Skip to content

[HEXAGON] Auto-vectorization (fp16) for v68 #12397

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Aug 18, 2022
Merged

[HEXAGON] Auto-vectorization (fp16) for v68 #12397

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
4dd2434
Auto-vectorization (fp16) for v68
Aug 12, 2022
78590b8
use tvm.testing.main in fp16 test of tanh_slice op
Aug 12, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions python/tvm/target/target.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -724,6 +724,12 @@ def create_llvm_options(cpu_ver, config): # pylint: disable=unused-argument

llvm_options = config["llvm_options"]

# To enable auto-vectorization for v68 target added the below llvm-option by default
if arch_version == 68:
if not llvm_options:
llvm_options = ""
llvm_options += " -force-hvx-float"

Copy link
Contributor

@cconvey cconvey Aug 17, 2022

Choose a reason for hiding this comment

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

Does this flag impose any new requirements regarding which LLVM versions TVM needs?

Copy link
Contributor

@cconvey cconvey Aug 17, 2022

Choose a reason for hiding this comment

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

Also, are there any potential downsides or other side-effects of using this flag? Asking because with compiler flags, the word "force" sometimes implies that the resulting behavior isn't always a good idea.

Copy link
Contributor

@kparzysz-quic kparzysz-quic Aug 17, 2022

Choose a reason for hiding this comment

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

It will work with any LLVM version that supports v69. It's fine to use it.

# TVM's option parser doesn't allow '=' in values, but '=' can
# appear in LLVM flags. Replace it with '@', since it's unlikely
# that '@' will be used in another context.
Expand Down
2 changes: 1 addition & 1 deletion tests/python/contrib/test_hexagon/topi/test_tanh_slice.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -106,4 +106,4 @@ def test_tanh(


if __name__ == "__main__":
sys.exit(pytest.main(sys.argv))
tvm.testing.main()