craig.topper added a comment.
Implemented @DavidKreitzer's suggestion in r333033
Repository:
rL LLVM
https://reviews.llvm.org/D47174
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
DavidKreitzer added a comment.
A bit of history: In icc, the f16<=>f32 conversion intrinsics are a bit of an
anomaly in that they can be implemented using either native code or emulation
code based on the target architecture switch. See https://godbolt.org/g/bQy7xY
(thanks, Craig, for the
This revision was automatically updated to reflect the committed changes.
Closed by commit rL333014: [X86] Move 128-bit f16c intrinsics to
__emmintrin_f16c.h include from emmintrin. (authored by ctopper, committed by ).
Herald added a subscriber: llvm-commits.
Changed prior to commit:
RKSimon added a comment.
In https://reviews.llvm.org/D47174#1108329, @craig.topper wrote:
> It is odd, but they really are split in the icc include files. So they got
> split a while back in clang to match the Intel Intrinsic Guide documentation.
OK - if that means we're matching latest
craig.topper added a comment.
It is odd, but they really are split in the icc include files. So they got
split a while back in clang to match the Intel Intrinsic Guide documentation.
Repository:
rC Clang
https://reviews.llvm.org/D47174
___
RKSimon added a comment.
Aren't all the instructions from the same CPUID bit? It seems odd to split them
across multiple files.
Repository:
rC Clang
https://reviews.llvm.org/D47174
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
rnk accepted this revision.
rnk added a comment.
This revision is now accepted and ready to land.
lgtm
Repository:
rC Clang
https://reviews.llvm.org/D47174
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
craig.topper added inline comments.
Comment at: lib/Headers/immintrin.h:72
-/* The 256-bit versions of functions in f16cintrin.h.
- Intel documents these as being in immintrin.h, and
Interesting this to note here, the 256-bit f16c intrinsics were being
craig.topper created this revision.
craig.topper added reviewers: RKSimon, spatel, echristo, DavidKreitzer.
Intel documents the 128-bit versions as being in emmintrin.h and the 256-bit
version as being in immintrin.h.
This patch makes a new __emmtrin_f16c.h to hold the 128-bit versions to be