add dfx for operator FusedDeepMoe

[wangyibo1005]

Motivation

Debugging and triaging issues in the FusedDeepMoe operator was hindered by fragmented, ad-hoc logging and error checks. This PR introduces a centralized DFX (Diagnostics, Fault management, and eXperience) mechanism so that failures are reported in a consistent, information-rich way, improving observability and long-term maintainability.

Modification

Add a lightweight DFX utility layer via new headers (e.g., dfx_base.h, ops_log.h, ops_error.h) to provide standardized logging and structured error-reporting macros.

Refactor FusedDeepMoe host-side logic (infer/tiling paths) to adopt the new macros, replacing legacy checks/logging patterns (e.g., migrating from OP_TILING_CHECK / OP_LOGE to OPS_ERR_IF / OPS_LOG_E).

Improve log context and robustness, including consistent module/package naming and more detailed diagnostic fields (e.g., richer contextual logging and structured inner-error reporting).

Result and Report

More actionable diagnostics: errors are now reported through a unified framework with standardized formatting and richer context, making operator issues easier to reproduce and root-cause.

Cleaner and more maintainable code: logging/error handling is centralized, reducing duplicated patterns and improving consistency across operator components.

Scope: changes are primarily DFX/instrumentation and refactoring (5 files changed; net +379/−103), with no intended change to the operator’s functional behavior. Detailed debug infomation can be viewed in plog now.

Before:

Without debug detail

After:

DEEPEP_OPS[FusedDeepMoeTilingFuncImpl][48856] OpName:[FusedDeepMoe] "HCCL_BUFFSIZE is too SMALL, epRankSize = 16, maxBs = 256, moeExpertNumPerRank = 16, " " tokenLength = 7168, " " NEEDED_HCCL_BUFFSIZE(epRankSize * maxBs * moeExpertNumPerRank * tokenLength * ** TOKEN_DTYPE_BYTE_SIZE * DOUBLE_BUFFER) = 1793MB, HCCL_BUFFSIZE=200MB" [ERROR] [048856,python]:

[gemini-code-assist]

Summary of Changes

Hello @wangyibo1005, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request implements a new, centralized DFX (Diagnostics, Fault management, and eXperience) logging and error handling mechanism. It introduces a set of utility headers (dfx_base.h, ops_log.h, ops_error.h) that define robust macros for logging various levels of information and reporting errors. The FusedDeepMoe operator's inference and tiling logic have been refactored to adopt this new framework, replacing older, less structured logging calls with the standardized DFX macros, thereby enhancing the overall observability and maintainability of the operator.

Highlights

• New DFX Logging Framework: Introduced dfx_base.h, ops_log.h, and ops_error.h to establish a standardized Diagnostics, Fault management, and eXperience (DFX) logging and error reporting system. This framework provides flexible macros for various log levels and detailed context information.
• Refactored FusedDeepMoe Operator: Updated fused_deep_moe_infer.cpp and fused_deep_moe_tiling.cpp to replace legacy OP_TILING_CHECK and OP_LOGE macros with the newly defined OPS_ERR_IF and OPS_LOG_E macros, integrating the operator into the new DFX framework for improved error handling and logging consistency.
• Enhanced Log Detail and Error Reporting: The new logging system provides more detailed log messages, including file, line, function, thread ID, and operation-specific information. New error macros like OPS_ERR_IF and OPS_INNER_ERR_STUB facilitate conditional error reporting and structured error messages.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request introduces a new DFX (logging/error handling) framework and refactors the FusedDeepMoe operator to use it. This is a good step towards standardized logging and error reporting, providing more context in logs and structured error handling. I've identified a few areas for improvement: a potential stack overflow due to a large stack allocation in a logging macro, a generic exception handler that could be more specific, a logical error in a validation check, and some code redundancy in the new macros. Addressing these points will make the new DFX framework more robust and maintainable.