Roadmap

A non-comprehensive list of future goals

• Stateful Dispatch
• Bulk Synchronous Data Parallel Kernel Launch
• Group Shared Variables
• Promises & Futures
• Asynchronous Algorithm Interface
• State-of-the-Art Counter-Based Random Number Generator
• Range-Based Interface
• Dynamic Backend
• Distributed Storage
• NUMA Systems
• multi-GPU
• Cluster
• Enable Thrust algorithms from CUDA __device__ Functions
• Block-Convergent Algorithms
• Improved Algorithm Performance
  • merge and set operations
  • radix sort
  • merge sort
• Improved Productivity when using Thrust
  • Concise Compilation Error Reporting
  • Generalized for_each and transform

A non-comprehensive list of non-goals (unlikely to implement)

• Support for Domain-Specific Functionality
• Multi-Dimensional Data Structures
• Support for Backend System-Specific Idiosyncracies