diff options
| author | Jerry Zhang <jerryzh@fb.com> | 2019-04-11 19:38:21 -0700 |
|---|---|---|
| committer | Facebook Github Bot <facebook-github-bot@users.noreply.github.com> | 2019-04-11 19:43:18 -0700 |
| commit | 7e73783c6fefb592a112859781cf662bfb0f408d (patch) | |
| tree | c150ce9eed1447c85a8bfc4c88885b074998e46a | |
| parent | 422b01e78889a9fd283d8fb86bb5527c5cae3cbf (diff) | |
| download | pytorch-7e73783c6fefb592a112859781cf662bfb0f408d.tar.gz pytorch-7e73783c6fefb592a112859781cf662bfb0f408d.tar.bz2 pytorch-7e73783c6fefb592a112859781cf662bfb0f408d.zip | |
Fix promoteTypes for QInt types (#19182)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/19182
This is a bug discovered by zafartahirov, right now if one of the tensor is QInt
type we'll return undefined, but actually we want to allow ops that accepts
Tensors of the same QInt type to work.
Reviewed By: zafartahirov
Differential Revision: D14909172
fbshipit-source-id: 492fd6403da8c56e180efe9d632a3b7fc879aecf
| -rw-r--r-- | c10/core/ScalarType.h | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/c10/core/ScalarType.h b/c10/core/ScalarType.h index 99e7975d35..b34ed427d3 100644 --- a/c10/core/ScalarType.h +++ b/c10/core/ScalarType.h @@ -246,6 +246,11 @@ static inline ScalarType promoteTypes(ScalarType a, ScalarType b) { "promoteTypes with complex numbers is not handled yet; figure out what the correct rules should be"); } + // For QInt types, we only allow exact match + if (isQIntType(a) && a == b) { + return a; + } + if (isQIntType(a) || isQIntType(b)) { AT_ERROR( "promoteTypes with quantized numbers is not handled yet; figure out what the correct rules should be"); |