Skip to content

[SYCL][ESIMD] Explicit SIMD memory-access APIs #1856

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.

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 1 commit into from
Jul 8, 2020

Conversation

DenisBakhvalov
Copy link
Contributor

This commit adds implementation of Explicit SIMD memory-access
APIs along with memory intrinsics used to implement operations
with SIMD classes.

Co-authored-by: Chen, Gang Y <[email protected]>
Co-authored-by: Chen, Kai Yu <[email protected]>
Co-authored-by: Bobrovsky, Konstantin S <[email protected]>
Co-authored-by: Ashar, Pratik J <[email protected]>

Signed-off-by: Denis Bakhvalov <[email protected]>

@DenisBakhvalov
Copy link
Contributor Author

Review only 4th commit:
[SYCL][ESIMD] Explicit SIMD memory-access APIs

@DenisBakhvalov DenisBakhvalov added the esimd Explicit SIMD feature label Jun 11, 2020
Copy link
Contributor

@Pennycook Pennycook left a comment

Choose a reason for hiding this comment

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

Thanks for renaming flat_load to gather and flat_store to scatter. I think this looks a lot clearer. It probably makes sense to rename the tests as well (e.g. flat_load.cpp to gather.cpp).

We should discuss the CacheHint template arguments and flat_atomic interface some more, but that shouldn't block this PR.

This commit adds implementation of Explicit SIMD memory-access
APIs along with memory intrinsics used to implement operations
with SIMD classes.

Co-authored-by: Chen, Gang Y <[email protected]>
Co-authored-by: Chen, Kai Yu <[email protected]>
Co-authored-by: Bobrovsky, Konstantin S <[email protected]>
Co-authored-by: Ashar, Pratik J <[email protected]>

Signed-off-by: Denis Bakhvalov <[email protected]>
@DenisBakhvalov DenisBakhvalov force-pushed the private/dbakhval-esimd-memory branch from a09386c to cdc3044 Compare July 8, 2020 00:09
@romanovvlad
Copy link
Contributor

romanovvlad commented Jul 8, 2020

@DenisBakhvalov Is this ready for review? If no, please mark as a draft. If yes, the patch changes 1500 LOC which is quite huge, could please split it up to several patches?

@dendibakh
Copy link
Contributor

@DenisBakhvalov Is this ready for review? If no, please mark as a draft. If yes, the patch changes 1500 LOC which is quite huge, could please split it up to several patches?

@romanovvlad , The patch is ready for review. I know, the patch is relatively big, but all changes serve the same logical purpose. It was reviewed earlier by @Pennycook.

Copy link
Contributor

@romanovvlad romanovvlad left a comment

Choose a reason for hiding this comment

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

@Pennycook Could you please approve if it's OK with you?

@@ -0,0 +1,23 @@
// RUN: %clangxx -fsycl -fsycl-explicit-simd -fsycl-device-only -fsyntax-only -Xclang -verify %s
// expected-no-diagnostics
Copy link
Contributor

Choose a reason for hiding this comment

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

Can we have tests with real execution rather than just frontend checks?

Copy link
Contributor

Choose a reason for hiding this comment

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

We would have functional tests once we upstream the whole ESIMD extension.

@againull againull merged commit aad7c1c into intel:sycl Jul 8, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
esimd Explicit SIMD feature
Projects
None yet
Development

Successfully merging this pull request may close these issues.

7 participants