TBB #06: C++ Optimized Memory Allocators and Thread-Safe Concurrent Containerss

How to build TBB enabled optimized C++ program using tbb allocator, scalable allocator, cache aligned allocator, and how to use thread-safe containers such as concurrent_vector.

We will also learn how to build debugging build for TBB enabled application with tbb.lib, tbbmalloc.lib, -ltbb -ltbbmalloc, etc.

For release build with optimization switch -O3
clang++ -std=c++17 -O3 alloc.cpp -ltbb -ltbbmalloc -o c.exe
g++ -std=c++17 -O3 alloc.cpp -ltbb -ltbbmalloc -o g.exe

For debugging build without optimization, debugging switch -g
clang++ -std=c++17 -O0 -g -DTBB_USE_DEBUG=1 alloc.cpp -ltbb_debug -ltbbmalloc_debug -o c.exe
g++ -std=c++17 -O0 -g -DTBB_USE_DEBUG=1 alloc.cpp -ltbb_debug -ltbbmalloc_debug -o g.exe

For Microsoft compiler optimized release build
cl /EHsc /std:c++17 /MD /O2 /DTBB_SUPPRESS_DEPRECATED_MESSAGES tbbmalloc.lib alloc.cpp /Fe: m.exe

For Microsoft Compiler for debugging build
cl /EHsc /std:c++17 /MDd /D_DEBUG /DTBB_SUPPRESS_DEPRECATED_MESSAGES tbbmalloc_debug.lib alloc.cpp /Fe: m.exe

Download Source Code:
http://www.TalkPlayFun.com/download/tbb_linear_algebra/06-Allocators.zip

References:

Threading Building Blocks (TBB)
https://www.threadingbuildingblocks.org/intel-tbb-tutorial

Documentation, Training, & Code Samples
https://software.intel.com/en-us/tbb/documentation#training

Allocator Concept
https://software.intel.com/en-us/node/506256

tbb::tbb_allocator - https://software.intel.com/en-us/node/506257
Template class for scalable memory allocation if available; possibly non-scalable otherwise.

tbb::scalable_allocator - https://software.intel.com/en-us/node/506258
A scalable_allocator allocates and frees memory in a way that scales with the number of processors. A scalable_allocator models the Allocator Concept. Using a scalable_allocator in place of std::allocator may improve program performance.

tbb::cache_aligned_allocator - https://software.intel.com/en-us/node/506260
Template class for allocating memory in a way that avoids false sharing. A cache_aligned_allocator allocates memory on cache line boundaries, in order to avoid false sharing.

tbb::zero_allocator - https://software.intel.com/en-us/node/506261
A zero_allocator allocates zeroed memory. A zero_allocator can be instantiated for any class A that models the Allocator concept. The default for A is tbb_allocator. A zero_allocator forwards allocation requests to A and zeros the allocation before returning it.

Containers - https://software.intel.com/en-us/node/506169

tbb::concurrent_vector - https://software.intel.com/en-us/node/506203
Template class for a vector that can be concurrently grown and accessed.

TBB #01: Installing C++ Compilers and Environment Settings
https://www.youtube.com/watch?v=W8t3yJC05lk&list=PL1_C6uWTeBDFvAj45SFsdM98mESge4zm_&index=2