[PATCH] D44661: [libcxx] optimize reduce(), hmin(), hmax() by reordering the operations.
Zhihao Yuan via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Sun Mar 25 01:27:40 PDT 2018
lichray added inline comments.
================
Comment at: libcxx/include/experimental/simd:2196
+ typename _SimdType::value_type>::type
+__hmax(const _SimdType& __v) {
auto __acc = __v[0];
----------------
These conditions are too long, consider
```
__hmax_impl(const _SimdType& __v, true_type, ...);
__hmax_impl(const _SimdType& __v, false_type, true_type);
__hmax_impl(const _SimdType& __v, false_type, false_type);
```
https://reviews.llvm.org/D44661
More information about the cfe-commits
mailing list