Skip to content

Conversation

@wrongtest-intellif
Copy link
Contributor

Try fix #12566

@vinx13
Copy link
Member

vinx13 commented Aug 24, 2022

cc @shingjan

Copy link

@shingjan shingjan left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Thanks for sending this fix!

case 16: {
os << "__float2half_rn";
os << '(' << std::scientific << op->value << 'f' << ')';
os << "__float2half_rn" << '(';
Copy link

@shingjan shingjan Aug 24, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would that be possible that some tests are added? We can probably re-use my repro here

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

the case is added

@wrongtest-intellif wrongtest-intellif force-pushed the fix_cuda_fp16_inf_literal branch from c04edc7 to 2614cb1 Compare August 25, 2022 07:29
@vinx13 vinx13 merged commit bb00a15 into apache:main Aug 25, 2022
xinetzone pushed a commit to daobook/tvm that referenced this pull request Nov 25, 2022
* Fix cuda codegen's fp16 inf literal

* add relay testcase
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Bug] [TIR] float infinity not supported in TIR with CUDA target

3 participants