Inside the Intel Compiler
While the Intel compiler is strictly ANSI-compliant, there are options to cover many GCC extensions, such as long long int, zero-length arrays or macros with variable number of arguments. GCC-style inline assembly code is also supported. DWARF2 debugging information is provided to use with standard debuggers such as GDB. Certain Microsoft extensions are also enabled, such as __declspec attributes, along with support for Microsoft-style inline assembly code.
In addition to inline assembly code, the Intel compiler also supports MMX and SSE/SSE2 intrinsics. These allow access to the processor-specific extensions without the performance and correctness problems often caused by using inline assembly that can interfere with the analysis and transformations of the Intel compiler. By using the provided intrinsics, the programmer can take advantage of specific instructions but still receive the benefits of register allocation, scheduling and other optimizations.
The Intel compiler for Linux is a state-of-the-art compiler that delivers performance among the best in the industry, using sophisticated techniques to enable advanced features of Intel IA-32 architectures. More information can be found at developer.intel.com/software/products/compilers.
Thanks to Zia Ansari and David Kreitzer for their help in describing some of the technical details of the compiler. We also thank all the other members of the Intel compiler team.
Intel, Pentium, Itanium and MMX are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries.