diff options
author | Antonio Sanchez <cantonios@google.com> | 2020-11-17 15:32:44 -0800 |
---|---|---|
committer | Antonio Sánchez <cantonios@google.com> | 2020-11-18 20:32:35 +0000 |
commit | 17268b155d54422f1294130c0fb8c178757d911a (patch) | |
tree | 2be3d541729f3e9be6a180a58270bae10156df4f /Eigen/src/Core/arch/NEON/PacketMath.h | |
parent | 41d5d5334b8a4e364dfd88dcd91f6cd38834b8ed (diff) |
Add bit_cast for half/bfloat to/from uint16_t, fix TensorRandom
The existing `TensorRandom.h` implementation makes the assumption that
`half` (`bfloat16`) has a `uint16_t` member `x` (`value`), which is not
always true. This currently fails on arm64, where `x` has type `__fp16`.
Added `bit_cast` specializations to allow casting to/from `uint16_t`
for both `half` and `bfloat16`. Also added tests in
`half_float`, `bfloat16_float`, and `cxx11_tensor_random` to catch
these errors in the future.
Diffstat (limited to 'Eigen/src/Core/arch/NEON/PacketMath.h')
0 files changed, 0 insertions, 0 deletions