[UIL] Add scalar IR API, on-device net construction, and per-peer staging fix

[MC952-arch]

PR Category

UIL

PR Types

New Features

PR Description

This PR updates the Device IR / LLVM-IR-facing API surface by adopting explicit suffixes for the two calling styles: C-suffixed struct-based wrappers (existing path) and a new S-suffixed scalar, struct-free API intended for Triton/LLVM IR consumers. It also expands the device IR test coverage to exercise the new scalar entry points.

[Copilot]

Pull request overview

This PR updates the Device IR / LLVM-IR-facing API surface by adopting explicit suffixes for the two calling styles: C-suffixed struct-based wrappers (existing path) and a new S-suffixed scalar, struct-free API intended for Triton/LLVM IR consumers. It also expands the device IR test coverage to exercise the new scalar entry points.

Changes:

• Renames existing IR wrapper entry points to C-suffixed symbols and updates kernels/tests accordingly.
• Introduces scalar enums (flagcxCoopKind_t, flagcxTeamKind_t) and a new S-suffixed scalar IR API (+ implementation).
• Extends the device IR unit test to add S1–S6 scalar IR kernel tests and updates exported IR symbol lists.

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
test/unittest/device_api/test_device_ir.cpp	Adds scalar IR (S1–S6) test coverage and includes scalar results in overall pass/fail aggregation.
test/device_api/device_ir.h	Declares scalar IR (S-suffixed) kernel launchers used by unit tests.
test/device_api/device_ir.cu	Updates kernels to call renamed C-suffixed wrappers and adds scalar IR kernels exercising S APIs.
flagcx/adaptor/include/device_api/flagcx_device_enums.h	Adds scalar enum types used by the S API to avoid struct instantiation in IR.
flagcx/adaptor/include/device_api/flagcx_device_core.h	Exposes scalar enums via core header for shared native + IR integration.
bindings/ir/public_symbols.txt	Updates exported symbol list to include C/S-suffixed APIs and scalar additions.
bindings/ir/flagcx_device_wrapper.h	Renames wrapper declarations to C-suffixed names and adds API documentation/deprecation note.
bindings/ir/flagcx_device_wrapper_impl.h	Renames wrapper implementations to C-suffixed names and includes scalar IR implementations.
bindings/ir/flagcx_device_scalar_ir.h	Adds new struct-free scalar IR API declarations for Triton/LLVM IR users.
bindings/ir/flagcx_device_scalar_ir_impl.h	Implements scalar IR APIs (coop/team/barrier/net) as inline device functions.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 12 out of 12 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 9 comments.

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 22 out of 22 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 21 out of 21 changed files in this pull request and generated 1 comment.