Version 6.13, part of Release 2018b, includes the following enhancements:

  • GPU Support: View details of GPU support on over 600 function pages, and browse GPU support for functions by toolbox
  • GPU Functionality: Use new and enhanced gpuArray functions, such as spline interpolation and vecnorm
  • Support for NVIDIA CUDA 9.1: Update to CUDA Toolkit 9.1
  • Distributed Arrays: Use new and enhanced distributed array functionality, including support for vecnorm and writing to Amazon S3 and Azure
  • Improved Scalability: Use up to 1024 workers per parallel pool
  • Streamlined Cloud Cluster Setup: Create new Cloud Center clusters directly from the MATLAB desktop

See the Release Notes for details.

Version 6.12, part of Release 2018a, includes the following enhancements:

  • Parfeval callbacks: New afterAll and afterEach methods for parallel futures
  • Slurm support: Slurm is a fully supported scheduler
  • Support for NVIDIA Volta: Update to CUDA 9, support for Volta class GPUs
  • Improved file mirroring: Performance of file mirroring for generic scheduler integration
  • ​Parfor performance improvements: More efficient broadcast variables in parfor for non-local clusters

See the Release Notes for details.

Version 6.11, part of Release 2017b, includes the following enhancements:

  • Improved Parallel Language Performance: Execute parallel language constructs with reduced overhead
  • Tall Array Support: Use tall arrays with Windows client access to Linux Spark clusters
  • Improved Parallel Pool Robustness: Run pools without Message Passing Interface (MPI) by default, making pools resilient to workers crashing​​​​
  • Improved MATLAB Integration with Third-Party Schedulers: Use the Generic Profile Wizard for easier installation and setup of MATLAB Distributed Computing Server
  • Cloud Storage: Work with data in Microsoft Azure Blob Storage

See the Release Notes for details.

Version 6.10, part of Release 2017a, includes the following enhancements:

  • Access to Intermediate Results and Updates in Parallel Computations: Poll for messages or data from different workers during parallel workflows
  • Tall Array Support: Use parallel execution environments for tall timetables and enhanced tall array functions
  • More Responsive Job Monitor: Automatic updates for new, submitted, or deleted jobs or tasks

See the Release Notes for details.

Version 6.9, part of Release 2016b, includes the following enhancements:

  • Parallel Support for Tall Arrays: Process big data with tall arrays in parallel on your desktop, MATLAB Distributed Computing Server, and Spark clusters
  • Support for GPU arrays: Use enhanced gpuArray functions, including new sparse iterative solver bicg
  • Parallel Menu Enhancement: Use the new menu items in the Parallel Menu to configure and manage cloud based resources
  • New Data Types in Distributed Arrays: Use enhanced functions for creating distributed arrays of: datetime; duration; calendarDuration; string; categorical; and table
  • Loading Distributed Arrays: Load distributed arrays in parallel using datastore
  • Cluster Profile Validation: Choose which validation stages run and the number of MATLAB workers to use

See the Release Notes for details.

Version 6.8, part of Release 2016a, includes the following enhancements:

  • GPU Support for Sparse Matrices: Use enhanced gpuArray functions for sparse matrices on GPUs
  • Support for Distributed Arrays: Use enhanced distributed array functions including sparse input to direct (mldivide) and iterative solvers (cgs and pcg)
  • GPU-Accelerated Deep Learning: Use Deep Learning Toolbox to train deep convolutional neural networks with GPU-enabled acceleration for image classification tasks
  • GPU-enabled MATLAB Functions: Accelerate applications using GPU-enabled MATLAB functions for linear equations, descriptive statistics and set operations
  • Parallel-Enabled Gradient Estimation: Accelerate more nonlinear solvers in the Optimization Toolbox with parallel finite difference estimation of gradients and Jacobians
  • Hadoop Kerberos Support: Improved support for Hadoop in a Kerberos authenticated environment
  • Increased Data Transfer Limits: Transfer data up to 4GB in size between client and workers in any job using a MATLAB job scheduler cluster

See the Release Notes for details.