The NDv2-series virtual machine is a new addition to the GPU family designed for the needs of the most demanding GPU-accelerated AI, machine learning, simulation, and HPC workloads.
NDv2 is powered by 8 NVIDIA Tesla V100 NVLINK-connected GPUs, each with 32 GB of GPU memory. Each NDv2 VM also has 40 non-HyperThreaded Intel Xeon Platinum 8168 (Skylake) cores and 672 GiB of system memory.
NDv2 instances provide excellent performance for HPC and AI workloads utilizing CUDA GPU-optimized computation kernels, and the many AI, ML, and analytics tools that support GPU acceleration 'out-of-box,' such as TensorFlow, Pytorch, Caffe, RAPIDS, and other frameworks.
Critically, the NDv2 is built for both computationally intense scale-up (harnessing 8 GPUs per VM) and scale-out (harnessing multiple VMs working together) workloads. The NDv2 series now supports 100-Gigabit InfiniBand EDR backend networking, similar to that available on the HB series of HPC VM, to allow high-performance clustering for parallel scenarios including distributed training for AI and ML. This backend network supports all major InfiniBand protocols, including those employed by NVIDIA’s NCCL2 libraries, allowing for seamless clustering of GPUs.
When enabling InfiniBand on the ND40rs_v2 VM, please use the 4.7-22.214.171.124 Mellanox OFED driver.
Due to increased GPU memory, the new ND40rs_v2 VM requires the use of Generation 2 VMs and marketplace images.
Please note: The ND40s_v2 featuring 16 GB of per-GPU memory is no longer available for preview and has been superceded by the updated ND40rs_v2.
|Size||vCPU||Memory: GiB||Temp Storage (SSD): GiB||GPU||GPU Memory: GiB||Max data disks||Max uncached disk throughput: IOPS / MBps||Max network bandwidth||Max NICs|
|Standard_ND40rs_v2||40||672||2948||8 V100 32 GB (NVLink)||32||32||80000 / 800||24000 Mbps||8|
Size table definitions
Storage capacity is shown in units of GiB or 1024^3 bytes. When you compare disks measured in GB (1000^3 bytes) to disks measured in GiB (1024^3) remember that capacity numbers given in GiB may appear smaller. For example, 1023 GiB = 1098.4 GB.
Disk throughput is measured in input/output operations per second (IOPS) and MBps where MBps = 10^6 bytes/sec.
Data disks can operate in cached or uncached modes. For cached data disk operation, the host cache mode is set to ReadOnly or ReadWrite. For uncached data disk operation, the host cache mode is set to None.
If you want to get the best performance for your VMs, you should limit the number of data disks to two disks per vCPU.
Expected network bandwidth is the maximum aggregated bandwidth allocated per VM type across all NICs, for all destinations. For more information, see Virtual machine network bandwidth.
Upper limits aren't guaranteed. Limits offer guidance for selecting the right VM type for the intended application. Actual network performance will depend on several factors including network congestion, application loads, and network settings. For information on optimizing network throughput, see Optimize network throughput for Azure virtual machines. To achieve the expected network performance on Linux or Windows, you may need to select a specific version or optimize your VM. For more information, see Bandwidth/Throughput testing (NTTTCP).
Supported operating systems and drivers
To take advantage of the GPU capabilities of Azure N-series VMs, NVIDIA GPU drivers must be installed.
The NVIDIA GPU Driver Extension installs appropriate NVIDIA CUDA or GRID drivers on an N-series VM. Install or manage the extension using the Azure portal or tools such as Azure PowerShell or Azure Resource Manager templates. For general information about VM extensions, see Azure virtual machine extensions and features.
If you choose to install NVIDIA GPU drivers manually, see N-series GPU driver setup for Linux.
- General purpose
- Memory optimized
- Storage optimized
- GPU optimized
- High performance compute
- Previous generations
Learn more about how Azure compute units (ACU) can help you compare compute performance across Azure SKUs.