Selecting Instance Types#

An instance is a server hosted by a cloud provider and instance types (or machine types) are defined by combinations of CPU, memory, storage, and networking capacity (see GCP supported instance types and AWS supported instance types for more details).

Specifying memory and CPU#

When you create a cluster, you can provide a range of required resources, and Coiled will select available, compatible instance types for you. In the following example, Coiled will select instance types with between 2 and 8 CPUs and between 32 and 64 GiB of memory:

from coiled import Cluster
from distributed import Client

cluster = Cluster(worker_cpu=[2, 8], worker_memory=["32GiB", "64GiB"])
client = Client(cluster)

When Coiled identifies a number of compatible instance types, they are prioritized in order of lowest to highest estimated cost.

Requesting specific instance types#

To allow for more fine-grain control of the type of cluster you create, you can provide a list of specific instance types for the scheduler and workers using the scheduler_vm_types and worker_vm_types keyword arguments.

For example:

import coiled

cluster = coiled.Cluster(
    scheduler_vm_types=["t3.large", "t3.xlarge"],
    worker_vm_types=["m6i.large", "m6in.large"],

It’s recommended you specify more than one instance type in your list to avoid instance availability issues in the cloud provider and region that you are using Coiled. Lists of instance types are prioritized in order of decreasing priority.

Allowable Instance Types#

You can use coiled.list_instance_types() to return a dictionary of allowed instance types, with values for each instance type and keys containing a dictionary of characteristics for that instance type.

You can specify a single keyword argument or a combination (i.e. backend, cores, memory, arch, and gpus) to filter the results.

import coiled

# Filter instances by cloud provider, cores, and memory
coiled.list_instance_types(backend="aws", cores=2, memory="8 Gib")

You can also provide list_instance_types with a range of values, for example:

coiled.list_instance_types(backend="gcp", cores=[2, 8])

For those interested in GPUs, you can use gpu to specify an exact number or a range of GPUs. The following, for example, will return all Google Cloud instance types containing at least one, but not more than 4 GPUs.

coiled.list_instance_types(backend="gcp", gpus=[1, 2])


For AWS, GPUs are tied to the instance type, but for GCP you can add different GPU types to GPU enabled instances (see GPUs).

You can also use arch to filter by CPU architecture. The following, for example, will return available ARM-based instance types (see Using ARM on Coiled):

coiled.list_instance_types(backend="aws", arch="arm64")