feat: support CCL Send and Recv

[GordonYang1]

Summary

This PR adds Send and Recv support to the shared CCL backend abstraction, including native bindings through the existing NCCL and MCCL provider layers for the public infinicclSend() and infinicclRecv() APIs. It also adds shared communicator and data-type handling for CCL point-to-point operations, keeps inter-only communicators on the existing OpenMPI staging path during mixed-backend flows, includes CCL-only plus OpenMPI-assisted Send/Recv example programs with correctness and one-way communication-bandwidth reporting, and enforces and documents NCCL 2.10 as the minimum supported version.

Changes

• Public API and Dispatch

  • Enable the existing infinicclSend() and infinicclRecv() APIs for configured CCL backends through generated bridge dispatch.
  • Keep an inter-only communicator on its initialized OpenMPI provider when a mixed OpenMPI/CCL build selects the CCL implementation before native rank initialization.
  • Use the native CCL communicator once infinicclCommInitRank() has initialized the matching intra-node provider.
  • Reuse the existing public signatures without introducing a new communication API.

• Common CCL Implementation

  • Add shared CCL Send and Recv implementations following the existing provider-oriented AllReduce structure.
  • Validate the CCL communicator instance and map InfiniCCL data types through the configured provider before dispatch.
  • Return NotSupported when the selected provider cannot represent the requested data type.

• Existing CCL Provider Bindings

  • Extend the existing NCCL and MCCL API wrappers with thin bindings to ncclSend(), ncclRecv(), mcclSend(), and mcclRecv().
  • Register Send and Recv with the existing NCCL and MCCL provider layers.

• Examples and Validation

  • Add a thread-per-GPU single-node CCL Send/Recv example using one shared unique ID.
  • Add an OpenMPI-assisted CCL Send/Recv example that uses OpenMPI for rank discovery and unique-ID broadcast, then initializes a native CCL communicator for GPU point-to-point communication.
  • Validate a matched one-way transfer from rank 0 to rank 1 and propagate validation failures through the process exit status.
  • Report payload size, average profiled time, algorithm bandwidth as payload_bytes / time, and bus bandwidth with the point-to-point correction factor of 1 in all three examples.

• Documentation

  • Enforce NCCL 2.10 at compile time and document it as the minimum supported version because the existing NCCL type map uses ncclAvg.

Platform and Backend Affected

Platform

• CPU
• NVIDIA GPU
• Iluvatar GPU
• MetaX GPU
• Moore Threads GPU
• Cambricon MLU
• HYGON DCU

Backend

• OpenMPI
• MPICH
• NCCL
• MCCL

Performance Impact

• No performance impact
• Performance improved
• Performance regression possible

This adds GPU-native CCL paths for Send and Recv, avoiding the existing MPI host-staging paths when a supported CCL backend is selected. Existing MPI point-to-point operations are intended to remain unchanged.

Known Issues & Future Work

• CCL operation support now covers AllReduce, Send, and Recv; other CCL operations remain future work.
• Send and Recv inherit the backend/device combinations and data type support of the existing CCL providers; this PR does not add a new provider or device integration.
• The CCL examples validate one matched, one-way float32 transfer from rank 0 to rank 1 on the default stream. Multi-peer, bidirectional, grouped, additional-data-type, and non-default-stream coverage remain future work. The heterogeneous MPI Send/Recv regression also transfers between ranks 0 and 1 on the NVIDIA node, so it is not a cross-vendor point-to-point test.

Test Results

The implementation commit is a0874c9ba781c110ab1325f7b0ded2d7f8d31c9e, a single commit directly based on ef4045a2d99837c75c2acd90aae57dacb83172d2. The attached evidence contains exactly 15 hash-verified canonical logs from the current commit: all 15 targets passed, with Correct: YES 17 times and Correct: NO 0 times. The extra two positive results are the additional expected validation cases in the broadcast log.

All four Send/Recv paths passed with a 1,048,576-element Float32 payload (4.00 MiB), 2 warm-up iterations, and 20 profiled iterations:

Path	Test topology	Payload	Time	Alg BW	Bus BW
Pure NCCL	8 NVIDIA A100 GPUs	4.00 MiB	0.096 ms	43.69 GB/s	43.69 GB/s
OpenMPI + NCCL hybrid	8 NVIDIA A100 GPUs	4.00 MiB	0.062 ms	67.24 GB/s	67.24 GB/s
Heterogeneous OpenMPI	8 NVIDIA A100 + 8 MetaX C550 GPUs	4.00 MiB	2.189 ms	1.92 GB/s	1.92 GB/s
Pure MCCL	8 MetaX C550 GPUs	4.00 MiB	0.128 ms	32.69 GB/s	32.69 GB/s

The reported time is the elapsed time averaged over the 20 profiled one-way transfers after the 2 warm-up transfers. Algorithm bandwidth uses payload_bytes / time; bus bandwidth is identical because the point-to-point correction factor is 1. The values were independently recomputed while accounting for the displayed time being rounded to three decimal places. They are included as execution evidence, not as a cross-platform benchmark comparison.

• The pure NCCL and OpenMPI+NCCL configurations each passed both selected targets, 2/2 and 2/2, on all 8 A100 GPUs.
• The heterogeneous OpenMPI configuration passed all 9 MPI targets on 16 ranks. Ranks 0-7 mapped to NVIDIA devices 0-7, and ranks 8-15 mapped to MetaX devices 0-7.
• The pure MCCL configuration passed both selected targets on all 8 MetaX C550 GPUs, with ranks 0-7 mapped to devices 0-7.
• All five formal invocations returned zero. There were no harness retries, MetaX vendor queue retries, timeouts, missing canonical logs, or recorded finalization errors.

Test Involved Platform

• CPU
• NVIDIA GPU
• Iluvatar GPU
• MetaX GPU
• Moore Threads GPU
• Cambricon MLU
• HYGON DCU

Test Involved Backend

• OpenMPI
• MPICH
• NCCL
• MCCL

Pure CCL (NCCL) on single-node NVIDIA:
ccl_all_reduce.log
ccl_broadcast.log

CCL + MPI on single-node NVIDIA:
ccl_mpi_hybrid_all_reduce.log
ccl_mpi_hybrid_broadcast.log

MPI on Heterogeneous Cluster:
mpi_all_gather.log
mpi_all_reduce.log
mpi_all_to_all.log
mpi_broadcast.log
mpi_gather.log
mpi_reduce.log
mpi_reduce_scatter.log
mpi_scatter.log
mpi_send_recv.log

Pure CCL (MCCL) on single-node MetaX:
ccl_all_reduce.log
ccl_broadcast.log

---

Checklist

Every contributor must verify every item below before requesting
review. Tick each box only after the check has actually been performed —
do not tick speculatively. If an item truly does not apply, replace the
checkbox with N/A and briefly explain why in an inline comment.

Title, Branch, and Commits

• PR title follows Conventional Commits (e.g. feat: …, fix(nccl): …).
• Branch name follows <type>/xxx-yyyy-zzzz where <type> matches the PR title's Conventional Commits type and words are joined with hyphens (see CONTRIBUTING.md §Branches).
• Each commit message follows Conventional Commits.
• Small PR is a single squashable commit; or, for a large PR, every commit is meaningful, well-formed, and independently reviewable (see CONTRIBUTING.md §Pull Requests).
• No stray merge commits from master — the branch is rebased cleanly on top of the current master.
• No fixup! / squash! / wip commits remain.

Scope and Design

• Changes are minimal — no unrelated modifications were introduced (CONTRIBUTING.md §Code/General).
• No dead code, commented-out blocks, debug prints, printf/std::cout/print(...) left behind, or TODO without an owner and issue link.
• No unrelated formatting churn that would obscure the diff.
• Public API changes (if any) are intentional, documented, and reflected in affected callers/tests.

General Code Hygiene

• The code is self-explanatory; comments were added only where the intent or rationale is non-obvious (CONTRIBUTING.md §Code/General).
• Every modified or added file ends with a single trailing newline (CONTRIBUTING.md §Code/General).
• No trailing whitespace, inconsistent indentation, or mixed formatting styles remain.
• Identifiers referenced in comments or error messages are wrapped in Markdown backticks (e.g. the `AllReduce` implementation) (CONTRIBUTING.md §Code/General).
• All comments and error messages are in English (CONTRIBUTING.md §Code/General).
• Comments and error messages are complete sentences — capitalized first letter, terminal punctuation — unless the language/framework convention says otherwise (CONTRIBUTING.md §Code/General; §Python).

C++ Specific (if C++ files changed)

• Code follows the Google C++ Style Guide strictly.
• clang-format (version 16, per .github/workflows/clang-format.yml) has been run against all modified applicable files; the diff is clean.
• No exceptions are thrown. Error paths use assert with messages that include at least __FILE__, __LINE__, and __func__ (CONTRIBUTING.md §C++).
• Error and warning message wording follows the LLVM Coding Standards (CONTRIBUTING.md §C++).
• N/A- Constructor initializer list order matches member declaration order (CONTRIBUTING.md §C++).
• Exactly one blank line between classes, between classes and functions, and between functions (CONTRIBUTING.md §C++).
• Exactly one blank line between members (functions and variables) within a class (CONTRIBUTING.md §C++).
• Exactly one blank line before and after the contents of a namespace (CONTRIBUTING.md §C++).

Python Specific (if Python files changed)

• N/A- Code is PEP 8 compliant; ruff check passes cleanly on CI (see `.github/workflows/ruff.yml).
• N/A- ruff format --check passes cleanly — if not, run ruff format and commit the result.
• N/A- Comments are complete English sentences, starting with a capital letter and ending with punctuation; Markdown backticks are used for code references (CONTRIBUTING.md §Python).
• N/A- Framework-specific conventions (e.g. lowercase pytest.skip messages without terminal period) are honored where applicable (CONTRIBUTING.md §Python).
• N/A- No blank line between the function signature and the body when there is no docstring or comment (CONTRIBUTING.md §Python).
• N/A- A blank line is present before and after if, for, and similar control-flow statements (CONTRIBUTING.md §Python).
• N/A- A blank line appears before each return, except when it directly follows a control-flow statement (CONTRIBUTING.md §Python).
• N/A- Docstrings (if any) follow PEP 257 (CONTRIBUTING.md §Python).
• N/A- Type hints are added / kept consistent with the surrounding code.

Testing

• All applicable example programs have been built and tested successfully on at least one supported heterogeneous cluster setup.

Build, CI, and Tooling

• N/A- New backends or devices have been added to auto-detection in CMakeLists.txt under if(AUTO_DETECT_DEVICES) or to if(AUTO_DETECT_BACKENDS) if applicable.
• Both CI workflows (clang-format.yml, ruff.yml) are green locally (or expected to be green on CI).

Documentation

• README.md, CONTRIBUTING.md, or inline docs updated when behavior, build flags, or developer workflow changed.
• N/A- Any user-visible breaking change is called out explicitly under "Summary" and in the commit/PR title with a ! or BREAKING CHANGE: footer.

Security and Safety

• No secrets, access tokens, internal URLs, customer data, or personal hardware identifiers have been committed.
• N/A- Third-party code is license-compatible and attributed.
• No unsafe pointer arithmetic, uninitialized reads, or missing bounds checks were introduced.