Console
Start free

gcloud compute tpus tpu-vm create

NAME
gcloud compute tpus tpu-vm create - create a new Cloud TPU VM node
SYNOPSIS
gcloud compute tpus tpu-vm create (TPU : --zone=ZONE) --version=VERSION [--async] [--boot-disk=[KEY=VALUE,…]] [--data-disk=[mode=MODE],[source=SOURCE]] [--description=DESCRIPTION] [--internal-ips] [--labels=[KEY=VALUE,…]] [--metadata=[KEY=VALUE,…]] [--metadata-from-file=[KEY=VALUE,…]] [--network=NETWORK; default="default"] [--preemptible] [--queue-count=QUEUE_COUNT] [--range=RANGE] [--reserved] [--scopes=[SCOPES,…]] [--service-account=SERVICE_ACCOUNT] [--shielded-secure-boot] [--spot] [--subnetwork=SUBNETWORK] [--tags=[TAGS,…]] [--accelerator-type=ACCELERATOR_TYPE; default="v2-8"     | --topology=TOPOLOGY --type=TYPE] [GCLOUD_WIDE_FLAG …]
DESCRIPTION
Create a new Cloud TPU VM node.
EXAMPLES
To create a TPU VM node with ID my-tpu in the default user project, network and compute/zone (with other defaults supplied by API), run: 
gcloud compute tpus tpu-vm create my-tpu
 To create a TPU VM node in a specific network, run: 
gcloud compute tpus tpu-vm create my-tpu --zone=us-central1-a --network=my-tf-network --description='My TPU VM' --version='v2-alpha'

To create a small TPU VM v2 pod, run: 

gcloud compute tpus tpu-vm create my-tpu --zone=us-central1-a --accelerator-type='v2-32' --description='My TPU VM' --version='v2-alpha'
POSITIONAL ARGUMENTS
Tpu resource - Name of the Cloud TPU VM node to create. The arguments in this group can be used to specify the attributes of this resource. (NOTE) Some attributes are not given arguments in this group but can be set in other ways.

To set the project attribute:

This must be specified.
TPU
ID of the tpu or fully qualified identifier for the tpu.

To set the tpu attribute:

This positional argument must be specified if any of the other arguments in this group are specified.
--zone=ZONE
Zone of the Cloud TPU. If not specified, will use default compute/zone. To set the zone attribute:
REQUIRED FLAGS
--version=VERSION
Runtime version for the TPU, such as 2.3.

For a list of available versions run:

gcloud compute tpus tpu-vm versions list
OPTIONAL FLAGS
--async
Return immediately, without waiting for the operation in progress to complete.
--boot-disk=[KEY=VALUE,…]
Specifies the boot disk configuration. 
gcloud compute tpus tpu-vm create --boot-disk kms-key=<full_kms_key_name_here>
 The following keys are allowed:
kms-key
Specifies the fully qualified Cloud KMS cryptokey name which will be used to protect the disk. KMS cryptokey name format: projects/<kms-project>/locations/<kms-location>/keyRings/<kms-keyring>/cryptoKeys/<key-name>
--data-disk=[mode=MODE],[source=SOURCE]
Additional data disks for the TPU VM.

This flag must be repeated to provide multiple data disks. For example: 

gcloud compute tpus tpu-vm create --data-disk source=projects/my-project/zones/us-central1-c/disks/my-disk,mode=read-only
 The following keys are allowed:
source
Specifies the full path to an existing disk. Required. The disk must be in the same zone.
mode
Specifies the mode in which to attach this disk. Valid options are 'read-write', 'read-only'. If not specified, the default is 'read-write'.
--description=DESCRIPTION
Text description of the TPU.
--internal-ips
Indicate that the IP addresses for the node should be internal. The default is that external IP addresses will be associated with the TPU workers.
--labels=[KEY=VALUE,…]
Resource labels to represent user-provided metadata. See https://cloud.google.com/compute/docs/labeling-resources for details.
--metadata=[KEY=VALUE,…]
List of comma-separated metadata key-value pairs for the Cloud TPU VM node. Example: --metadata='key1=value1,key2=value2'
--metadata-from-file=[KEY=VALUE,…]
Same as --metadata except the value for the entry will be read from a local file. Example: --metadata-from-file='key1=value1.txt'
--network=NETWORK; default="default"
Network that this TPU will be a part of.
--preemptible
If provided, the TPU will be preemptible and time-limited. It may be preempted to free up resources for standard TPUs, and will only be able to run for a limited amount of time. Preemptible TPUs cannot be restarted.
--queue-count=QUEUE_COUNT
Specifies the networking queue count for TPU VM instances. Both Rx and Tx queues will be set to this number. If it's not specified, a default queue count will be assigned. For Virtio-net, each interface will get min(floor(#vCPU / #vNIC), 32) queues. For gVNIC, each interface will get min(floor(#vCPU / #vNIC / 2), 16) queues.
--range=RANGE
CIDR Range for the TPU. The IP range that the TPU will select an IP address from. Must be in CIDR notation and a /29 range, for example 192.168.0.0/29. Errors will occur if the CIDR range has already been used for a currently existing TPU, the CIDR range conflicts with any networks in the user's provided network, or the provided network is peered with another network that is using that CIDR range.
--reserved
When specified, will attempt to create the TPU node under reservations made in the current project. The reservations can be made separately but used in aggregated form. i.e., the user can make a reservation of 128 V2 TPUs and later on make another reservation of 128 V2 TPUs then creates a v2-256 TPU instance. If there exists no reservation or not sufficient amount of reserved cores under the project, the request will fail due to lack of capacity.
--scopes=[SCOPES,…]
List of comma-separated scopes to be made available for the service account.
--service-account=SERVICE_ACCOUNT
Email address of the service account. If empty, default Google Compute Engine service account will be used.
--shielded-secure-boot
Specifies that the TPU instances are created with secure boot enabled. This implicitly makes them Shielded VM instances.
--spot
If specified, create this VM as a spot VM. Spot VMs make unused capacity available at highly discounted rates. Spot VMs may be preempted at any time if the capacity is needed, but unless preempted there is no limit on runtime duration. Spot VM TPUs cannot be restarted, and must be recreated again.
--subnetwork=SUBNETWORK
Subnetwork that this TPU will be a part of.
--tags=[TAGS,…]
Tags to apply to the TPU Node. Tags are used to identify valid sources or targets for network firewalls. See https://cloud.google.com/vpc/docs/add-remove-network-tags for more details.
Specify a TPU node in either chips or cores At most one of these can be specified: Core based TPU configuration
--accelerator-type=ACCELERATOR_TYPE; default="v2-8"
TPU accelerator type for the TPU. If not specified, this defaults to v2-8.

For a list of available accelerator types run:

gcloud compute tpus tpu-vm accelerator-types list Chip based TPU configuration
--topology=TOPOLOGY
Chip topology for TPU. This flag argument must be specified if any of the other arguments in this group are specified.
--type=TYPE
Type of TPU. TYPE must be one of: v2, v3, v4. This flag argument must be specified if any of the other arguments in this group are specified.
GCLOUD WIDE FLAGS
These flags are available to all commands: --access-token-file, --account, --billing-project, --configuration, --flags-file, --flatten, --format, --help, --impersonate-service-account, --log-http, --project, --quiet, --trace-token, --user-output-enabled, --verbosity.

Run $ gcloud help for details.

API REFERENCE
This command uses the tpu/v2 API. The full documentation for this API can be found at: https://cloud.google.com/tpu/
NOTES
This variant is also available: 
gcloud alpha compute tpus tpu-vm create