The NCCads H100 v5 series of virtual machines are a new addition to the Azure GPU family. In this VM SKU, Trusted Execution Environment (TEE) spans confidential VM on the CPU and attached GPU, enabling secure offload of data, models, and computation to the GPU.
The NCCads H100 v5 series is powered by 4th-generation AMD EPYC™ Genoa processors and NVIDIA H100 Tensor Core GPU. These VMs feature 1 NVIDIA H100 NVL GPUs with 94 GB memory, 40 non-multithreaded AMD EPYC Genoa processor cores, and 320 GiB of system memory. These VMs are ideal for real-world Applied AI workloads, such as:
GPU-accelerated analytics and databases
Batch inferencing with heavy pre- and post-processing
1Temp disk speed often differs between RR (Random Read) and RW (Random Write) operations. RR operations are typically faster than RW operations. The RW speed is usually slower than the RR speed on series where only the RR speed value is listed.
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.
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.
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).
Accelerator (GPUs, FPGAs, etc.) info for each size
Size Name
Accelerators (Qty.)
Accelerator-Memory (GB)
Standard_NCC40ads_H100_v5
1
94
Creating VM
To get started with NCCads H100 v5 VMs and for information regarding confidential GPU driver and attesetation configuration, see confidential GPU Configuration