projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b5e3dac) | patch
[flang] Remove fp128 support for llvm.round and llvm.trunc
authorPeixin Qiao <qiaopeixin@huawei.com>
Tue, 26 Jul 2022 12:21:51 +0000 (20:21 +0800)
committerPeixin Qiao <qiaopeixin@huawei.com>
Tue, 26 Jul 2022 12:21:51 +0000 (20:21 +0800)
commit418d2338f9bad03e002ca1a301ce537102a2f142
tree1b780a7af70fd4d884087e8f475f95e30af9cebatree | snapshot
parentb5e3dac33d42cdf8a3b19b1f64b726e700363dedcommit | diff
[flang] Remove fp128 support for llvm.round and llvm.trunc

The fp128 in llvm.round and llvm.trunc is not supported in X86_64 for
now. Revert the support. To support quad precision for llvm.round and
llvm.trunc, it may should be supported using runtime.

Reviewed By: Jean Perier

Differential Revision: https://reviews.llvm.org/D130556
flang/lib/Lower/IntrinsicCall.cpp diff | blob | history
flang/test/Lower/Intrinsics/aint.f90 diff | blob | history
flang/test/Lower/Intrinsics/anint.f90 diff | blob | history
flang/test/Lower/math-lowering.f90 diff | blob | history
Domain: System / Toolchain;