List of available Amazon Web Services EU instances

The list below shows available AWS instances that you can choose and specify as a value for the sbg:AWSInstanceType hint. See the AWS page on instance types for details on pricing.

List of instances with ephemeral storage

Name	Cores	RAM [GB]	Storage [GB]
i2.xlarge	4	30.5	800
i2.2xlarge	8	61	1600
i2.4xlarge	16	122	3200
i2.8xlarge	32	244	6400
x1.16xlarge	64	976	1920
x1.32xlarge	128	1952	1920

List of instances with variable attached EBS storage

Storage size can be changed to anything between 2GB and 4 TB.
Learn more from our EBS Customization Documentation

Name	Cores	RAM [GB]
c4.large	2	3.75
c4.xlarge	4	7.5
c4.2xlarge	8	15
c4.4xlarge	16	30
c4.8xlarge	36	60
c5.large	2	4
c5.xlarge	4	8
c5.2xlarge	8	16
c5.4xlarge	16	32
c5.9xlarge	36	72
c5.12xlarge	48	96
c5.18xlarge	72	144
c5.24xlarge	96	192
m4.large	2	8
m4.xlarge	4	16
m4.2xlarge	8	32
m4.4xlarge	16	64
m4.10xlarge	40	160
m4.16xlarge	64	256
m5.large	2	8
m5.xlarge	4	16
m5.2xlarge	8	32
m5.4xlarge	16	64
m5.8xlarge	32	128
m5.12xlarge	48	192
m5.16xlarge	64	256
m5.24xlarge	96	384
r4.large	2	15.25
r4.xlarge	4	30.5
r4.2xlarge	8	61
r4.4xlarge	16	122
r4.8xlarge	32	244
r4.16xlarge	64	488
r5.large	2	16
r5.xlarge	4	32
r5.2xlarge	8	64
r5.4xlarge	16	128
r5.8xlarge	32	256
r5.12xlarge	48	384
r5.16xlarge	64	512
r5.24xlarge	96	768

GPU Instances

The Platform also supports the following powerful, scalable GPU instances that deliver high performance compute in the cloud. Designed for general-purpose GPU compute applications using CUDA and OpenCL, these instances are ideally suited for machine learning, molecular modeling, genomics, rendering, and other workloads requiring massive parallel floating point processing power.

Name	GPUs	vCPUs	RAM (GiB)
p2.xlarge	1	4	61
p2.8xlarge	8	32	488
p2.16xlarge	16	64	732
p3.2xlarge	1 Tesla v100	8	61
p3.8xlarge	4 Tesla v100	32	244
p3.16xlarge	8 Tesla v100	64	488
g4dn.xlarge	1	4	16
g4dn.2xlarge	1	8	32
g4dn.4xlarge	1	16	64
g4dn.8xlarge	1	32	128
g4dn.16xlarge	1	64	256
g4dn.12xlarge	4	48	192

Creating Docker images containing tools that are run on GPU instances is similar to the process of creating Docker images with tools that are designed for CPU instances. The only major difference is that GPU tools have additional requirements for interaction with GPUs, which can be either OpenCL or CUDA. NVIDIA drivers come preinstalled and optimized according to the Amazon best practice for the specific instance family and are accessible from the Docker container. It is recommended to use one of Docker images provided by NVIDIA as the base image. For tools that require CUDA, the list of supported images are available at https://hub.docker.com/r/nvidia/cuda/, and for tools that are based on OpenCL at https://hub.docker.com/r/nvidia/opencl. The rest of the procedure for creating and uploading a Docker image is the same as for tools designed to run on CPU instances. In case you have any problems with the setup, please contact our Support Team.

When creating a Docker image containing GPU tools, it should be taken into account that older binaries are usually built for older GPU architectures and might not work on newer GPUs. If that is the case, these binaries can’t be used, and new ones should be built from source code.