Intel® C++ Compiler 16.0 User and Reference Guide
Enables or disables compiler interpretation of SIMD pragmas.
Linux and OS X: | -simd -no-simd |
Windows: | /Qsimd /Qsimd- |
None
-simd |
SIMD pragmas are enabled. |
This option enables or disables compiler interpretation of SIMD pragmas.
To disable interpretation of SIMD pragmas, specify -no-simd (Linux* OS and OS X*) or /Qsimd- (Windows* OS). Note that the compiler may still vectorize loops based on its own heuristics (leading to generation of SIMD instructions) even when -no-simd (or /Qsimd-) is specified.
To disable all compiler vectorization, use the "-no-vec -no-simd" (Linux* OS and OS X*) or "/Qvec- /Qsimd-" (Windows* OS) compiler options. The option -no-vec (and /Qvec-) disables all auto-vectorization, including vectorization of array notation statements. The option -no-simd (and /Qsimd-) disables vectorization of loops that have SIMD pragmas.
If you specify option -mia32 or option /arch:IA32, SIMD pragmas are disabled by default. When you specify option -mia32 or option /arch:IA32, vector instructions cannot be used. Therefore, you cannot explicitly enable SIMD pragmas by specifying option [Q]simd.
None