Improve model tracer infra

[Aswinmcw]

Ticket

NA

Problem description

Model tracer is using c++ graph tracer which is not perfect

What's changed

Remove the graph tracer dependencies and used python tracer to make it more effective
Left old infra intact till we completely migrate to new infra
Added new infra files as _v2
This PR just adds the infra changes, upcoming PR will contain changes to the sweep test that will work with _v2 loader

Checklist

• 
• 
• 
• New/Existing tests provide coverage for changes

Model tests

If your changes cover model-related code, you should run tests corresponding to affected models and platforms (Single card, T3K, Galaxy). "Choose your pipeline" workflows facilitate running multiple kinds of tests in a single run. Each offers models-mandatory and models-extended presets.
The former includes a minimal set of tests, to be run always. The latter extends that with additional ones - use your best judgement in deciding which is the most appropriate for your PR.

• 

  • models-mandatory preset (runs: Device perf regressions and Frequent model and ttnn tests)
  • models-extended preset (runs: the mandatory tests, plus Demo and Model perf tests)
  • other selection - specify runs

• 

  • models-mandatory preset (runs: Unit tests)
  • models-extended preset (runs: the mandatory tests, plus Demo and Model perf tests)
  • other selection - specify runs

• 

  • models-mandatory preset (runs: Quick tests)
  • models-extended preset (runs: the mandatory tests, plus Demo and Model perf tests)
  • other selection - specify runs

[Copilot]

Pull request overview

Copilot reviewed 12 out of 13 changed files in this pull request and generated 7 comments.

Comments suppressed due to low confidence (1)

tests/sweep_framework/load_ttnn_ops_data_v2.py:213

• The V2 schema (model_tracer/destructively_create_ttnn_ops_schema_v2.sql) defines ttnn_ops_v2.ttnn_mesh_config with only (mesh_shape, device_count), but this loader still inserts/selects placement_type/shard_dim/distribution_shape columns. This will fail at runtime against the V2 schema. Update get_or_create_mesh_config and its callers to match the V2 table definition (and adjust the mesh_key uniqueness accordingly), or target the V1 schema if placement fields are required.

    if mesh_key not in mesh_config_cache:
        cur.execute(
            """
            INSERT INTO ttnn_ops_v2.ttnn_mesh_config (mesh_shape, device_count, placement_type, shard_dim, distribution_shape)
            VALUES (%s, %s, %s, %s, %s)
            ON CONFLICT (mesh_shape, device_count, placement_type, shard_dim, distribution_shape) DO NOTHING
            RETURNING ttnn_mesh_config_id
        """,
            (mesh_shape, device_count, placement_type, shard_dim, distribution_shape),

[stevendae]

Do not merge until my approval is complete.

[stevendae]

this merge would still continue to use the v1 schema until sweeps_parameter_generator.py imports master_config_loader_v2.py.

I see the intent. Implement changes without changing core workflow just yet.