Subject: CVS commit: pkgsrc/parallel/threadingbuildingblocks
From: Ryo ONODERA
Date: 2020-09-05 13:34:08
Message id: 20200905113408.47207FB28@cvs.NetBSD.org

Log Message:
threadingbuildingblocks: Update to 2020.3

Changelog:
Intel TBB 2020 Update 3
TBB_INTERFACE_VERSION == 11103

Changes (w.r.t. Intel TBB 2020 Update 2):

Changes affecting backward compatibility:

- Changed body type concept of the flow::input_node.
    Set TBB_DEPRECATED_INPUT_NODE_BODY to 1 to compile with the previous
    concept of the body type.

Bugs fixed:

- Fixed compilation errors in C++20 mode due to ambiguity of comparison
    operators. Inspired by Barry Revzin
    (https://github.com/oneapi-src/oneTBB/pull/251).

Open-source contributions integrated:

- Fixed an issue in TBBBuild.cmake that causes the build with no arguments
    to fail (https://github.com/oneapi-src/oneTBB/pull/233) by tttapa.
- Added cmake/{TBBConfig,TBBConfigVersion}.cmake to Git ignore list
    (https://github.com/oneapi-src/oneTBB/pull/239) by Eisuke Kawashima.

------------------------------------------------------------------------
Intel TBB 2020 Update 2
TBB_INTERFACE_VERSION == 11102

Changes (w.r.t. Intel TBB 2020 Update 1):

- Cross-allocator copying constructor and copy assignment operator for
    concurrent_vector are deprecated.
- Added input_node to the flow graph API. It acts like a source_node
    except for being inactive by default; source_node is deprecated.
- Allocator template parameter for flow graph nodes is deprecated. Set
    TBB_DEPRECATED_FLOW_NODE_ALLOCATOR to 1 to avoid compilation errors.
- Flow graph preview hetero-features are deprecated.

Bugs fixed:

- Fixed the task affinity mechanism to prevent unlimited memory
    consumption in case the number of threads is explicitly decreased.
- Fixed memory leak related NUMA support functionality in task_arena.

------------------------------------------------------------------------
Intel TBB 2020 Update 1
TBB_INTERFACE_VERSION == 11101

Changes (w.r.t. Intel TBB 2020):

Preview features:

- The NUMA support library (tbbbind) no more depends on the main
    TBB library.

Bugs fixed:

- Fixed the issue of task_arena constraints not propagated on
    copy construction.
- Fixed TBBGet.cmake script broken by TBB package name changes
    (https://github.com/intel/tbb/issues/209).

------------------------------------------------------------------------
Intel TBB 2020
TBB_INTERFACE_VERSION == 11100

Changes (w.r.t. Intel TBB 2019 Update 9):

- Extended task_arena interface to simplify development of NUMA-aware
    applications.
- Added warning notifications when the deprecated functionality is used.

Open-source contributions integrated:

- Fixed various build warnings
    (https://github.com/intel/tbb/pull/179) by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 2019 Update 9
TBB_INTERFACE_VERSION == 11009

Changes (w.r.t. Intel TBB 2019 Update 8):

- Multiple APIs are deprecated. For details, please see
    Deprecated Features appendix in the TBB reference manual.
- Added C++17 deduction guides for flow graph nodes.

Preview Features:

- Added isolated_task_group class that allows multiple threads to add
    and execute tasks sharing the same isolation.
- Extended the flow graph API to simplify connecting nodes.
- Added erase() by heterogeneous keys for concurrent ordered containers.
- Added a possibility to suspend task execution at a specific point
    and resume it later.

Bugs fixed:

- Fixed the emplace() method of concurrent unordered containers to
    destroy a temporary element that was not inserted.
- Fixed a bug in the merge() method of concurrent unordered
    containers.
- Fixed behavior of a continue_node that follows buffering nodes.
- Fixed compilation error caused by missed stdlib.h when CMake
    integration is used (https://github.com/intel/tbb/issues/195).
    Inspired by Andrew Penkrat.

Open-source contributions integrated:

- Added support for move-only types to tbb::parallel_pipeline
    (https://github.com/intel/tbb/pull/159) by Raf Schietekat.
- Fixed detection of clang version when CUDA toolkit is installed
    (https://github.com/intel/tbb/pull/150) by Guilherme Amadio.

------------------------------------------------------------------------
Intel TBB 2019 Update 8
TBB_INTERFACE_VERSION == 11008

Changes (w.r.t. Intel TBB 2019 Update 7):

Bugs fixed:

- Fixed a bug in TBB 2019 Update 7 that could lead to incorrect memory
    reallocation on Linux (https://github.com/intel/tbb/issues/148).
- Fixed enqueuing tbb::task into tbb::task_arena not to fail on threads
    with no task scheduler initialized
    (https://github.com/intel/tbb/issues/116).

------------------------------------------------------------------------
Intel TBB 2019 Update 7
TBB_INTERFACE_VERSION == 11007

Changes (w.r.t. Intel TBB 2019 Update 6):

- Added TBBMALLOC_SET_HUGE_SIZE_THRESHOLD parameter to set the lower
    bound for allocations that are not released back to OS unless
    a cleanup is explicitly requested.
- Added zip_iterator::base() method to get the tuple of underlying
    iterators.
- Improved async_node to never block a thread that sends a message
    through its gateway.
- Extended decrement port of the tbb::flow::limiter_node to accept
    messages of integral types.
- Added support of Windows* to the CMake module TBBInstallConfig.
- Added packaging of CMake configuration files to TBB packages built
    using build/build.py script
    (https://github.com/intel/tbb/issues/141).

Changes affecting backward compatibility:

- Removed the number_of_decrement_predecessors parameter from the
    constructor of flow::limiter_node. To allow its usage, set
    TBB_DEPRECATED_LIMITER_NODE_CONSTRUCTOR macro to 1.

Preview Features:

- Added ordered associative containers:
    concurrent_{map,multimap,set,multiset} (requires C++11).

Open-source contributions integrated:

- Fixed makefiles to properly obtain the GCC version for GCC 7
    and later (https://github.com/intel/tbb/pull/147) by Timmmm.

------------------------------------------------------------------------
Intel TBB 2019 Update 6
TBB_INTERFACE_VERSION == 11006

Changes (w.r.t. Intel TBB 2019 Update 5):

- Added support for Microsoft* Visual Studio* 2019.
- Added support for enqueuing tbb::task into tbb::task_arena
    (https://github.com/01org/tbb/issues/116).
- Improved support for allocator propagation on concurrent_hash_map
    assigning and swapping.
- Improved scalable_allocation_command cleanup operations to release
    more memory buffered by the calling thread.
- Separated allocation of small and large objects into distinct memory
    regions, which helps to reduce excessive memory caching inside the
    TBB allocator.

Preview Features:

- Removed template class gfx_factory from the flow graph API.

------------------------------------------------------------------------
Intel TBB 2019 Update 5
TBB_INTERFACE_VERSION == 11005

Changes (w.r.t. Intel TBB 2019 Update 4):

- Associating a task_scheduler_observer with an implicit or explicit
    task arena is now a fully supported feature.
- Added a CMake module TBBInstallConfig that allows to generate and
    install CMake configuration files for TBB packages.
    Inspired by Hans Johnson (https://github.com/01org/tbb/pull/119).
- Added node handles, methods merge() and unsafe_extract() to concurrent
    unordered containers.
- Added constructors with Compare argument to concurrent_priority_queue
    (https://github.com/01org/tbb/issues/109).
- Controlling the stack size of worker threads is now supported for
    Universal Windows Platform.
- Improved tbb::zip_iterator to work with algorithms that swap values
    via iterators.
- Improved support for user-specified allocators in concurrent_hash_map,
    including construction of allocator-aware data types.
- For ReaderWriterMutex types, upgrades and downgrades now succeed if
    the mutex is already in the requested state.
    Inspired by Niadb (https://github.com/01org/tbb/pull/122).

Preview Features:

- The task_scheduler_observer::may_sleep() method has been removed.

Bugs fixed:

- Fixed the issue with a pipeline parallel filter executing serially if
    it follows a thread-bound filter.
- Fixed a performance regression observed when multiple parallel
    algorithms start simultaneously.

------------------------------------------------------------------------
Intel TBB 2019 Update 4
TBB_INTERFACE_VERSION == 11004

Changes (w.r.t. Intel TBB 2019 Update 3):

- global_control class is now a fully supported feature.
- Added deduction guides for tbb containers: concurrent_hash_map,
    concurrent_unordered_map, concurrent_unordered_set.
- Added tbb::scalable_memory_resource function returning
    std::pmr::memory_resource interface to the TBB memory allocator.
- Added tbb::cache_aligned_resource class that implements
    std::pmr::memory_resource with cache alignment and no false sharing.
- Added rml::pool_msize function returning the usable size of a memory
    block allocated from a given memory pool.
- Added default and copy constructors for tbb::counting_iterator
    and tbb::zip_iterator.
- Added TBB_malloc_replacement_log function to obtain the status of
    dynamic memory allocation replacement (Windows* only).
- CMake configuration file now supports release-only and debug-only
    configurations (https://github.com/01org/tbb/issues/113).
- TBBBuild CMake module takes the C++ version from CMAKE_CXX_STANDARD.

Bugs fixed:

- Fixed compilation for tbb::concurrent_vector when used with
    std::pmr::polymorphic_allocator.

Open-source contributions integrated:

- TBB_INTERFACE_VERSION is included into TBB version in CMake
    configuration (https://github.com/01org/tbb/pull/100)
    by Hans Johnson.
- Fixed detection of C++17 deduction guides for Visual C++*
    (https://github.com/01org/tbb/pull/112) by Marian Klymov.

------------------------------------------------------------------------
Intel TBB 2019 Update 3
TBB_INTERFACE_VERSION == 11003

Changes (w.r.t. Intel TBB 2019 Update 2):

- Added tbb::transform_iterator.
- Added new Makefile target 'profile' to flow graph examples enabling
    additional support for Intel(R) Parallel Studio XE tools.
- Added TBB_MALLOC_DISABLE_REPLACEMENT environment variable to switch off
    dynamic memory allocation replacement on Windows*. Inspired by
    a contribution from Edward Lam.

Preview Features:

- Extended flow graph API to support relative priorities for functional
    nodes, specified as an optional parameter to the node constructors.

Open-source contributions integrated:

- Enabled using process-local futex operations
    (https://github.com/01org/tbb/pull/58) by Andrey Semashev.

------------------------------------------------------------------------
Intel TBB 2019 Update 2
TBB_INTERFACE_VERSION == 11002

Changes (w.r.t. Intel TBB 2019 Update 1):

- Added overloads for parallel_reduce with default partitioner and
    user-supplied context.
- Added deduction guides for tbb containers: concurrent_vector,
    concurrent_queue, concurrent_bounded_queue,
    concurrent_priority_queue.
- Reallocation of memory objects >1MB now copies and frees memory if
    the size is decreased twice or more, trading performance off for
    reduced memory usage.
- After a period of sleep, TBB worker threads now prefer returning to
    their last used task arena.

Bugs fixed:

- Fixed compilation of task_group.h when targeting macOS* 10.11 or
    earlier (https://github.com/conda-forge/tbb-feedstock/issues/42).

Open-source contributions integrated:

- Added constructors with HashCompare argument to concurrent_hash_map
    (https://github.com/01org/tbb/pull/63) by arewedancer.

------------------------------------------------------------------------
Intel TBB 2019 Update 1
TBB_INTERFACE_VERSION == 11001

Changes (w.r.t. Intel TBB 2019):

- Doxygen documentation could be built with 'make doxygen' command now.

Changes affecting backward compatibility:

- Enforced 8 byte alignment for tbb::atomic<long long> and
    tbb::atomic<double>. On IA-32 architecture it may cause layout
    changes in structures that use these types.

Bugs fixed:

- Fixed an issue with dynamic memory allocation replacement on Windows*
    occurred for some versions of ucrtbase.dll.
- Fixed possible deadlock in tbbmalloc cleanup procedure during process
    shutdown. Inspired by a contribution from Edward Lam.
- Fixed usage of std::uncaught_exception() deprecated in C++17
   (https://github.com/01org/tbb/issues/67).
- Fixed a crash when a local observer is activated after an arena
    observer.
- Fixed compilation of task_group.h by Visual C++* 15.7 with
    /permissive- option (https://github.com/01org/tbb/issues/53).
- Fixed tbb4py to avoid dependency on Intel(R) C++ Compiler shared
    libraries.
- Fixed compilation for Anaconda environment with GCC 7.3 and higher.

Open-source contributions integrated:

- Fix various warnings when building with Visual C++
    (https://github.com/01org/tbb/pull/70) by Edward Lam.

------------------------------------------------------------------------
Intel TBB 2019
TBB_INTERFACE_VERSION == 11000

Changes (w.r.t. Intel TBB 2018 Update 5):

- Lightweight policy for functional nodes in the flow graph is now
    a fully supported feature.
- Reservation support in flow::write_once_node and flow::overwrite_node
    is now a fully supported feature.
- Support for Flow Graph Analyzer and improvements for
    Intel(R) VTune(TM) Amplifier become a regular feature enabled by
    TBB_USE_THREADING_TOOLS macro.
- Added support for std::new_handler in the replacement functions for
    global operator new.
- Added C++14 constructors to concurrent unordered containers.
- Added tbb::counting_iterator and tbb::zip_iterator.
- Fixed multiple -Wextra warnings in TBB source files.

Preview Features:

- Extracting nodes from a flow graph is deprecated and disabled by
    default. To enable, use TBB_DEPRECATED_FLOW_NODE_EXTRACTION macro.

Changes affecting backward compatibility:

- Due to internal changes in the flow graph classes, recompilation is
    recommended for all binaries that use the flow graph.

Open-source contributions integrated:

- Added support for OpenBSD by Anthony J. Bentley.

------------------------------------------------------------------------
Intel TBB 2018 Update 6
TBB_INTERFACE_VERSION == 10006

Changes (w.r.t. Intel TBB 2018 Update 5):

Bugs fixed:

- Fixed an issue with dynamic memory allocation replacement on Windows*
    occurred for some versions of ucrtbase.dll.

------------------------------------------------------------------------
Intel TBB 2018 Update 5
TBB_INTERFACE_VERSION == 10005

Changes (w.r.t. Intel TBB 2018 Update 4):

Preview Features:

- Added user event tracing API for Intel(R) VTune(TM) Amplifier and
    Flow Graph Analyzer.

Bugs fixed:

- Fixed the memory allocator to properly support transparent huge pages.
- Removed dynamic exception specifications in tbbmalloc_proxy for C++11
    and later (https://github.com/01org/tbb/issues/41).
- Added -flifetime-dse=1 option when building with GCC on macOS*
    (https://github.com/01org/tbb/issues/60).

Open-source contributions integrated:

- Added ARMv8 support by Siddhesh Poyarekar.
- Avoid GCC warnings for clearing an object of non-trivial type
    (https://github.com/01org/tbb/issues/54) by Daniel Arndt.

------------------------------------------------------------------------
Intel TBB 2018 Update 4
TBB_INTERFACE_VERSION == 10004

Changes (w.r.t. Intel TBB 2018 Update 3):

Preview Features:

- Improved support for Flow Graph Analyzer and Intel(R) VTune(TM)
    Amplifier in the task scheduler and generic parallel algorithms.
- Default device set for opencl_node now includes all the devices from
    the first available OpenCL* platform.
- Added lightweight policy for functional nodes in the flow graph. It
    indicates that the node body has little work and should, if possible,
    be executed immediately upon receiving a message, avoiding task
    scheduling overhead.

------------------------------------------------------------------------
Intel TBB 2018 Update 3
TBB_INTERFACE_VERSION == 10003

Changes (w.r.t. Intel TBB 2018 Update 2):

Preview Features:

- Added template class blocked_rangeNd for a generic multi-dimensional
    range (requires C++11). Inspired by a contribution from Jeff Hammond.

Bugs fixed:

- Fixed a crash with dynamic memory allocation replacement on
    Windows* for applications using system() function.
- Fixed parallel_deterministic_reduce to split range correctly when used
     with static_partitioner.
- Fixed a synchronization issue in task_group::run_and_wait() which
    caused a simultaneous call to task_group::wait() to return
    prematurely.

------------------------------------------------------------------------
Intel TBB 2018 Update 2
TBB_INTERFACE_VERSION == 10002

Changes (w.r.t. Intel TBB 2018 Update 1):

- Added support for Android* NDK r16, macOS* 10.13, Fedora* 26.
- Binaries for Universal Windows Driver (vc14_uwd) now link with static
    Microsoft* runtime libraries, and are only available in commercial
    releases.
- Extended flow graph documentation with more code samples.

Preview Features:

- Added a Python* module for multi-processing computations in numeric
    Python* libraries.

Bugs fixed:

- Fixed constructors of concurrent_hash_map to be exception-safe.
- Fixed auto-initialization in the main thread to be cleaned up at
    shutdown.
- Fixed a crash when tbbmalloc_proxy is used together with dbghelp.
- Fixed static_partitioner to assign tasks properly in case of nested
    parallelism.

------------------------------------------------------------------------
Intel TBB 2018 Update 1
TBB_INTERFACE_VERSION == 10001

Changes (w.r.t. Intel TBB 2018):

- Added lambda-friendly overloads for parallel_scan.
- Added support of static and simple partitioners in
    parallel_deterministic_reduce.

Preview Features:

- Added initial support for Flow Graph Analyzer to parallel_for.
- Added reservation support in overwrite_node and write_once_node.

Bugs fixed:

- Fixed a potential deadlock scenario in the flow graph that affected
    Intel TBB 2018.

------------------------------------------------------------------------
Intel TBB 2018
TBB_INTERFACE_VERSION == 10000

Changes (w.r.t. Intel TBB 2017 Update 7):

- Introduced Parallel STL, an implementation of the C++ standard
    library algorithms with support for execution policies. For more
    information, see Getting Started with Parallel STL
    (https://software.intel.com/en-us/get-started-with-pstl).
- this_task_arena::isolate() function is now a fully supported feature.
- this_task_arena::isolate() function and task_arena::execute() method
    were extended to pass on the value returned by the executed functor
    (requires C++11).
- task_arena::enqueue() and task_group::run() methods extended to accept
    move-only functors.
- A flow graph now spawns all tasks into the same task arena,
    and waiting for graph completion also happens in that arena.
- Improved support for Flow Graph Analyzer in async_node, opencl_node,
    and composite_node.
- Added support for Android* NDK r15, r15b.
- Added support for Universal Windows Platform.
- Increased minimally supported version of macOS*
    (MACOSX_DEPLOYMENT_TARGET) to 10.11.

Changes affecting backward compatibility:

- Internal layout changes in some flow graph classes;
- Several undocumented methods are removed from class graph,
    including set_active() and is_active().
- Due to incompatible changes, the namespace version is updated
    for the flow graph; recompilation is recommended for all
    binaries that use the flow graph classes.

Preview Features:

- opencl_node can be used with any graph object; class opencl_graph
    is removed.
- graph::wait_for_all() now automatically waits for all not yet consumed
    async_msg objects.
- Improved concurrent_lru_cache::handle_object to support C++11 move
    semantics, default construction, and conversion to bool.

Bugs fixed:

- Fixed a bug preventing use of streaming_node and opencl_node with
    Clang; inspired by a contribution from Francisco Facioni.
- Fixed this_task_arena::isolate() function to work correctly with
    parallel_invoke and parallel_do algorithms.
- Fixed a memory leak in composite_node.
- Fixed an assertion failure in debug tbbmalloc binaries when
    TBBMALLOC_CLEAN_ALL_BUFFERS is used.

------------------------------------------------------------------------
Intel TBB 2017 Update 8
TBB_INTERFACE_VERSION == 9108

Changes (w.r.t. Intel TBB 2017 Update 7):

Bugs fixed:

- Fixed an assertion failure in debug tbbmalloc binaries when
    TBBMALLOC_CLEAN_ALL_BUFFERS is used.

------------------------------------------------------------------------
Intel TBB 2017 Update 7
TBB_INTERFACE_VERSION == 9107

Changes (w.r.t. Intel TBB 2017 Update 6):

- In the huge pages mode, the memory allocator now is also able to use
    transparent huge pages.

Preview Features:

- Added support for Intel TBB integration into CMake-aware
    projects, with valuable guidance and feedback provided by Brad King
    (Kitware).

Bugs fixed:

- Fixed scalable_allocation_command(TBBMALLOC_CLEAN_ALL_BUFFERS, 0)
    to process memory left after exited threads.

------------------------------------------------------------------------
Intel TBB 2017 Update 6
TBB_INTERFACE_VERSION == 9106

Changes (w.r.t. Intel TBB 2017 Update 5):

- Added support for Android* NDK r14.

Preview Features:

- Added a blocking terminate extension to the task_scheduler_init class
    that allows an object to wait for termination of worker threads.

Bugs fixed:

- Fixed compilation and testing issues with MinGW (GCC 6).
- Fixed compilation with /std:c++latest option of VS 2017
    (https://github.com/01org/tbb/issues/13).

------------------------------------------------------------------------
Intel TBB 2017 Update 5
TBB_INTERFACE_VERSION == 9105

Changes (w.r.t. Intel TBB 2017 Update 4):

- Added support for Microsoft* Visual Studio* 2017.
- Added graph/matmult example to demonstrate support for compute offload
    to Intel(R) Graphics Technology in the flow graph API.
- The "compiler" build option now allows to specify a full path to the
    compiler.

Changes affecting backward compatibility:

- Constructors for many classes, including graph nodes, concurrent
    containers, thread-local containers, etc., are declared explicit and
    cannot be used for implicit conversions anymore.

Bugs fixed:

- Added a workaround for bug 16657 in the GNU C Library (glibc)
    affecting the debug version of tbb::mutex.
- Fixed a crash in pool_identify() called for an object allocated in
    another thread.

------------------------------------------------------------------------
Intel TBB 2017 Update 4
TBB_INTERFACE_VERSION == 9104

Changes (w.r.t. Intel TBB 2017 Update 3):

- Added support for C++11 move semantics in parallel_do.
- Added support for FreeBSD* 11.

Changes affecting backward compatibility:

- Minimal compiler versions required for support of C++11 move semantics
    raised to GCC 4.5, VS 2012, and Intel(R) C++ Compiler 14.0.

Bugs fixed:

- The workaround for crashes in the library compiled with GCC 6
    (-flifetime-dse=1) was extended to Windows*.

------------------------------------------------------------------------
Intel TBB 2017 Update 3
TBB_INTERFACE_VERSION == 9103

Changes (w.r.t. Intel TBB 2017 Update 2):

- Added support for Android* 7.0 and Android* NDK r13, r13b.

Preview Features:

- Added template class gfx_factory to the flow graph API. It implements
    the Factory concept for streaming_node to offload computations to
    Intel(R) processor graphics.

Bugs fixed:

- Fixed a possible deadlock caused by missed wakeup signals in
    task_arena::execute().

Open-source contributions integrated:

- A build fix for Linux* s390x platform by Jerry J.

------------------------------------------------------------------------
Intel TBB 2017 Update 2
TBB_INTERFACE_VERSION == 9102

Changes (w.r.t. Intel TBB 2017 Update 1):

- Removed the long-outdated support for Xbox* consoles.

Bugs fixed:

- Fixed the issue with task_arena::execute() not being processed when
    the calling thread cannot join the arena.
- Fixed dynamic memory allocation replacement failure on macOS* 10.12.

------------------------------------------------------------------------
Intel TBB 2017 Update 1
TBB_INTERFACE_VERSION == 9101

Changes (w.r.t. Intel TBB 2017):

Bugs fixed:

- Fixed dynamic memory allocation replacement failures on Windows* 10
    Anniversary Update.
- Fixed emplace() method of concurrent unordered containers to not
    require a copy constructor.

------------------------------------------------------------------------
Intel TBB 2017
TBB_INTERFACE_VERSION == 9100

Changes (w.r.t. Intel TBB 4.4 Update 5):

- static_partitioner class is now a fully supported feature.
- async_node class is now a fully supported feature.
- Improved dynamic memory allocation replacement on Windows* OS to skip
    DLLs for which replacement cannot be done, instead of aborting.
- Intel TBB no longer performs dynamic memory allocation replacement
    for Microsoft* Visual Studio* 2008.
- For 64-bit platforms, quadrupled the worst-case limit on the amount
    of memory the Intel TBB allocator can handle.
- Added TBB_USE_GLIBCXX_VERSION macro to specify the version of GNU
    libstdc++ when it cannot be properly recognized, e.g. when used
    with Clang on Linux* OS. Inspired by a contribution from David A.
- Added graph/stereo example to demonstrate tbb::flow::async_msg.
- Removed a few cases of excessive user data copying in the flow graph.
- Reworked split_node to eliminate unnecessary overheads.
- Added support for C++11 move semantics to the argument of
    tbb::parallel_do_feeder::add() method.
- Added C++11 move constructor and assignment operator to
    tbb::combinable template class.
- Added tbb::this_task_arena::max_concurrency() function and
    max_concurrency() method of class task_arena returning the maximal
    number of threads that can work inside an arena.
- Deprecated tbb::task_arena::current_thread_index() static method;
    use tbb::this_task_arena::current_thread_index() function instead.
- All examples for commercial version of library moved online:
    https://software.intel.com/en-us/product-code-samples. Examples are
    available as a standalone package or as a part of Intel(R) Parallel
    Studio XE or Intel(R) System Studio Online Samples packages.

Changes affecting backward compatibility:

- Renamed following methods and types in async_node class:
    Old                   New
    async_gateway_type => gateway_type
    async_gateway()    => gateway()
    async_try_put()    => try_put()
    async_reserve()    => reserve_wait()
    async_commit()     => release_wait()
- Internal layout of some flow graph nodes has changed; recompilation
    is recommended for all binaries that use the flow graph.

Preview Features:

- Added template class streaming_node to the flow graph API. It allows
    a flow graph to offload computations to other devices through
    streaming or offloading APIs.
- Template class opencl_node reimplemented as a specialization of
    streaming_node that works with OpenCL*.
- Added tbb::this_task_arena::isolate() function to isolate execution
    of a group of tasks or an algorithm from other tasks submitted
    to the scheduler.

Bugs fixed:

- Added a workaround for GCC bug #62258 in std::rethrow_exception()
    to prevent possible problems in case of exception propagation.
- Fixed parallel_scan to provide correct result if the initial value
    of an accumulator is not the operation identity value.
- Fixed a memory corruption in the memory allocator when it meets
    internal limits.
- Fixed the memory allocator on 64-bit platforms to align memory
    to 16 bytes by default for all allocations bigger than 8 bytes.
- As a workaround for crashes in the Intel TBB library compiled with
    GCC 6, added -flifetime-dse=1 to compilation options on Linux* OS.
- Fixed a race in the flow graph implementation.

Open-source contributions integrated:

- Enabling use of C++11 'override' keyword by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.4 Update 6
TBB_INTERFACE_VERSION == 9006

Changes (w.r.t. Intel TBB 4.4 Update 5):

- For 64-bit platforms, quadrupled the worst-case limit on the amount
    of memory the Intel TBB allocator can handle.

Bugs fixed:

- Fixed a memory corruption in the memory allocator when it meets
    internal limits.
- Fixed the memory allocator on 64-bit platforms to align memory
    to 16 bytes by default for all allocations bigger than 8 bytes.
- Fixed parallel_scan to provide correct result if the initial value
    of an accumulator is not the operation identity value.
- As a workaround for crashes in the Intel TBB library compiled with
    GCC 6, added -flifetime-dse=1 to compilation options on Linux* OS.

------------------------------------------------------------------------
Intel TBB 4.4 Update 5
TBB_INTERFACE_VERSION == 9005

Changes (w.r.t. Intel TBB 4.4 Update 4):

- Modified graph/fgbzip2 example to remove unnecessary data queuing.

Preview Features:

- Added a Python* module which is able to replace Python's thread pool
    class with the implementation based on Intel TBB task scheduler.

Bugs fixed:

- Fixed the implementation of 64-bit tbb::atomic for IA-32 architecture
    to work correctly with GCC 5.2 in C++11/14 mode.
- Fixed a possible crash when tasks with affinity (e.g. specified via
    affinity_partitioner) are used simultaneously with task priority
    changes.

------------------------------------------------------------------------
Intel TBB 4.4 Update 4
TBB_INTERFACE_VERSION == 9004

Changes (w.r.t. Intel TBB 4.4 Update 3):

- Removed a few cases of excessive user data copying in the flow graph.
- Improved robustness of concurrent_bounded_queue::abort() in case of
    simultaneous push and pop operations.

Preview Features:

- Added tbb::flow::async_msg, a special message type to support
    communications between the flow graph and external asynchronous
    activities.
- async_node modified to support use with C++03 compilers.

Bugs fixed:

- Fixed a bug in dynamic memory allocation replacement for Windows* OS.
- Fixed excessive memory consumption on Linux* OS caused by enabling
    zero-copy realloc.
- Fixed performance regression on Intel(R) Xeon Phi(tm) coprocessor with
    auto_partitioner.

------------------------------------------------------------------------
Intel TBB 4.4 Update 3
TBB_INTERFACE_VERSION == 9003

Changes (w.r.t. Intel TBB 4.4 Update 2):

- Modified parallel_sort to not require a default constructor for values
    and to use iter_swap() for value swapping.
- Added support for creating or initializing a task_arena instance that
    is connected to the arena currently used by the thread.
- graph/binpack example modified to use multifunction_node.
- For performance analysis, use Intel(R) VTune(TM) Amplifier XE 2015
    and higher; older versions are no longer supported.
- Improved support for compilation with disabled RTTI, by omitting its use
    in auxiliary code, such as assertions. However some functionality,
    particularly the flow graph, does not work if RTTI is disabled.
- The tachyon example for Android* can be built using Android Studio 1.5
    and higher with experimental Gradle plugin 0.4.0.

Preview Features:

- Added class opencl_subbufer that allows using OpenCL* sub-buffer
    objects with opencl_node.
- Class global_control supports the value of 1 for
    max_allowed_parallelism.

Bugs fixed:

- Fixed a race causing "TBB Warning: setaffinity syscall failed" message.
- Fixed a compilation issue on OS X* with Intel(R) C++ Compiler 15.0.
- Fixed a bug in queuing_rw_mutex::downgrade() that could temporarily
    block new readers.
- Fixed speculative_spin_rw_mutex to stop using the lazy subscription
    technique due to its known flaws.
- Fixed memory leaks in the tool support code.

------------------------------------------------------------------------
Intel TBB 4.4 Update 2
TBB_INTERFACE_VERSION == 9002

Changes (w.r.t. Intel TBB 4.4 Update 1):

- Improved interoperability with Intel(R) OpenMP RTL (libiomp) on Linux:
    OpenMP affinity settings do not affect the default number of threads
    used in the task scheduler. Intel(R) C++ Compiler 16.0 Update 1
    or later is required.
- Added a new flow graph example with different implementations of the
    Cholesky Factorization algorithm.

Preview Features:

- Added template class opencl_node to the flow graph API. It allows a
    flow graph to offload computations to OpenCL* devices.
- Extended join_node to use type-specified message keys. It simplifies
    the API of the node by obtaining message keys via functions
    associated with the message type (instead of node ports).
- Added static_partitioner that minimizes overhead of parallel_for and
    parallel_reduce for well-balanced workloads.
- Improved template class async_node in the flow graph API to support
    user settable concurrency limits.

Bugs fixed:

- Fixed a possible crash in the GUI layer for library examples on Linux.

------------------------------------------------------------------------
Intel TBB 4.4 Update 1
TBB_INTERFACE_VERSION == 9001

Changes (w.r.t. Intel TBB 4.4):

- Added support for Microsoft* Visual Studio* 2015.
- Intel TBB no longer performs dynamic replacement of memory allocation
    functions for Microsoft Visual Studio 2005 and earlier versions.
- For GCC 4.7 and higher, the intrinsics-based platform isolation layer
    uses __atomic_* built-ins instead of the legacy __sync_* ones.
    This change is inspired by a contribution from Mathieu Malaterre.
- Improvements in task_arena:
    Several application threads may join a task_arena and execute tasks
    simultaneously. The amount of concurrency reserved for application
    threads at task_arena construction can be set to any value between
    0 and the arena concurrency limit.
- The fractal example was modified to demonstrate class task_arena
    and moved to examples/task_arena/fractal.

Bugs fixed:

- Fixed a deadlock during destruction of task_scheduler_init objects
    when one of destructors is set to wait for worker threads.
- Added a workaround for a possible crash on OS X* when dynamic memory
    allocator replacement (libtbbmalloc_proxy) is used and memory is
    released during application startup.
- Usage of mutable functors with task_group::run_and_wait() and
    task_arena::enqueue() is disabled. An attempt to pass a functor
    which operator()() is not const will produce compilation errors.
- Makefiles and environment scripts now properly recognize GCC 5.0 and
    higher.

Open-source contributions integrated:

- Improved performance of parallel_for_each for inputs allowing random
    access, by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.4
TBB_INTERFACE_VERSION == 9000

Changes (w.r.t. Intel TBB 4.3 Update 6):

- The following features are now fully supported:
    tbb::flow::composite_node;
    additional policies of tbb::flow::graph_node::reset().
- Platform abstraction layer for Windows* OS updated to use compiler
    intrinsics for most atomic operations.
- The tbb/compat/thread header updated to automatically include
    C++11 <thread> where available.
- Fixes and refactoring in the task scheduler and class task_arena.
- Added key_matching policy to tbb::flow::join_node, which removes
    the restriction on the type that can be compared-against.
- For tag_matching join_node, tag_value is redefined to be 64 bits
    wide on all architectures.
- Expanded the documentation for the flow graph with details about
    node semantics and behavior.
- Added dynamic replacement of C11 standard function aligned_alloc()
    under Linux* OS.
- Added C++11 move constructors and assignment operators to
    tbb::enumerable_thread_specific container.
- Added hashing support for tbb::tbb_thread::id.
- On OS X*, binaries that depend on libstdc++ are not provided anymore.
    In the makefiles, libc++ is now used by default; for building with
    libstdc++, specify stdlib=libstdc++ in the make command line.

Preview Features:

- Added a new example, graph/fgbzip2, that shows usage of
    tbb::flow::async_node.
- Modification to the low-level API for memory pools:
    added a function for finding a memory pool by an object allocated
    from that pool.
- tbb::memory_pool now does not request memory till the first allocation
    from the pool.

Changes affecting backward compatibility:

- Internal layout of flow graph nodes has changed; recompilation is
    recommended for all binaries that use the flow graph.
- Resetting a tbb::flow::source_node will immediately activate it,
    unless it was created in inactive state.

Bugs fixed:

- Failure at creation of a memory pool will not cause process
    termination anymore.

Open-source contributions integrated:

- Supported building TBB with Clang on AArch64 with use of built-in
    intrinsics by David A.

------------------------------------------------------------------------
Intel TBB 4.3 Update 6
TBB_INTERFACE_VERSION == 8006

Changes (w.r.t. Intel TBB 4.3 Update 5):

- Supported zero-copy realloc for objects >1MB under Linux* via
    mremap system call.
- C++11 move-aware insert and emplace methods have been added to
    concurrent_hash_map container.
- install_name is set to @rpath/<library name> on OS X*.

Preview Features:

- Added template class async_node to the flow graph API. It allows a
    flow graph to communicate with an external activity managed by
    the user or another runtime.
- Improved speed of flow::graph::reset() clearing graph edges.
    rf_extract flag has been renamed rf_clear_edges.
- extract() method of graph nodes now takes no arguments.

Bugs fixed:

- concurrent_unordered_{set,map} behaves correctly for degenerate
    hashes.
- Fixed a race condition in the memory allocator that may lead to
    excessive memory consumption under high multithreading load.

------------------------------------------------------------------------
Intel TBB 4.3 Update 5
TBB_INTERFACE_VERSION == 8005

Changes (w.r.t. Intel TBB 4.3 Update 4):

- Added add_ref_count() method of class tbb::task.

Preview Features:

- Added class global_control for application-wide control of allowed
    parallelism and thread stack size.
- memory_pool_allocator now throws the std::bad_alloc exception on
    allocation failure.
- Exceptions thrown for by memory pool constructors changed from
    std::bad_alloc to std::invalid_argument and std::runtime_error.

Bugs fixed:

- scalable_allocator now throws the std::bad_alloc exception on
    allocation failure.
- Fixed a race condition in the memory allocator that may lead to
    excessive memory consumption under high multithreading load.
- A new scheduler created right after destruction of the previous one
    might be unable to modify the number of worker threads.

Open-source contributions integrated:

- (Added but not enabled) push_front() method of class tbb::task_list
    by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.3 Update 4
TBB_INTERFACE_VERSION == 8004

Changes (w.r.t. Intel TBB 4.3 Update 3):

- Added a C++11 variadic constructor for enumerable_thread_specific.
    The arguments from this constructor are used to construct
    thread-local values.
- Improved exception safety for enumerable_thread_specific.
- Added documentation for tbb::flow::tagged_msg class and
    tbb::flow::output_port function.
- Fixed build errors for systems that do not support dynamic linking.
- C++11 move-aware insert and emplace methods have been added to
    concurrent unordered containers.

Preview Features:

- Interface-breaking change: typedefs changed for node predecessor and
    successor lists, affecting copy_predecessors and copy_successors
    methods.
- Added template class composite_node to the flow graph API. It packages
    a subgraph to represent it as a first-class flow graph node.
- make_edge and remove_edge now accept multiport nodes as arguments,
    automatically using the node port with index 0 for an edge.

Open-source contributions integrated:

- Draft code for enumerable_thread_specific constructor with multiple
    arguments (see above) by Adrien Guinet.
- Fix for GCC invocation on IBM* Blue Gene*
    by Jeff Hammond and Raf Schietekat.
- Extended testing with smart pointers for Clang & libc++
    by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.3 Update 3
TBB_INTERFACE_VERSION == 8003

Changes (w.r.t. Intel TBB 4.3 Update 2):

- Move constructor and assignment operator were added to unique_lock.

Preview Features:

- Time overhead for memory pool destruction was reduced.

Open-source contributions integrated:

- Build error fix for iOS* by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.3 Update 2
TBB_INTERFACE_VERSION == 8002

Changes (w.r.t. Intel TBB 4.3 Update 1):

- Binary files for 64-bit Android* applications were added as part of the
    Linux* OS package.
- Exact exception propagation is enabled for Intel C++ Compiler on OS X*.
- concurrent_vector::shrink_to_fit was optimized for types that support
    C++11 move semantics.

Bugs fixed:

- Fixed concurrent unordered containers to insert elements much faster
    in debug mode.
- Fixed concurrent priority queue to support types that do not have
    copy constructors.
- Fixed enumerable_thread_specific to forbid copying from an instance
    with a different value type.

Open-source contributions integrated:

- Support for PathScale* EKOPath* Compiler by Erik Lindahl.

------------------------------------------------------------------------
Intel TBB 4.3 Update 1
TBB_INTERFACE_VERSION == 8001

Changes (w.r.t. Intel TBB 4.3):

- The ability to split blocked_ranges in a proportion, used by
    affinity_partitioner since version 4.2 Update 4, became a formal
    extension of the Range concept.
- More checks for an incorrect address to release added to the debug
    version of the memory allocator.
- Different kind of solutions for each TBB example were merged.

Preview Features:

- Task priorities are re-enabled in preview binaries.

Bugs fixed:

- Fixed a duplicate symbol when TBB_PREVIEW_VARIADIC_PARALLEL_INVOKE is
    used in multiple compilation units.
- Fixed a crash in __itt_fini_ittlib seen on Ubuntu 14.04.
- Fixed a crash in memory release after dynamic replacement of the
    OS X* memory allocator.
- Fixed incorrect indexing of arrays in seismic example.
- Fixed a data race in lazy initialization of task_arena.

Open-source contributions integrated:

- Fix for dumping information about gcc and clang compiler versions
    by Misty De Meo.

------------------------------------------------------------------------
Intel TBB 4.3
TBB_INTERFACE_VERSION == 8000

Changes (w.r.t. Intel TBB 4.2 Update 5):

- The following features are now fully supported: flow::indexer_node,
    task_arena, speculative_spin_rw_mutex.
- Compatibility with C++11 standard improved for tbb/compat/thread
    and tbb::mutex.
- C++11 move constructors have been added to concurrent_queue and
    concurrent_bounded_queue.
- C++11 move constructors and assignment operators have been added to
    concurrent_vector, concurrent_hash_map, concurrent_priority_queue,
    concurrent_unordered_{set,multiset,map,multimap}.
- C++11 move-aware emplace/push/pop methods have been added to
    concurrent_vector, concurrent_queue, concurrent_bounded_queue,
    concurrent_priority_queue.
- Methods to insert a C++11 initializer list have been added:
    concurrent_vector::grow_by(), concurrent_hash_map::insert(),
    concurrent_unordered_{set,multiset,map,multimap}::insert().
- Testing for compatibility of containers with some C++11 standard
    library types has been added.
- Dynamic replacement of standard memory allocation routines has been
    added for OS X*.
- Microsoft* Visual Studio* projects for Intel TBB examples updated
    to VS 2010.
- For open-source packages, debugging information (line numbers) in
    precompiled binaries now matches the source code.
- Debug information was added to release builds for OS X*, Solaris*,
    FreeBSD* operating systems and MinGW*.
- Various improvements in documentation, debug diagnostics and examples.

Preview Features:

- Additional actions on reset of graphs, and extraction of individual
    nodes from a graph (TBB_PREVIEW_FLOW_GRAPH_FEATURES).
- Support for an arbitrary number of arguments in parallel_invoke
   (TBB_PREVIEW_VARIADIC_PARALLEL_INVOKE).

Changes affecting backward compatibility:

- For compatibility with C++11 standard, copy and move constructors and
    assignment operators are disabled for all mutex classes. To allow
    the old behavior, use TBB_DEPRECATED_MUTEX_COPYING macro.
- flow::sequencer_node rejects messages with repeating sequence numbers.
- Changed internal interface between tbbmalloc and tbbmalloc_proxy.
- Following deprecated functionality has been removed:
    old debugging macros TBB_DO_ASSERT & TBB_DO_THREADING_TOOLS;
    no-op depth-related methods in class task;
    tbb::deprecated::concurrent_queue;
    deprecated variants of concurrent_vector methods.
- register_successor() and remove_successor() are deprecated as methods
    to add and remove edges in flow::graph; use make_edge() and
    remove_edge() instead.

Bugs fixed:

- Fixed incorrect scalable_msize() implementation for aligned objects.
- Flow graph buffering nodes now destroy their copy of forwarded items.
- Multiple fixes in task_arena implementation, including for:
    inconsistent task scheduler state inside executed functions;
    incorrect floating-point settings and exception propagation;
    possible stalls in concurrent invocations of execute().
- Fixed floating-point settings propagation when the same instance of
    task_group_context is used in different arenas.
- Fixed compilation error in pipeline.h with Intel Compiler on OS X*.
- Added missed headers for individual components to tbb.h.

Open-source contributions integrated:

- Range interface addition to parallel_do, parallel_for_each and
    parallel_sort by Stephan Dollberg.
- Variadic template implementation of parallel_invoke
    by Kizza George Mbidde (see Preview Features).
- Improvement in Seismic example for MacBook Pro* with Retina* display
    by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.2 Update 5
TBB_INTERFACE_VERSION == 7005

Changes (w.r.t. Intel TBB 4.2 Update 4):

- The second template argument of class aligned_space<T,N> now is set
    to 1 by default.

Preview Features:

- Better support for exception safety, task priorities and floating
    point settings in class task_arena.
- task_arena::current_slot() has been renamed to
    task_arena::current_thread_index().

Bugs fixed:

- Task priority change possibly ignored by a worker thread entering
    a nested parallel construct.
- Memory leaks inside the task scheduler when running on
    Intel(R) Xeon Phi(tm) coprocessor.

Open-source contributions integrated:

- Improved detection of X Window support for Intel TBB examples
    and other feedback by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.2 Update 4
TBB_INTERFACE_VERSION == 7004

Changes (w.r.t. Intel TBB 4.2 Update 3):

- Added possibility to specify floating-point settings at invocation
    of most parallel algorithms (including flow::graph) via
    task_group_context.
- Added dynamic replacement of malloc_usable_size() under
    Linux*/Android* and dlmalloc_usable_size() under Android*.
- Added new methods to concurrent_vector:
    grow_by() that appends a sequence between two given iterators;
    grow_to_at_least() that initializes new elements with a given value.
- Improved affinity_partitioner for better performance on balanced
    workloads.
- Improvements in the task scheduler, including better scalability
    when threads search for a task arena, and better diagnostics.
- Improved allocation performance for workloads that do intensive
    allocation/releasing of same-size objects larger than ~8KB from
    multiple threads.
- Exception support is enabled by default for 32-bit MinGW compilers.
- The tachyon example for Android* can be built for all targets
    supported by the installed NDK.
- Added Windows Store* version of the tachyon example.
- GettingStarted/sub_string_finder example ported to offload execution
    on Windows* for Intel(R) Many Integrated Core Architecture.

Preview Features:

- Removed task_scheduler_observer::on_scheduler_leaving() callback.
- Added task_scheduler_observer::may_sleep() callback.
- The CPF or_node has been renamed indexer_node. The input to
    indexer_node is now a list of types. The output of indexer_node is
    a tagged_msg type composed of a tag and a value. For indexer_node,
    the tag is a size_t.

Bugs fixed:

- Fixed data races in preview extensions of task_scheduler_observer.
- Added noexcept(false) for destructor of task_group_base to avoid
    crash on cancellation of structured task group in C++11.

Open-source contributions integrated:

- Improved concurrency detection for BG/Q, and other improvements
    by Raf Schietekat.
- Fix for crashes in enumerable_thread_specific in case if a contained
    object is too big to be constructed on the stack by Adrien Guinet.

------------------------------------------------------------------------
Intel TBB 4.2 Update 3
TBB_INTERFACE_VERSION == 7003

Changes (w.r.t. Intel TBB 4.2 Update 2):

- Added support for Microsoft* Visual Studio* 2013.
- Improved Microsoft* PPL-compatible form of parallel_for for better
    support of auto-vectorization.
- Added a new example for cancellation and reset in the flow graph:
    Kohonen self-organizing map (examples/graph/som).
- Various improvements in source code, tests, and makefiles.

Bugs fixed:

- Added dynamic replacement of _aligned_msize() previously missed.
- Fixed task_group::run_and_wait() to throw invalid_multiple_scheduling
    exception if the specified task handle is already scheduled.

Open-source contributions integrated:

- A fix for ARM* processors by Steve Capper.
- Improvements in std::swap calls by Robert Maynard.

------------------------------------------------------------------------
Intel TBB 4.2 Update 2
TBB_INTERFACE_VERSION == 7002

Changes (w.r.t. Intel TBB 4.2 Update 1):

- Enable C++11 features for Microsoft* Visual Studio* 2013 Preview.
- Added a test for compatibility of TBB containers with C++11
    range-based for loop.

Changes affecting backward compatibility:

- Internal layout changed for class tbb::flow::limiter_node.

Preview Features:

- Added speculative_spin_rw_mutex, a read-write lock class which uses
    Intel(R) Transactional Synchronization Extensions.

Bugs fixed:

- When building for Intel(R) Xeon Phi(tm) coprocessor, TBB programs
    no longer require explicit linking with librt and libpthread.

Open-source contributions integrated:

- Fixes for ARM* processors by Steve Capper, Leif Lindholm
    and Steven Noonan.
- Support for Clang on Linux by Raf Schietekat.
- Typo correction in scheduler.cpp by Julien Schueller.

------------------------------------------------------------------------
Intel TBB 4.2 Update 1
TBB_INTERFACE_VERSION == 7001

Changes (w.r.t. Intel TBB 4.2):

- Added project files for Microsoft* Visual Studio* 2010.
- Initial support of Microsoft* Visual Studio* 2013 Preview.
- Enable C++11 features available in Intel(R) C++ Compiler 14.0.
- scalable_allocation_mode(TBBMALLOC_SET_SOFT_HEAP_LIMIT, <size>) can be
    used to urge releasing memory from tbbmalloc internal buffers when
    the given limit is exceeded.

Preview Features:

- Class task_arena no longer requires linking with a preview library,
    though still remains a community preview feature.
- The method task_arena::wait_until_empty() is removed.
- The method task_arena::current_slot() now returns -1 if
    the task scheduler is not initialized in the thread.

Changes affecting backward compatibility:

- Because of changes in internal layout of graph nodes, the namespace
    interface number of flow::graph has been incremented from 6 to 7.

Bugs fixed:

- Fixed a race in lazy initialization of task_arena.
- Fixed flow::graph::reset() to prevent situations where tasks would be
    spawned in the process of resetting the graph to its initial state.
- Fixed decrement bug in limiter_node.
- Fixed a race in arc deletion in the flow graph.

Open-source contributions integrated:

- Improved support for IBM* Blue Gene* by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.2
TBB_INTERFACE_VERSION == 7000

Changes (w.r.t. Intel TBB 4.1 Update 4):

- Added speculative_spin_mutex, which uses Intel(R) Transactional
    Synchronization Extensions when they are supported by hardware.
- Binary files linked with libc++ (the C++ standard library in Clang)
    were added on OS X*.
- For OS X* exact exception propagation is supported with Clang;
    it requires use of libc++ and corresponding Intel TBB binaries.
- Support for C++11 initializer lists in constructor and assignment
    has been added to concurrent_hash_map, concurrent_unordered_set,
    concurrent_unordered_multiset, concurrent_unordered_map,
    concurrent_unordered_multimap.
- The memory allocator may now clean its per-thread memory caches
    when it cannot get more memory.
- Added the scalable_allocation_command() function for on-demand
    cleaning of internal memory caches.
- Reduced the time overhead for freeing memory objects smaller than ~8K.
- Simplified linking with the debug library for applications that use
    Intel TBB in code offloaded to Intel(R) Xeon Phi(tm) coprocessors.
    See an example in
    examples/GettingStarted/sub_string_finder/Makefile.
- Various improvements in source code, scripts and makefiles.

Changes affecting backward compatibility:

- tbb::flow::graph has been modified to spawn its tasks;
    the old behaviour (task enqueuing) is deprecated. This change may
    impact applications that expected a flow graph to make progress
    without calling wait_for_all(), which is no longer guaranteed. See
    the documentation for more details.
- Changed the return values of the scalable_allocation_mode() function.

Bugs fixed:

- Fixed a leak of parallel_reduce body objects when execution is
    cancelled or an exception is thrown, as suggested by Darcy Harrison.
- Fixed a race in the task scheduler which can lower the effective
    priority despite the existence of higher priority tasks.
- On Linux an error during destruction of the internal thread local
    storage no longer results in an exception.

Open-source contributions integrated:

- Fixed task_group_context state propagation to unrelated context trees
    by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.1 Update 4
TBB_INTERFACE_VERSION == 6105

Changes (w.r.t. Intel TBB 4.1 Update 3):

- Use /volatile:iso option with VS 2012 to disable extended
    semantics for volatile variables.
- Various improvements in affinity_partitioner, scheduler,
    tests, examples, makefiles.
- Concurrent_priority_queue class now supports initialization/assignment
    via C++11 initializer list feature (std::initializer_list<T>).

Bugs fixed:

- Fixed more possible stalls in concurrent invocations of
    task_arena::execute(), especially waiting for enqueued tasks.
- Fixed requested number of workers for task_arena(P,0).
- Fixed interoperability with Intel(R) VTune(TM) Amplifier XE in
    case of using task_arena::enqueue() from a terminating thread.

Open-source contributions integrated:

- Type fixes, cleanups, and code beautification by Raf Schietekat.
- Improvements in atomic operations for big endian platforms
    by Raf Schietekat.

------------------------------------------------------------------------
Intel TBB 4.1 Update 3
TBB_INTERFACE_VERSION == 6103

Changes (w.r.t. Intel TBB 4.1 Update 2):

- Binary files for Android* applications were added to the Linux* OS
    package.
- Binary files for Windows Store* applications were added to the
    Windows* OS package.
- Exact exception propagation (exception_ptr) support on Linux OS is
    now turned on by default for GCC 4.4 and higher.
- Stopped implicit use of large memory pages by tbbmalloc (Linux-only).
    Now use of large pages must be explicitly enabled with
    scalable_allocation_mode() function or TBB_MALLOC_USE_HUGE_PAGES
    environment variable.

Community Preview Features:

- Extended class task_arena constructor and method initialize() to
    allow some concurrency to be reserved strictly for application
    threads.
- New methods terminate() and is_active() were added to class
    task_arena.

Bugs fixed:

- Fixed initialization of hashing helper constant in the hash
    containers.
- Fixed possible stalls in concurrent invocations of
    task_arena::execute() when no worker thread is available to make
    progress.
- Fixed incorrect calculation of hardware concurrency in the presence
    of inactive processor groups, particularly on systems running
    Windows* 8 and Windows* Server 2012.

Open-source contributions integrated:

- The fix for the GUI examples on OS X* systems by Raf Schietekat.
- Moved some power-of-2 calculations to functions to improve readability
    by Raf Schietekat.
- C++11/Clang support improvements by arcata.
- ARM* platform isolation layer by Steve Capper, Leif Lindholm, Leo Lara
    (ARM).

------------------------------------------------------------------------
Intel TBB 4.1 Update 2
TBB_INTERFACE_VERSION == 6102

Changes (w.r.t. Intel TBB 4.1 Update 1):

- Objects up to 128 MB are now cached by the tbbmalloc. Previously
    the threshold was 8MB. Objects larger than 128 MB are still
    processed by direct OS calls.
- concurrent_unordered_multiset and concurrent_unordered_multimap
    have been added, based on Microsoft* PPL prototype.
- Ability to value-initialize a tbb::atomic<T> variable on construction
    in C++11, with const expressions properly supported.

Community Preview Features:

- Added a possibility to wait until all worker threads terminate.
    This is necessary before calling fork() from an application.

Bugs fixed:

- Fixed data race in tbbmalloc that might lead to memory leaks
    for large object allocations.
- Fixed task_arena::enqueue() to use task_group_context of target arena.
- Improved implementation of 64 bit atomics on ia32.

------------------------------------------------------------------------
Intel TBB 4.1 Update 1
TBB_INTERFACE_VERSION == 6101

Changes (w.r.t. Intel TBB 4.1):

- concurrent_vector class now supports initialization/assignment
    via C++11 initializer list feature (std::initializer_list<T>)
- Added implementation of the platform isolation layer based on
    Intel compiler atomic built-ins; it is supposed to work on
    any platform supported by compiler version 12.1 and newer.
- Using GetNativeSystemInfo() instead of GetSystemInfo() to support
    more than 32 processors for 32-bit applications under WOW64.
- The following form of parallel_for:
    parallel_for(first, last, [step,] f[, context]) now accepts an
    optional partitioner parameter after the function f.

Backward-incompatible API changes:

- The library no longer injects tuple in to namespace std.
    In previous releases, tuple was injected into namespace std by
    flow_graph.h when std::tuple was not available.  In this release,
    flow_graph.h now uses tbb::flow::tuple.  On platforms where
    std::tuple is available, tbb::flow::tuple is typedef'ed to
    std::tuple.  On all other platforms, tbb::flow::tuple provides
    a subset of the functionality defined by std::tuple. Users of
    flow_graph.h may need to change their uses of std::tuple to
    tbb::flow::tuple to ensure compatibility with non-C++11 compliant
    compilers.

Bugs fixed:

- Fixed local observer to be able to override propagated CPU state and
    to provide correct value of task_arena::current_slot() in callbacks.

------------------------------------------------------------------------
Intel TBB 4.1
TBB_INTERFACE_VERSION == 6100

Changes (w.r.t. Intel TBB 4.0 Update 5):

- _WIN32_WINNT must be set to 0x0501 or greater in order to use TBB
    on Microsoft* Windows*.
- parallel_deterministic_reduce template function is fully supported.
- TBB headers can be used with C++0x/C++11 mode (-std=c++0x) of GCC
    and Intel(R) Compiler.
- C++11 std::make_exception_ptr is used where available, instead of
    std::copy_exception from earlier C++0x implementations.
- Improvements in the TBB allocator to reduce extra memory consumption.
- Partial refactoring of the task scheduler data structures.
- TBB examples allow more flexible specification of the thread number,
    including arithmetic and geometric progression.

Bugs fixed:

- On Linux & OS X*, pre-built TBB binaries do not yet support exact
    exception propagation via C++11 exception_ptr. To prevent run time
    errors, by default TBB headers disable exact exception propagation
    even if the C++ implementation provides exception_ptr.

Community Preview Features:

- Added: class task_arena, for work submission by multiple application
    threads with thread-independent control of concurrency level.
- Added: task_scheduler_observer can be created as local to a master
    thread, to observe threads that work on behalf of that master.
    Local observers may have new on_scheduler_leaving() callback.

------------------------------------------------------------------------
Intel TBB 4.0 Update 5
TBB_INTERFACE_VERSION == 6005

Changes (w.r.t. Intel TBB 4.0 Update 4):

- Parallel pipeline optimization (directly storing small objects in the
    interstage data buffers) limited to trivially-copyable types for
    C++11 and a short list of types for earlier compilers.
- _VARIADIC_MAX switch is honored for TBB tuple implementation
    and flow::graph nodes based on tuple.
- Support of Cocoa framework was added to the GUI examples on OS X*
    systems.

Bugs fixed:

- Fixed a tv_nsec overflow bug in condition_variable::wait_for.
- Fixed execution order of enqueued tasks with different priorities.
- Fixed a bug with task priority changes causing lack of progress
    for fire-and-forget tasks when TBB was initialized to use 1 thread.
- Fixed duplicate symbol problem when linking multiple compilation
    units that include flow_graph.h on VC 10.

------------------------------------------------------------------------
Intel TBB 4.0 Update 4
TBB_INTERFACE_VERSION == 6004

Changes (w.r.t. Intel TBB 4.0 Update 3):

- The TBB memory allocator transparently supports large pages on Linux.
- A new flow_graph example, logic_sim, was added.
- Support for DirectX* 9 was added to GUI examples.

Community Preview Features:

- Added: aggregator, a new concurrency control mechanism.

Bugs fixed:

- The abort operation on concurrent_bounded_queue now leaves the queue
    in a reusable state. If a bad_alloc or bad_last_alloc exception is
    thrown while the queue is recovering from an abort, that exception
    will be reported instead of user_abort on the thread on which it
    occurred, and the queue will not be reusable.
- Steal limiting heuristic fixed to avoid premature stealing disabling
    when large amount of __thread data is allocated on thread stack.
- Fixed a low-probability leak of arenas in the task scheduler.
- In STL-compatible allocator classes, the method construct() was fixed
    to comply with C++11 requirements.
- Fixed a bug that prevented creation of fixed-size memory pools
    smaller than 2M.
- Significantly reduced the amount of warnings from various compilers.

Open-source contributions integrated:

- Multiple improvements by Raf Schietekat.
- Basic support for Clang on OS X* by Blas Rodriguez Somoza.
- Fixes for warnings and corner-case bugs by Blas Rodriguez Somoza
    and Edward Lam.

------------------------------------------------------------------------
Intel TBB 4.0 Update 3
TBB_INTERFACE_VERSION == 6003

Changes (w.r.t. Intel TBB 4.0 Update 2):

- Modifications to the low-level API for memory pools:
    added support for aligned allocations;
    pool policies reworked to allow backward-compatible extensions;
    added a policy to not return memory space till destruction;
    pool_reset() does not return memory space anymore.
- Class tbb::flow::graph_iterator added to iterate over all nodes
    registered with a graph instance.
- multioutput_function_node has been renamed multifunction_node.
    multifunction_node and split_node are now fully-supported features.
- For the tagged join node, the policy for try_put of an item with
    already existing tag has been defined: the item will be rejected.
- Matching the behavior on Windows, on other platforms the optional
    shared libraries (libtbbmalloc, libirml) now are also searched
    only in the directory where libtbb is located.
- The platform isolation layer based on GCC built-ins is extended.

Backward-incompatible API changes:

- a graph reference parameter is now required to be passed to the
    constructors of the following flow graph nodes: overwrite_node,
    write_once_node, broadcast_node, and the CPF or_node.
- the following tbb::flow node methods and typedefs have been renamed:
       Old                             New
    join_node and or_node:
       inputs()                 ->     input_ports()
       input_ports_tuple_type   ->     input_ports_type
    multifunction_node and split_node:
       ports_type               ->     output_ports_type

Bugs fixed:

- Not all logical processors were utilized on systems with more than
    64 cores split by Windows into several processor groups.

------------------------------------------------------------------------
Intel TBB 4.0 Update 2 commercial-aligned release
TBB_INTERFACE_VERSION == 6002

Changes (w.r.t. Intel TBB 4.0 Update 1 commercial-aligned release):

- concurrent_bounded_queue now has an abort() operation that releases
    threads involved in pending push or pop operations. The released
    threads will receive a tbb::user_abort exception.
- Added Community Preview Feature:  concurrent_lru_cache container,
    a concurrent implementation of LRU (least-recently-used) cache.

Bugs fixed:

- fixed a race condition in the TBB scalable allocator.
- concurrent_queue counter wraparound bug was fixed, which occurred when
    the number of push and pop operations exceeded ~>4 billion on IA32.
- fixed races in the TBB scheduler that could put workers asleep too
    early, especially in presence of affinitized tasks.

------------------------------------------------------------------------
Intel TBB 4.0 Update 1 commercial-aligned release
TBB_INTERFACE_VERSION == 6000 (forgotten to increment)

Changes (w.r.t. Intel TBB 4.0 commercial-aligned release):

- Memory leaks fixed in binpack example.
- Improvements and fixes in the TBB allocator.

------------------------------------------------------------------------
Intel TBB 4.0 commercial-aligned release
TBB_INTERFACE_VERSION == 6000

Changes (w.r.t. Intel TBB 3.0 Update 8 commercial-aligned release):

- concurrent_priority_queue is now a fully supported feature.
    Capacity control methods were removed.
- Flow graph is now a fully supported feature.
- A new memory backend has been implemented in the TBB allocator.
    It can reuse freed memory for both small and large objects, and
    returns unused memory blocks to the OS more actively.
- Improved partitioning algorithms for parallel_for and parallel_reduce
    to better handle load imbalance.
- The convex_hull example has been refactored for reproducible
    performance results.
- The major interface version has changed from 5 to 6.
    Deprecated interfaces might be removed in future releases.

Community Preview Features:

- Added: serial subset, i.e. sequential implementations of TBB generic
    algorithms (currently, only provided for parallel_for).
- Preview of new flow graph nodes:
    or_node (accepts multiple inputs, forwards each input separately
      to all successors),
    split_node (accepts tuples, and forwards each element of a tuple
      to a corresponding successor), and
    multioutput_function_node (accepts one input, and passes the input
    and a tuple of output ports to the function body to support outputs
    to multiple successors).
- Added: memory pools for more control on memory source, grouping,
    and collective deallocation.

------------------------------------------------------------------------
Intel TBB 3.0 Update 8 commercial-aligned release
TBB_INTERFACE_VERSION == 5008

Changes (w.r.t. Intel TBB 3.0 Update 7 commercial-aligned release):

- Task priorities become an official feature of TBB,
    not community preview as before.
- Atomics API extended, and implementation refactored.
- Added task::set_parent() method.
- Added concurrent_unordered_set container.

Open-source contributions integrated:

- PowerPC support by Raf Schietekat.
- Fix of potential task pool overrun and other improvements
    in the task scheduler by Raf Schietekat.
- Fix in parallel_for_each to work with std::set in Visual* C++ 2010.

Community Preview Features:

- Graph community preview feature was renamed to flow graph.
    Multiple improvements in the implementation.
    Binpack example was added for the feature.
- A number of improvements to concurrent_priority_queue.
    Shortpath example was added for the feature.
- TBB runtime loaded functionality was added (Windows*-only).
    It allows to specify which versions of TBB should be used,
    as well as to set directories for the library search.
- parallel_deterministic_reduce template function was added.

------------------------------------------------------------------------
Intel TBB 3.0 Update 7 commercial-aligned release
TBB_INTERFACE_VERSION == 5006 (forgotten to increment)

Changes (w.r.t. Intel TBB 3.0 Update 6 commercial-aligned release):

- Added implementation of the platform isolation layer based on
    GCC atomic built-ins; it is supposed to work on any platform
    where GCC has these built-ins.

Community Preview Features:

- Graph's dining_philosophers example added.
- A number of improvements to graph and concurrent_priority_queue.

Files:
RevisionActionfile
1.9modifypkgsrc/parallel/threadingbuildingblocks/Makefile
1.2modifypkgsrc/parallel/threadingbuildingblocks/PLIST
1.2modifypkgsrc/parallel/threadingbuildingblocks/buildlink3.mk
1.4modifypkgsrc/parallel/threadingbuildingblocks/distinfo
1.1removepkgsrc/parallel/threadingbuildingblocks/patches/patch-build_FreeBSD.gcc.inc