Pgardner/unary bcast llk

[pgardnerTT]

Ticket

tenstorrent/tt-metal#23354

Problem description

Adding ROW, COL, SCALAR unary_bcast for Float32, Int32, Uint32, Uint16

What's changed

New dataformat support

Type of change

• 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)
• Documentation update

Checklist

• All post commit CI passes
• Blackhole Post commit CI passes (if applicable)

[github-actions]

Thank you for your contribution! 🚀

You can run tt-metal integration tests by adding the blackhole-integration-tests and/or wormhole-integration-tests labels to this pull request.

If you want to run metal post-commit tests, you can add the metal-post-commit-tests label to this pull request.

📖 For more information, please refer to our CONTRIBUTING guide.

[Copilot]

Pull request overview

This PR adds ROW, SCALAR, and COL broadcast support for 32-bit data formats (Float32, Int32, UInt32) and UInt16 in the low-level kernel (LLK) library. The implementation targets both Wormhole B0 and Blackhole architectures.

Key Changes:

• Added broadcast type handling for ROW, SCALAR, and COL operations with unpack-to-dest support for 32-bit formats
• Refactored address register configuration logic to use dynamic register selection based on context
• Enhanced math datacopy operations with specialized D2B/B2D instruction sequences for different broadcast types and UInt16 format

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
tt_llk_wormhole_b0/llk_lib/llk_unpack_A.h	Adds ROW/SCALAR/COL broadcast MOP configurations, refactors address register logic, adjusts semaphore timing, and updates unpacker x_end configuration for broadcast types
tt_llk_wormhole_b0/llk_lib/llk_math_eltwise_unary_datacopy.h	Implements broadcast-specific D2B/B2D instruction sequences, adds UInt16 handling for MOVB2D operations, configures address modes per format, and adds ADDR_MOD_3 initialization
tt_llk_wormhole_b0/common/inc/cunpack_common.h	Updates int8_math_enabled check to include UInt32 format and adds else clause to clear debug feature disable register
tt_llk_blackhole/llk_lib/llk_unpack_A.h	Mirrors Wormhole changes for Blackhole architecture including broadcast MOP configurations and address register refactoring
tt_llk_blackhole/llk_lib/llk_math_eltwise_unary_datacopy.h	Implements same broadcast D2B/B2D sequences and UInt16 handling as Wormhole version with Blackhole-specific FPU mode settings
tt_llk_blackhole/common/inc/cunpack_common.h	Adds else clause to clear debug feature disable register matching Wormhole implementation
tests/python_tests/test_bcast.py	New comprehensive test file covering ROW/SCALAR/COL broadcast types with supported formats including constraint filtering and golden reference generation

---

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