_v2Standard_NC6s_v2 66 112112 736736 11 1212 44
Standard_NC12s_v2Standard_NC12s_v2 1212 224224 14741474 22 2424 88
Standard_NC24s_v2Standard_NC24s_v2 2424 448448 29482948 44 3232 88
Standard_NC24rs_v2*Standard_NC24rs_v2* 2424 448448 29482948 44 3232 88

1 GPU = 一个 P100 卡。1 GPU = one P100 card.

*支持 RDMA*RDMA capable

NCv3 系列NCv3-series

NCv3 系列 VM 采用 NVIDIA Tesla V100 GPU。NCv3-series VMs are powered by NVIDIA Tesla V100 GPUs. 这些 GPU 可提供 NCv2 系列的 1.5 倍计算性能。These GPUs can provide 1.5x the computational performance of the NCv2-series. 客户可将这些更新的 GPU 用于传统的 HPC 工作负荷,例如油藏模拟、DNA 测序、蛋白质分析、Monte Carlo 模拟和其他工作负荷。Customers can take advantage of these updated GPUs for traditional HPC workloads such as reservoir modeling, DNA sequencing, protein analysis, Monte Carlo simulations, and others. NC24rs v3 配置提供了针对紧密耦合的并行计算工作负荷优化的低延迟、高吞吐量网络接口。The NC24rs v3 configuration provides a low latency, high-throughput network interface optimized for tightly coupled parallel computing workloads.

重要

对于此大小系列,订阅中的 vCPU(核心)配额在每个区域中最初都设置为 0。For this size family, the vCPU (core) quota in your subscription is initially set to 0 in each region. 可以请求在某个可用区域提高此系列的 vCPU 配额Request a vCPU quota increase for this family in an available region.

大小Size vCPUvCPU 内存:GiBMemory: GiB 临时存储 (SSD) GiBTemp storage (SSD) GiB GPUGPU 最大数据磁盘数Max data disks 最大 NIC 数Max NICs
Standard_NC6s_v3Standard_NC6s_v3 66 112112 736736 11 1212 44
Standard_NC12s_v3Standard_NC12s_v3 1212 224224 14741474 22 2424 88
Standard_NC24s_v3Standard_NC24s_v3 2424 448448 29482948 44 3232 88
Standard_NC24rs_v3*Standard_NC24rs_v3* 2424 448448 29482948 44 3232 88

1 GPU = 一个 V100 卡。1 GPU = one V100 card.

*支持 RDMA*RDMA capable

ND 系列ND-series

ND 系列虚拟机是针对 AI 和深度学习工作负荷设计的 GPU 家族的新成员。The ND-series virtual machines are a new addition to the GPU family designed for AI and Deep Learning workloads. 它们在训练和推理方面性能卓越。They offer excellent performance for training and inference. ND 实例采用 NVIDIA Tesla P40 GPU。ND instances are powered by NVIDIA Tesla P40 GPUs. 这些实例可以针对单精度浮点运算和利用 Microsoft 认知工具包、TensorFlow、Caffe 及其他框架的 AI 工作负荷提供卓越的性能。These instances provide excellent performance for single-precision floating point operations, for AI workloads utilizing Microsoft Cognitive Toolkit, TensorFlow, Caffe, and other frameworks. ND 系列还提供了更大的 GPU 内存大小(24 GB),能够适应更大的神经网络模型。The ND-series also offers a much larger GPU memory size (24 GB), enabling to fit much larger neural net models. 与 NC 系列一样,ND 系列可通过 RDMA 和 InfiniBand 连接提供含辅助型低延迟、高吞吐量网络的配置,以便可运行跨多个 GPU 的大规模训练作业。Like the NC-series, the ND-series offers a configuration with a secondary low-latency, high-throughput network through RDMA, and InfiniBand connectivity so you can run large-scale training jobs spanning many GPUs.

重要

对于此大小系列,订阅中每个区域的 vCPU(核心)配额最初都设置为 0。For this size family, the vCPU (core) quota per region in your subscription is initially set to 0. 可以请求在某个可用区域提高此系列的 vCPU 配额Request a vCPU quota increase for this family in an available region.

大小Size vCPUvCPU 内存:GiBMemory: GiB 临时存储 (SSD) GiBTemp storage (SSD) GiB GPUGPU 最大数据磁盘数Max data disks 最大 NIC 数Max NICs
Standard_ND6sStandard_ND6s 66 112112 736736 11 1212 44
Standard_ND12sStandard_ND12s 1212 224224 14741474 22 2424 88
Standard_ND24sStandard_ND24s 2424 448448 29482948 44 3232 88
Standard_ND24rs*Standard_ND24rs* 2424 448448 29482948 44 3232 88

1 GPU = 一个 P40 卡。1 GPU = one P40 card.

*支持 RDMA*RDMA capable

NV 系列NV-series

NV 系列虚拟机采用 NVIDIA Tesla M60 GPU 和 NVIDIA GRID 技术,适用于桌面加速型应用程序和虚拟桌面,方便客户将其数据或模拟可视化。The NV-series virtual machines are powered by NVIDIA Tesla M60 GPUs and NVIDIA GRID technology for desktop accelerated applications and virtual desktops where customers are able to visualize their data or simulations. 用户可以在 NV 实例上直观显示其图形密集型工作流以获取高级图形功能,并可额外运行单精度工作负荷,例如编码和渲染。Users are able to visualize their graphics intensive workflows on the NV instances to get superior graphics capability and additionally run single precision workloads such as encoding and rendering.

NV 实例中的每个 GPU 都带有 GRID 许可证。Each GPU in NV instances comes with a GRID license. 使用此许可证,可以灵活地将 NV 实例用作单个用户的虚拟工作站,或将 25 个并发用户都连接到用于虚拟应用程序方案的 VM。This license gives you the flexibility to use an NV instance as a virtual workstation for a single user, or 25 concurrent users can connect to the VM for a virtual application scenario.

大小Size vCPUvCPU 内存:GiBMemory: GiB 临时存储 (SSD) GiBTemp storage (SSD) GiB GPUGPU 最大数据磁盘数Max data disks 最大 NIC 数Max NICs 虚拟工作站Virtual Workstations 虚拟应用程序Virtual Applications
Standard_NV6Standard_NV6 66 5656 340340 11 2424 11 11 2525
Standard_NV12Standard_NV12 1212 112112 680680 22 4848 22 22 5050
Standard_NV24Standard_NV24 2424 224224 14401440 44 6464 44 44 100100

1 GPU = 半块 M60 卡。1 GPU = one-half M60 card.

大小表定义Size table definitions

  • 存储容量的单位为 GiB 或 1024^3 字节。Storage capacity is shown in units of GiB or 1024^3 bytes. 比较以 GB(1000^3 字节)为单位的磁盘和以 GiB(1024^3 字节)为单位的磁盘时,请记住以 GiB 为单位的容量数显得更小。When comparing disks measured in GB (1000^3 bytes) to disks measured in GiB (1024^3) remember that capacity numbers given in GiB may appear smaller. 例如,1023 GiB = 1098.4 GBFor example, 1023 GiB = 1098.4 GB
  • 磁盘吞吐量的单位为每秒输入/输出操作数 (IOPS) 和 MBps,其中 MBps = 10^6 字节/秒。Disk throughput is measured in input/output operations per second (IOPS) and MBps where MBps = 10^6 bytes/sec.
  • 数据磁盘可以在缓存或非缓存模式下运行。Data disks can operate in cached or uncached modes. 对于缓存数据磁盘操作,主机缓存模式设置为 ReadOnlyReadWriteFor cached data disk operation, the host cache mode is set to ReadOnly or ReadWrite. 对于非缓存数据磁盘操作,主机缓存模式设置为 NoneFor uncached data disk operation, the host cache mode is set to None.
  • 若要获得 VM 的最佳性能,应将数据磁盘数限制为每 vCPU 2 个磁盘。If you want to get the best performance for your VMs, you should limit the number of data disks to 2 disks per vCPU.
  • 预期的网络带宽是指跨所有 NIC 为每个 VM 类型分配的最大聚合带宽,适用于所有目标。Expected network bandwidth is the maximum aggregated bandwidth allocated per VM type across all NICs, for all destinations. 不保证能够达到上述预期上限,但目的是提供相关指导,以便为目标应用程序选择适当的 VM 类型。Upper limits are not guaranteed, but are intended to provide guidance for selecting the right VM type for the intended application. 实际的网络性能将取决于许多因素,包括网络拥塞、应用程序负载和网络设置。Actual network performance will depend on a variety of factors including network congestion, application loads, and network settings. 有关如何优化网络吞吐量的信息,请参阅为 Windows 和 Linux 优化网络吞吐量For information on optimizing network throughput, see Optimizing network throughput for Windows and Linux. 若要在 Linux 或 Windows 中达到预期的网络性能,可能需要选择特定的版本,或者需要优化 VM。To achieve the expected network performance on Linux or Windows, it may be necessary to select a specific version or optimize your VM. 有关详细信息,请参阅如何针对虚拟机吞吐量进行可靠的测试For more information, see How to reliably test for virtual machine throughput.

支持的分发和驱动程序Supported distributions and drivers

若要利用运行 Linux 的 Azure N 系列 VM 的 GPU 功能,必须安装 NVIDIA GPU 驱动程序。To take advantage of the GPU capabilities of Azure N-series VMs running Linux, NVIDIA GPU drivers must be installed. NVIDIA GPU 驱动程序扩展可在 N 系列 VM 上安装适当的 NVIDIA CUDA 或 GRID 驱动程序。The NVIDIA GPU Driver Extension installs appropriate NVIDIA CUDA or GRID drivers on an N-series VM. 请使用 Azure 门户或工具(例如 Azure CLI 或 Azure 资源管理器模板)安装或管理该扩展。Install or manage the extension using the Azure portal or tools such as the Azure CLI or Azure Resource Manager templates. 有关受支持的分发版和部署步骤,请参阅 NVIDIA GPU 驱动程序扩展文档See the NVIDIA GPU Driver Extension documentation for supported distributions and deployment steps. 有关 VM 扩展的常规信息,请参阅 Azure 虚拟机扩展和功能For general information about VM extensions, see Azure virtual machine extensions and features.

如果选择手动安装 NVIDIA GPU 驱动程序,请参阅适用于 Linux 的 N 系列 GPU 驱动程序安装,了解受支持的分发版、驱动程序以及安装和验证步骤。If you choose to install NVIDIA GPU drivers manually, see N-series GPU driver setup for Linux for supported distributions, drivers, and installation and verification steps.

部署注意事项Deployment considerations

  • 有关 N 系列 VM 的可用性,请查看可用产品(按区域)For availability of N-series VMs, see Products available by region.

  • N 系列 VM 只能按 Resource Manager 部署模型部署。N-series VMs can only be deployed in the Resource Manager deployment model.

  • N 系列的 VM 在对其磁盘支持的 Azure 存储类型方面有所不同。N-series VMs differ in the type of Azure Storage they support for their disks. NC 和 NV VM 仅支持标准磁盘存储 (HDD) 所支持的 VM 磁盘。NC and NV VMs only support VM disks that are backed by Standard Disk Storage (HDD). NCv2、ND 和 NCv3 VM 仅支持高级磁盘存储 (SSD) 所支持的 VM 磁盘。NCv2, ND, and NCv3 VMs only support VM disks that are backed by Premium Disk Storage (SSD).

  • 如果需要部署的 N 系列 VM 较多,请考虑使用即用即付订阅或其他购买选项。If you want to deploy more than a few N-series VMs, consider a pay-as-you-go subscription or other purchase options. 如果使用的是 Azure 免费帐户,则仅可以使用有限数量的 Azure 计算核心。If you're using an Azure free account, you can use only a limited number of Azure compute cores.

  • 可能需要提高 Azure 订阅中的核心配额(按区域)以及单独针对 NC、NCv2、NCv3、ND 或 NV 核心的配额。You might need to increase the cores quota (per region) in your Azure subscription, and increase the separate quota for NC, NCv2, NCv3, ND, or NV cores. 若要请求提高配额,可免费提出在线客户支持请求To request a quota increase, open an online customer support request at no charge. 默认限制可能因订阅类别而异。Default limits may vary depending on your subscription category.

  • 不应在 Ubuntu NC VM 上安装 X server 或使用 Nouveau 驱动程序的其他系统。You shouldn't install X server or other systems that use the Nouveau driver on Ubuntu NC VMs. 在安装 NVIDIA GPU 驱动程序之前,需要禁用 Nouveau 驱动程序。Before installing NVIDIA GPU drivers, you need to disable the Nouveau driver.

其他大小Other sizes

后续步骤Next steps

了解有关 Azure 计算单元 (ACU) 如何帮助你跨 Azure SKU 比较计算性能的详细信息。Learn more about how Azure compute units (ACU) can help you compare compute performance across Azure SKUs.