Move CCL ops to ttnn - Githubissues

tenstorrent / tt-metal

:metal: TT-NN operator library, and TT-Metalium low level kernel programming model.

Apache License 2.0

303 stars 26 forks source link

Move CCL ops to ttnn #9486

Open ayerofieiev-tt opened 2 weeks ago

ayerofieiev-tt commented 2 weeks ago

Consolidate Tensor Manipulation ops in ttnn

Plan

all_gather

[x] replace c++ usage (not used)
[x] replace python usage #9517
[x] remove python binding
[ ] consolidate into proper structure

line_all_gather

[x] replace c++ usage (not used)
[ ] Add ttnn op
[ ] replace python usage
[ ] remove python binding
[ ] consolidate into proper structure

reduce_scatter

[x] replace c++ usage (not used)
[ ] Add ttnn op
[ ] replace python usage
[ ] remove python binding
[ ] consolidate into proper structure

CCL common (test and source) after above are done

[ ] remaining standalone and ccl common files under tt_eager/tt_dnn/op_library/ccl
- [ ] ccl_common, README, ccl_host_datastructures, and contents of tt_eager/tt_dnn/op_library/ccl/shared_with_host/
[ ] Unit tests under tests/tt_eager/ops/ccl

Testing

For the best coverage, I recommend to run these workflows. If some of them fails, check if it is the same fail as on main:

Post commit CI
[post-commit] models tests
Device perf regressions and output report
Model perf regressions and output report
Nightly fast dispatch model regression CI
T3000 frequent tests
T3000 demo tests

SeanNijjar commented 3 days ago

Added a few extra bullets for all the CCL common code/tests