remove deprecated qkv/kv_packed apis

[sudhakarsingh27]

Description

Please include a brief summary of the changes, relevant motivation and context.

Fixes # (issue)

Type of change

• Documentation change (change only to the documentation, either a fix or a new content)
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Infra/Build change
• Code refactoring

Changes

Please list the changes introduced in this PR:

• Change A
• Change B

Checklist:

• I have read and followed the contributing guidelines
• The functionality is complete
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[greptile-apps]

Greptile Summary

removed four deprecated fused attention APIs (nvte_fused_attn_fwd_qkvpacked, nvte_fused_attn_bwd_qkvpacked, nvte_fused_attn_fwd_kvpacked, nvte_fused_attn_bwd_kvpacked) and their associated helper functions

• removed function declarations from header file (~290 lines)
• removed function implementations from cpp file (~588 lines)
• removed internal helper functions (make_tensor_view, calculate_qkv_stride, calculate_qkv_unpacked_shape, calculate_kv_stride, calculate_kv_unpacked_shape) that were only used by deprecated APIs (~68 lines)
• no remaining references to these functions in the codebase
• users should migrate to nvte_fused_attn_fwd and nvte_fused_attn_bwd with separate Q, K, V tensors

Confidence Score: 5/5

• safe to merge - clean removal of deprecated code with no remaining references
• all removed functions were properly deprecated, no remaining usage in codebase, helper functions only used by deprecated APIs, straightforward deletion
• no files require special attention

Important Files Changed

Filename	Overview
transformer_engine/common/include/transformer_engine/fused_attn.h	removed deprecated function declarations for qkvpacked and kvpacked APIs (4 functions, ~290 lines)
transformer_engine/common/fused_attn/fused_attn.cpp	removed deprecated function implementations and helper functions (~650 lines total)

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Deprecated APIs] --> B[nvte_fused_attn_fwd_qkvpacked]
    A --> C[nvte_fused_attn_bwd_qkvpacked]
    A --> D[nvte_fused_attn_fwd_kvpacked]
    A --> E[nvte_fused_attn_bwd_kvpacked]
    
    B --> F[Helper Functions]
    C --> F
    D --> F
    E --> F
    
    F --> G[make_tensor_view]
    F --> H[calculate_qkv_stride]
    F --> I[calculate_qkv_unpacked_shape]
    F --> J[calculate_kv_stride]
    F --> K[calculate_kv_unpacked_shape]
    
    L[Migration Path] --> M[nvte_fused_attn_fwd]
    L --> N[nvte_fused_attn_bwd]
    
    M --> O[Separate Q, K, V tensors]
    N --> O
    
    style A fill:#ff9999
    style F fill:#ff9999
    style L fill:#99ff99
    style O fill:#99ff99

Loading

_{Last reviewed commit: 849077c}

[greptile-apps]

_{2 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}