What’s New in Intel® TBB 2.2

With Intel® Threading Building Blocks version 2.2, now available for download, come several new features and improvements. Here’s a quick summary of what to look forward to:

Improved performance

    • Improved performance of the scalable memory allocator

    • Significant redesign of the task scheduler for better performance and scalability

    • Improved performance of the concurrent hash map

    • Better performance of the affinity partitioner

    • The auto partitioner is now the default partitioner for loop templates instead of the simple partitioner

Improved experience

    • Intel® TBB 2.2 supports automatic initialization of the task scheduler (task_scheduler_init is now optional)

    • New high-level interface task_group for writing tasks

    • Default implementation of HashCompare simplifies usage of concurrent_hash_map

    • Expanded support of lambda expressions makes it easier to read and maintain code when using a lambda-capable compiler such as the Intel® C++ Compiler 11.0 and its successors as well as GCC and Microsoft* C++ Compilers

New and improved parallel algorithms

    • New algorithms parallel_invoke and parallel_for_each

    • New serial unordered and thread bound filters in pipeline

    • Simplified interface of parallel_for for common loops

New and improved data containers

    • New classes enumerable_thread_specific and combinable to support cross-platform thread local storage and associated algorithms

    • Unbounded non-blocking interface for concurrent_queue and new blocking concurrent_bounded_queue

    • Improved interfaces and behavior guarantees for concurrent_vector

New in scalable memory allocator

    • Improved support for memory requests bigger than 8K

    • Intel® TBB 2.2 supports automatic replacement of the OS allocator with its scalable memory allocator (Microsoft Windows* and Linux* OS)

We plan to describe most of these changes in more detail over the next few weeks.

Para obter informações mais completas sobre otimizações do compilador, consulte nosso aviso de otimização.