[CK TILE ENGINE] GEMM Multi D Restructure

[ThruptiRajLakshmanaGowda]

Proposed changes

Currently CK Tile Engine GEMM Multi D operator had a design which was taking longer time to build, with this new design the build time will be reduced.

Checklist

Please put an x into the boxes that apply. You can also fill these out after creating the PR. If you're not sure, please don't hesitate to ask.

• I have added tests relevant to the introduced functionality, and the unit tests are passing locally
• I have added the test to REGRESSION_TESTS list defined at the top of CMakeLists.txt in tests/CMakeLists.txt, IF the test takes more than 30 seconds to run.
• I have added inline documentation which enables the maintainers with understanding the motivation
• I have removed the stale documentation which is no longer relevant after this pull request
• (If this change is user-facing) I have added release notes which provide the end users with a brief summary of the improvement from this pull request
• I have run clang-format on all changed files
• Any dependent changes have been merged

Discussion

If this is a relatively large or complex change, feel free to start a discussion by explaining why you chose the solution you did and what alternatives you considered

[Copilot]

Pull Request Overview

This PR refactors the GEMM Multi D kernel generation system from a monolithic build approach to an individual kernel compilation model with parallel generation support. Key improvements include:

• Renaming output tensor from "E" to "C" for consistency
• Replacing batch dispatcher with individual kernel benchmarking
• Adding parallel kernel generation using Python's ProcessPoolExecutor
• New validation utilities for tile configurations
• JSON-based benchmark results with comprehensive metrics

Reviewed Changes

Copilot reviewed 17 out of 17 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
gemm_multi_d_profiler.hpp	Updated to benchmark single kernels; renamed E to C tensor references
gemm_multi_d_instance_builder.py	Complete rewrite for individual kernel generation with parallel processing
gemm_multi_d_benchmark.py	New Python-based benchmark orchestration tool with JSON output
gemm_multi_d_common.hpp	New file consolidating common types and helper functions
commons/validation_utils.py	New comprehensive validation utilities for tile configurations
gemm_multi_d_benchmark_single.cpp	New single-kernel benchmark executable
CMakeLists.txt	Refactored to build individual kernel targets instead of monolithic libraries
configs/*.json	Updated with new persistent field and k_block_per_cu parameter
Jenkinsfile	Simplified CI to use Python benchmark orchestrator

Comments suppressed due to low confidence (2)

tile_engine/ops/gemm_multi_d/gemm_multi_d_instance_builder.py:1

• Debug print statement left in production code. This should be removed or converted to proper logging.

#!/usr/bin/env python

tile_engine/ops/gemm_multi_d/gemm_multi_d_benchmark.hpp:68

• [nitpick] Unnecessary concatenation of two string literals. The second << \"\\n\" can be combined with the previous string literal as << \"\\n\".

---

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

[ThomasNing]

Other part LGTM overall

Review addressed