aboutsummaryrefslogtreecommitdiff
path: root/llvm/lib/CodeGen/MachineFunction.cpp
diff options
context:
space:
mode:
authorMatt Arsenault <Matthew.Arsenault@amd.com>2023-04-26 21:57:10 -0400
committerMatt Arsenault <Matthew.Arsenault@amd.com>2023-06-28 14:50:16 -0400
commit003b58f65bdd5d9c7d0c1b355566c9ef430c0e7d (patch)
tree9e28609c5c61dcc4b13fa95b66c4f7fa530140fc /llvm/lib/CodeGen/MachineFunction.cpp
parentee476996bec7f170928505a4c5b7715183cfbada (diff)
downloadllvm-003b58f65bdd5d9c7d0c1b355566c9ef430c0e7d.zip
llvm-003b58f65bdd5d9c7d0c1b355566c9ef430c0e7d.tar.gz
llvm-003b58f65bdd5d9c7d0c1b355566c9ef430c0e7d.tar.bz2
IR: Add llvm.frexp intrinsic
Add an intrinsic which returns the two pieces as multiple return values. Alternatively could introduce a pair of intrinsics to separately return the fractional and exponent parts. AMDGPU has native instructions to return the two halves, but could use some generic legalization and optimization handling. For example, we should be able to handle legalization of f16 on older targets, and for bf16. Additionally antique targets need a hardware workaround which would be better handled in the backend rather than in library code where it is now.
Diffstat (limited to 'llvm/lib/CodeGen/MachineFunction.cpp')
0 files changed, 0 insertions, 0 deletions