__faststorefence
Microsoft Specific
Guarantees that every previous memory reference, including both load and store memory references, is globally visible before any subsequent memory reference.
Syntax
void __faststorefence();
Requirements
Intrinsic | Architecture |
---|---|
__faststorefence |
x64 |
Header file <intrin.h>
Remarks
Generates a full memory barrier instruction sequence that guarantees load and store operations issued before the intrinsic are globally visible before execution continues. The effect is comparable to but faster than the _mm_mfence
intrinsic on all x64 platforms.
On the AMD64 platform, this routine generates an instruction that is a faster store fence than the sfence
instruction. For time-critical code, use this intrinsic instead of _mm_sfence
only on AMD64 platforms. On Intel x64 platforms, the _mm_sfence
instruction is faster.
This routine is only available as an intrinsic.
END Microsoft Specific