Fast Memory Block Operations

Fast Memory Block Operations

The slow way to do things in arrays is one element at a time. The faster way is to use the standard memory block functions on the whole array. There are a number of standard functions that operate on array data or memory blocks and they are very fast. Memory block operations in the standard C++ libraries are implemented using fast assembly language behind the scenes. The main functions in the standard C++ library that operate on binary bytes in a memory block are:

• memset — set bytes to a value, or clear bytes to zero.
• memcpy — copy bytes in a block (non-overlapping).
• memmove — copy bytes, allowing overlapping blocks.
• memcmp — compare two blocks of bytes.
• memchr — search for a byte in a block.

Note that unlike the standard string functions (such as strlen), these functions do not assume a block is null-terminated by a zero byte. Zero is simply a binary value, and these functions don't stop at a zero byte. All of these functions operate on a block of memory with a fixed byte length.

Each compiler environment typically offers some extra non-standard byte-wise functions that are also fast. Some of the less standardized C++ intrinsics that operate on memory blocks include:

• _memccpy — copy bytes up to a specified sentinel byte.
• memicmp / _memicmp — compare bytes ignoring letter case.
• bcopy — copy a block of bytes.
• bzero — clear a block of bytes to zero.
• bcmp — compare bytes in two blocks.
• _byteswap_uint64 — swap the bytes of an integer (Microsoft).
• __builtin_bswap16 — swap the bytes in an integer (GCC). There are versions for 32-bit and 64-bit.

• Next: • Up: Table of Contents • Buy: Generative AI in C++: Coding Transformers and LLMs

The new AI programming book by Aussie AI co-founders: AI coding in C++ Transformer engine speedups LLM models Phone and desktop AI Code examples Research citations Get your copy from Amazon: Generative AI in C++