- 신규로 표시
- 북마크
- 구독
- 소거
- RSS 피드 구독
- 강조
- 인쇄
- 부적절한 컨텐트 신고
링크가 복사됨
- 신규로 표시
- 북마크
- 구독
- 소거
- RSS 피드 구독
- 강조
- 인쇄
- 부적절한 컨텐트 신고
_Interlocked* intrinsics are MSVC-specific (Intel compiler also has to support them for compatibility and MinGW translates them to its own intrinsics). They are not available on platforms other than Windows. Intel Intrinsics Guide documents portable intrinsics supported by most major compilers, and that is a good thing since it avoids the confusion.
- 신규로 표시
- 북마크
- 구독
- 소거
- RSS 피드 구독
- 강조
- 인쇄
- 부적절한 컨텐트 신고
Hello,
well what do you miss exatly: https://software.intel.com/sites/landingpage/IntrinsicsGuide/
You have intrinsics, can filter them, there is also AVX, AVX2, FMA and further extensions. Intrinsics that go into single instructions (which is true for many but not all) have timing information about latency, throughput for all recent architectures.
Greetings,
Christian
- 신규로 표시
- 북마크
- 구독
- 소거
- RSS 피드 구독
- 강조
- 인쇄
- 부적절한 컨텐트 신고
- 신규로 표시
- 북마크
- 구독
- 소거
- RSS 피드 구독
- 강조
- 인쇄
- 부적절한 컨텐트 신고
Hello,
ok I see. Are those extensions not listed under one of the pointer under 'other'? If you hover the mouse over there and wait some time, you get a loft of subcategories. Otherwise, for compiler specific you will have to look at the compiler guide.
For intel, check out this link: https://software.intel.com/en-us/node/514071
At the left, there is the navigation, you should find all intel compiler intrinsics. When it comes to MSVC, I think, the will not provide any timing specifications ... If you know to what instruction the intrincis boild down, you should find them in the appropriate intel architecture and/or optimization manual. Normally, you find all instructions there with timing information.
