Amazon SageMaker , a fully-managed service to build, train, and deploy machine learning models, now supports random search as a tuning strategy and multiple hyperparameter scaling options when using Automatic Model Tuning.

AWS customers can now seamlessly accelerate their High Performance Computing (HPC), machine-learning, and other batch jobs through AWS Batch simply by specifying the number of GPUs each job requires. Starting today, you can use AWS Batch to specify the number and type of accelerators your jobs require as job definition input variables, alongside the current options of vCPU and memory. AWS Batch will scale up instances appropriate for your jobs based on the required number of GPUs and isolate the accelerators according to each job’s needs, so only the appropriate containers can access them.

Hardware-based compute accelerators such as Graphics Processing Units (GPUs) enable users to increase application throughput and decrease latency with purpose-built hardware. Until now, AWS Batch users wanting to take advantage of accelerators needed to build a custom AMI and install the appropriate drivers, and have AWS Batch scale GPU accelerated EC2 P-type instances based on their vCPU and memory characteristics. Now, customers can simply specify the desired number and type of GPUs, similar to how they can specify vCPU and memory, and Batch will launch the EC2 P-type instances needed to run the jobs. Additionally, Batch isolates the GPU to the container, so each container gets the appropriate amount of resources it needs.

Learn more about GPU support on AWS Batch here.
 

Amazon FSx for Lustre is now available in the AWS Asia Pacific (Sydney) Region.

Amazon FSx for Windows File Server is now available in the AWS Asia Pacific (Sydney) Region. 

AWS RoboMaker makes it easy to develop, test, and deploy intelligent robotics applications at scale. Today we have added support for the Gazebo 9 engine, a leading robotics simulator. Running simulation jobs on Gazebo 9 provides new features and benefits such as .obj mesh support, GUI plotting utility, improved actor animation, and improved shadow. For a full list of Gazebo 9 features, visit the Gazebo home page .

AWS RoboMaker is available in US East (N. Virginia), US West (Oregon), and EU (Ireland) regions. To get started, run a sample simulation job in the RoboMaker console or explore the RoboMaker webpage .
 

Amazon Elastic Container Service for Kubernetes (Amazon EKS) can now send log data from the Kubernetes control plane to Amazon CloudWatch Logs. These logs make it easier to monitor changes made to and performance of your Amazon EKS clusters.

The AWS Serverless Application Repository is now available in EU (Paris) and EU (Stockholm) regions. The addition of the EU (Paris) and EU (Stockholm) regions increases the availability of the AWS Serverless Application Repository, which offers region support for Asia Pacific (Mumbai, Singapore, Sydney, Tokyo), Canada (Central), EU (Frankfurt, Ireland, London, Paris), South America (São Paulo), US West (N. California, Oregon), and US East (N. Virginia, Ohio). The AWS Serverless Application Repository is now available in 16 regions.

Amazon Redshift automatically takes incremental snapshots (backups) of your data every 8 hours or 5 GB per node of data change. You now get more information and control over a snapshot including the ability to control the automatic snapshot’s schedule.

Customers in the EU (Paris) Region can now use AWS Secrets Manager to manage secrets such as database passwords and API keys needed to access their applications, services, and IT resources. 

Amazon Comprehend is a natural language processing (NLP) service that uses machine learning to discover insights and relationships in text. Starting today, you can assign tags to custom classifier and entity type resources. This enables you to allocate costs and get detailed billing reports across your custom models.

Each tag is a simple label consisting of a customer-defined key and an optional value that can make it easier to manage, search for, and filter resources. You can allocate costs by creating tagged Billing Groups and mapping individual “Things” to those groups.

All usage and cost for a model will inherit the tags of the Billing Group to which it belongs. IAM policies also support tag-based conditions, enabling you to constrain IAM permissions based on specific tags or tag values (when leveraging tag-based conditions for access control, make sure to also define and restrict who can modify those tags).

These features are provided at no additional cost. Click here to learn more about how to create and use tags in Amazon Comprehend . See AWS Tagging Strategies for general best practices for using tags with AWS resources.