Create an AI-optimized MIG with A4X Max

This document describes how to create a managed instance group (MIG) that uses A4X Max accelerator-optimized machine types. To learn about compute instance and cluster creation options, see Deployment options overview page.

A4X Max instance type

A Compute Engine instance, or compute instance, is a computing resource hosted on Google's infrastructure that can be either a virtual machine (VM) or a bare metal instance. A4X Max instances are available as bare metal instances, which differ from VM instances by providing direct, non-virtualized access to the underlying physical hardware. To learn more about the A4X Max machine type, see A4X Max series in the Compute Engine documentation.

Create a MIG if you want to manage multiple A4X Max instances as a single entity. To learn more about MIGs, see Managed instance groups in the Compute Engine documentation.

Limitations

When you use A4X Max machine types to create a MIG, the following limitations apply:

If you create a regional MIG that uses RDMA networking, then the MIG can only create A4X Max instances in the zone where you configured the network profile for RDMA.
You can't configure instance flexibility in the MIG.
If you apply a workload policy to a MIG, you cannot change the policy in the MIG when the group has A4X Max instances in it. To change the policy, you must first resize the MIG to zero.
You can't use an instance template that specifies a placement policy to create a MIG that uses a workload policy.

You must use the reservation-bound provisioning model. Other provisioning models are not supported.
You can't use a resize request to add A4X Max instances to the MIG. You must set the target size of the MIG to add the instances.

Before you begin

Before creating a MIG, if you haven't already done so, complete the following steps:

Choose a consumption option: your choice of consumption option determines how you get and use GPU resources.
To learn more, see Choose a consumption option.
Obtain capacity: the process to obtain capacity differs for each consumption option.
To learn about the process to obtain capacity for your chosen consumption option, see Capacity overview.

Note: When you request A4X Max or A4X capacity, you obtain it in the all capacity mode. This mode is the only supported reservation operational mode for A4X Max or A4X machine types. For more information about all capacity mode, see Reservation operational mode.

Required roles

To get the permissions that you need to create a MIG, ask your administrator to grant you the Compute Instance Admin (v1) (roles/compute.instanceAdmin.v1) IAM role on the project. For more information about granting roles, see Manage access to projects, folders, and organizations.

This predefined role contains the permissions required to create a MIG. To see the exact permissions that are required, expand the Required permissions section:

Required permissions

The following permissions are required to create a MIG:

To create a MIG: compute.instanceGroupManagers.create on the project

You might also be able to get these permissions with custom roles or other predefined roles.

A4X Max fundamentals

An A4X Max cluster is organized into a hierarchy of blocks and sub-blocks to facilitate large-scale, non-blocking network performance. Understanding this topology is key when reserving capacity and deploying workloads.

A4X Max instance: An A4X Max instance is a single A4X Max machine type that has 4 GPUs attached.
Sub-block: A sub-block is the fundamental unit of A4X Max capacity. For A4X Max, a sub-block consists of 18 A4X Max instances (72 GPUs); these instances form an NVLink domain and are connected using a multi-node NVLink system. You create an A4X Max sub-block by applying a workload policy that specifies a 1x72 topology.
Block: An A4X Max block is composed of 25 sub-blocks (NVLink domains), totalling up to 450 A4X Max instances (1,800 GPUs). The sub-blocks are rail-aligned for efficient scaling. Each sub-block requires a MIG. Therefore, for a single A4X Max block, you can create 25 MIGs.

The following table shows the supported topology options for A4X Max instances:

Topology (`acceleratorTopology`)	Number of GPUs	Number of instances
`1x72`	72	18

Overview

Creating a MIG with the A4X Max machine type includes the following steps:

Create VPC networks
Create a workload policy
Create an instance template
Create a MIG

Create VPC networks

Tip: If you are setting up a quick test, you can skip this step and specify a single NIC --network-interface=nic-type=IDPF instead.

To set up the network for A4X Max machine types, create two VPC networks for the following network interfaces:

1 regular VPC network with two subnets for the IDPF network interfaces (NICs). These are used for host to host communication.
1 VPC network with the RoCE network profile for the CX-8 NICs when creating multiple A4X Max subblocks. The RoCE VPC network uses a single subnet named default-subnet-1-RDMA_NAME_PREFIX-net that is automatically provided, and all eight CX-8 NICs use this subnet. These NICs use RDMA over Converged Ethernet (RoCE), providing the high-bandwidth, low-latency communication that's essential for scaling out to multiple A4X Max subblocks. For a single A4X Max subblock, you can skip this VPC network because within a single subblock, direct GPU to GPU communication is handled by the multi-node NVLink.

For more information about NIC arrangement, see Review network bandwidth and NIC arrangement.

Create the networks either manually by following the instruction guides or automatically by using the provided script.

Instruction guides

To create the networks, you can use the following instructions:

To create the regular VPC networks for the gVNICs, see Create and manage Virtual Private Cloud networks.
To create the RoCE VPC network, see Create a Virtual Private Cloud network for RDMA NICs.

For these VPC networks, we recommend setting the maximum transmission unit (MTU) to a larger value. For A4X Max machine types, the recommended MTU is 8896 bytes. To review the recommended MTU settings for other GPU machine types, see MTU settings for GPU machine types.

Script

To create the networks, follow these steps.

Use the following script to create regular VPC networks for the IDPF NICs.

  
    #!/bin/bash

    # Create regular VPC network for the IDPF NICs
    gcloud compute networks create IDPF_NETWORK_PREFIX-net \
      --subnet-mode=custom \
      --mtu=8896 \
      --enable-ula-internal-ipv6

    # Create subnets for the IDPF NICs
    for N in $(seq 0 1); do
      gcloud compute networks subnets create IDPF_NETWORK_PREFIX-$N \
        --network=IDPF_NETWORK_PREFIX-net \
        --region=REGION \
        --stack-type=IPV6_ONLY \
        --ipv6-access-type=INTERNAL
    done

    gcloud compute firewall-rules create IDPF_NETWORK_PREFIX-internal \
      --network=IDPF_NETWORK_PREFIX-net \
      --action=ALLOW \
      --rules=tcp:0-65535,udp:0-65535,58 \
      --source-ranges=IP_RANGE

If you require multiple A4X Max subblocks, use the following script to create the RoCE VPC network and subnets for the four CX-8 NICs on each A4X Max instance.

Important: If your deployment consists of only a single A4X Max subblock, you can skip this step.
```
  
    #!/bin/bash

    # List and make sure network profiles exist in the machine type's zone
    gcloud compute network-profiles list --filter "location.name=ZONE"

    # Create network for RDMA NICs
    gcloud compute networks create RDMA_NAME_PREFIX-net \
      --network-profile=ZONE-vpc-roce-metal \
      --subnet-mode custom \
      --mtu=8896

    # For RoCE VPC networks for bare metal instances, a single subnet named
    # default-subnet-1-RDMA_NAME_PREFIX-net is automatically provided.
    # For more details, see https://cloud.google.com/vpc/docs/rdma-network-profiles.

  
```
Replace the following:
- IDPF_NETWORK_PREFIX: the custom name prefix to use for the regular VPC networks and subnets for the IDPF NICs.
- RDMA_NAME_PREFIX: the custom name prefix to use for the RoCE VPC network and subnets for the CX-8 NICs.
- ZONE: specify a zone in which the machine type that you want to use is available, such as us-central1-a. For information about regions, see GPU availability by regions and zones.
- REGION: the region where you want to create the subnets. This region must correspond to the zone specified. For example, if your zone is us-central1-a, then your region is us-central1.
- IP_RANGE: the IP range to use for the SSH firewall rules.
Optional: To verify that the VPC network resources are created successfully, check the network settings in the Google Cloud console:
1. In the Google Cloud console, go to the VPC networks page.
  Go to VPC networks
2. Search the list for the networks that you created in the previous step.
3. To view the subnets, firewall rules, and other network settings, click the name of the network.

Create a workload policy

You cannot update a workload policy after you create it. To make changes in a workload policy, you must create a new one.

To create a workload policy, select one of the following options:

gcloud

To create a workload policy, use the gcloud compute resource-policies create workload-policy command.

For inter-accelerator network configuration, specify the --accelerator-topology flag in the command.

Use the following command for A4X Max machine type. The accelerator topology of 1x72 indicates 72 GPUs connected via a NVLink within a block.

gcloud compute resource-policies create workload-policy WORKLOAD_POLICY_NAME \
    --type=high-throughput \
    --accelerator-topology=1x72 \
    --region=REGION

Replace the following:

WORKLOAD_POLICY_NAME: the name of the workload policy.
REGION: the region where you want to create the workload policy. Specify a region in which you want to create the MIG and the machine type that you want to use is available. For information about regions, see GPU availability by regions and zones.

REST

To create a workload policy, make a POST request to the resourcePolicies.insert method.

For inter-accelerator network configuration, specify the acceleratorTopology field in the request.

Make the following request for A4X Max machine type. The accelerator topology of 1x72 indicates 72 GPUs connected via a NVLink within a block.

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/resourcePolicies
  {
    "name": "WORKLOAD_POLICY_NAME"
    "workloadPolicy": {
      "type": "HIGH_THROUGHPUT",
      "acceleratorTopology": "1x72"
    }
  }

Replace the following:

PROJECT_ID: your project ID
REGION: the region where you want to create the workload policy. Specify a region in which you want to create the MIG and the machine type that you want to use is available. For information about regions, see GPU availability by regions and zones.
WORKLOAD_POLICY_NAME: the name of the workload policy.

Create an instance template

Specify the instance properties for a MIG by creating an instance template.

To create an instance template, select one of the following options.

The following commands also set the access scope for your instances. To simplify permissions management, Google recommends that you set the access scope on an instance to cloud-platform access and then use IAM roles to define what services the instance can access. For more information, see Scopes best practice.

gcloud

To create a regional instance template, use the gcloud compute instance-templates create command.

gcloud compute instance-templates create INSTANCE_TEMPLATE_NAME \
    --machine-type=a4x-maxgpu-4g-metal \
    --image-family=IMAGE_FAMILY \
    --image-project=IMAGE_PROJECT \
    --instance-template-region=REGION \
    --boot-disk-type=hyperdisk-balanced \
    --boot-disk-size=DISK_SIZE \
    --scopes=cloud-platform \
    --network-interface=nic-type=IDPF,network=IDPF_NETWORK_PREFIX-net,stack-type=IPV6_ONLY,subnet=IDPF_NETWORK_PREFIX-sub-0 \
    --network-interface=nic-type=IDPF,network=IDPF_NETWORK_PREFIX-net,stack-type=IPV6_ONLY,subnet=IDPF_NETWORK_PREFIX-sub-1,no-address \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --network-interface=subnet=default-subnet-1-RDMA_NAME_PREFIX-net,stack-type=IPV6_ONLY,nic-type=mrdma \
    --reservation-affinity=specific \
    --reservation=RESERVATION \
    --provisioning-model=RESERVATION_BOUND \
    --instance-termination-action=DELETE \
    --maintenance-policy=TERMINATE \
    --restart-on-failure

Replace the following:

INSTANCE_TEMPLATE_NAME: the name of the instance template.
IMAGE_FAMILY: the image family of the OS image that you want to use. For a list of supported operating systems, see Supported operating systems.
IMAGE_PROJECT: the project ID of the OS image.
REGION: the region where you want to create the instance template. Specify a region in which the machine type that you want to use is available. For information about regions, see GPU availability by regions and zones.
DISK_SIZE: the size of the boot disk in GB.
IDPF_NETWORK_PREFIX: the name prefix that you specified when creating the VPC networks and subnets that use IDPF NICs.
RDMA_NAME_PREFIX: the name prefix that you specified when creating the VPC networks and subnets that use RDMA NICs.
RESERVATION: the reservation name, a block, or a subblock within a reservation. To get the reservation name or the available blocks, see View reserved capacity. Based on your requirements for instance placement, choose one of the following:
- To create A4X Max instances on any single block:
```
    projects/RESERVATION_OWNER_PROJECT_ID/reservations/RESERVATION_NAME
    
```
- To create A4X Max instances on a specific block:
```
    projects/RESERVATION_OWNER_PROJECT_ID/reservations/RESERVATION_NAME/reservationBlocks/RESERVATION_BLOCK_NAME
    
```
- To create A4X Max instances in a specific subblock:
```
    projects/RESERVATION_OWNER_PROJECT_ID/reservations/RESERVATION_NAME/reservationBlocks/RESERVATION_BLOCK_NAME/reservationSubBlocks/RESERVATION_SUBBLOCK_NAME
    
```
Tip: If the reservation exists in the current project, then you can omit projects/RESERVATION_OWNER_PROJECT_ID/reservations/ from the reservation value.

REST

To create a regional instance template, make a POST request to the regionInstanceTemplates.insert method.

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/instanceTemplates
{
  "name":"INSTANCE_TEMPLATE_NAME",
  "properties":{
    "machineType":"a4x-maxgpu-4g-metal",
    "disks":[
      {
        "boot":true,
        "initializeParams":{
          "diskSizeGb":"DISK_SIZE",
          "diskType":"hyperdisk-balanced",
          "sourceImage":"projects/IMAGE_PROJECT/global/images/family/IMAGE_FAMILY"
        },
        "mode":"READ_WRITE",
        "type":"PERSISTENT"
      }
    ],
    "serviceAccounts": [
      {
        "email": "default",
        "scopes": [
          "https://www.googleapis.com/auth/cloud-platform"
        ]
      }
    ],
    "networkInterfaces": [
    {
      "accessConfigs": [
        {
          "name": "external-nat",
          "type": "ONE_TO_ONE_NAT"
        }
      ],
      "network": "projects/NETWORK_PROJECT_ID/global/networks/IDPF_NETWORK_PREFIX-net",
      "nicType": "IDPF",
      "stackType": "IPV6_ONLY",
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/IDPF_NETWORK_PREFIX-sub-0"
    },
    {
      "network": "projects/NETWORK_PROJECT_ID/global/networks/IDPF_NETWORK_PREFIX-net",
      "nicType": "IDPF",
      "stackType": "IPV6_ONLY",
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/IDPF_NETWORK_PREFIX-sub-1"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    },
    {
      "subnetwork": "projects/NETWORK_PROJECT_ID/region/REGION/subnetworks/default-subnet-1-RDMA_NAME_PREFIX-net",
      "nicType": "MRDMA",
      "stackType": "IPV6_ONLY"
    }
  ],
    "reservationAffinity":{
        "consumeReservationType":"SPECIFIC_RESERVATION",
        "key":"compute.googleapis.com/reservation-name",
        "values":[
          "RESERVATION"
        ]
      },
    "scheduling":{
        "provisioningModel":"RESERVATION_BOUND",
        "instanceTerminationAction":"DELETE",
        "onHostMaintenance": "TERMINATE",
        "automaticRestart":true
      }
  }
}

Replace the following:

INSTANCE_TEMPLATE_NAME: the name of the instance template.
IMAGE_FAMILY: the image family of the OS image that you want to use. For a list of supported operating systems, see Supported operating systems.
IMAGE_PROJECT: the project ID of the OS image.
REGION: the region where you want to create the instance template. Specify a region in which the machine type that you want to use is available. For information about regions, see GPU availability by regions and zones.
DISK_SIZE: the size of the boot disk in GB.
NETWORK_PROJECT_ID: the project ID of the network.
IDPF_NETWORK_PREFIX: the name prefix that you specified when creating the VPC networks and subnets that use IDPF NICs.
REGION: the region of the subnetwork.
RDMA_NAME_PREFIX: the name prefix that you specified when creating the VPC networks and subnets that use RDMA NICs.
RESERVATION: the reservation name, a block, or a subblock within a reservation. To get the reservation name or the available blocks, see View reserved capacity. Based on your requirements for instance placement, choose one of the following:
- To create A4X Max instances on any single block:
```
    projects/RESERVATION_OWNER_PROJECT_ID/reservations/RESERVATION_NAME
    
```
- To create A4X Max instances on a specific block:
```
    projects/RESERVATION_OWNER_PROJECT_ID/reservations/RESERVATION_NAME/reservationBlocks/RESERVATION_BLOCK_NAME
    
```
- To create A4X Max instances in a specific subblock:
```
    projects/RESERVATION_OWNER_PROJECT_ID/reservations/RESERVATION_NAME/reservationBlocks/RESERVATION_BLOCK_NAME/reservationSubBlocks/RESERVATION_SUBBLOCK_NAME
    
```
Tip: If the reservation exists in the current project, then you can omit projects/RESERVATION_OWNER_PROJECT_ID/reservations/ from the reservation value.

Create a MIG

When creating a MIG using the A4X Max machine type, specify the instance count by using the MIG's target size. You can set the target size during MIG creation to either the required number of instances or start with zero and increase it later. Note that you cannot use resize requests to add instances to an A4X Max MIG.

To obtain a GPU topology of 1x72, create a MIG with 18 A4X Max instances. When you create the MIG, apply the workload policy that specifies the acceleratorTopology field. Applying the policy ensures that Compute Engine creates all 18 A4X Max instances in one sub-block to use an NVLink domain. If a sub-block lacks capacity for all 18 instances, then any managed instance that can't be created immediately will remain in CREATING state until capacity becomes available. These managed instances represent the compute instances that the MIG creates when there is capacity.

When you apply a workload policy with the acceleratorTopology field set to 1x72, you cannot create more than 18 A4X Max instances in a MIG. Specifying more than 18 instances causes the MIG creation to fail. To create multiple NVLink domains, create a separate MIG for each domain and apply the same workload policy to each MIG.

To create a MIG, select one of the following options:

gcloud

To create a MIG with a specified target size, use the instance-groups managed create command.

Create a zonal or regional MIG as follows:

To create a zonal MIG, use the following command:

gcloud compute instance-groups managed create MIG_NAME \
  --template=INSTANCE_TEMPLATE_URL \
  --size=TARGET_SIZE \
  --workload-policy=WORKLOAD_POLICY_URL \
  --zone=ZONE

To create a regional MIG, use the following command:

gcloud compute instance-groups managed create MIG_NAME \
    --template=INSTANCE_TEMPLATE_URL \
    --size=TARGET_SIZE \
    --workload-policy=WORKLOAD_POLICY_URL \
    --region=REGION

Replace the following:

MIG_NAME: the name of the MIG.
INSTANCE_TEMPLATE_URL: the URL of the instance template that you want to use to create instances in the MIG. The URL can contain either the ID or name of the instance template. Specify one of the following values:
- For a regional instance template: projects/PROJECT_ID/regions/REGION/instanceTemplates/INSTANCE_TEMPLATE_ID
- For a global instance template: INSTANCE_TEMPLATE_ID
TARGET_SIZE: the number of instances that you want in the MIG.
WORKLOAD_POLICY_URL: Optional: the URL of the workload policy. For example, projects/example-project/regions/us-central1/resourcePolicies/example-workload-policy. If you don't want to use a workload policy, then you can remove the --workload-policy flag.
ZONE: the zone in which you want to create the MIG. If you use a workload policy, then specify a zone within the policy's region.
REGION: the region in which you want to create the MIG. If you use a workload policy, then specify the same region as that of the policy. For a regional MIG, instead of a region, you can specify the zones in that region by using the --zones flag.

REST

To create a MIG with a specified target size, make a POST request.

Create a zonal or regional MIG as follows:

To create a zonal MIG, make a POST request to the instanceGroupManagers.insert method.

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/instanceGroupManagers
{
  "versions": [
    {
      "instanceTemplate": "INSTANCE_TEMPLATE_URL"
    }
  ],
  "name": "MIG_NAME",
  "instanceTemplate": "INSTANCE_TEMPLATE_URL",
  "targetSize": "TARGET_SIZE",
  "resourcePolicies": {
    "workloadPolicy": WORKLOAD_POLICY_URL
  }
}

To create a regional MIG, make a POST request to the regionInstanceGroupManagers.insert method.

  POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION/instanceGroupManagers
  {
    "versions": [
      {
        "instanceTemplate": "INSTANCE_TEMPLATE_URL"
      }
    ],
    "name": "MIG_NAME",
    "instanceTemplate": "INSTANCE_TEMPLATE_URL",
    "targetSize": "TARGET_SIZE",
    "resourcePolicies": {
      "workloadPolicy": WORKLOAD_POLICY_URL
    }
  }

Replace the following:

PROJECT_ID: the project ID.
ZONE: the zone in which you want to create the MIG. Specify a zone within the workload policy's region.
REGION: the region in which you want to create the MIG. Specify the same region as that of the workload policy.
INSTANCE_TEMPLATE_URL: the URL of the instance template that you want to use to create instances in the MIG. The URL can contain either the ID or name of the instance template. Specify one of the following values:
- For a regional instance template: projects/PROJECT_ID/regions/REGION/instanceTemplates/INSTANCE_TEMPLATE_ID
- For a global instance template: INSTANCE_TEMPLATE_ID
MIG_NAME: the name of the MIG.
TARGET_SIZE: the number of instances that you want in the MIG.
WORKLOAD_POLICY_URL: Optional: the URL of the workload policy. For example, projects/example-project/regions/us-central1/resourcePolicies/example-workload-policy.

Create an AI-optimized MIG with A4X Max Stay organized with collections Save and categorize content based on your preferences.

A4X Max instance type

Limitations

Before you begin

Required roles

Required permissions

A4X Max fundamentals

Overview

Create VPC networks

Instruction guides

Script

Create a workload policy

gcloud

REST

Create an instance template

gcloud

REST

Create a MIG

gcloud

REST

What's next?

Create an AI-optimized MIG with A4X Max