Remove restriction of input_nsticks_per_core % w == 0 #15205
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
ttnn/cpp/ttnn/operations/pool/upsample/device/upsample_program_factory_multicore.cpp
Outdated
Show resolved
Hide resolved
ttnn/cpp/ttnn/operations/pool/upsample/device/upsample_program_factory_multicore.cpp
Outdated
Show resolved
Hide resolved
…ded tensor inputs. Signed-off-by: Nilaykumar Patel <[email protected]>
Signed-off-by: Nilaykumar Patel <[email protected]>
ToDo: commonize code. Signed-off-by: Nilaykumar Patel <[email protected]>
nkpatel-tt
force-pushed
the
nkpatel/upsample2d-generalization-remove-restrictions-input_width
branch
from
607825e to
e39f225
Compare
Signed-off-by: Nilaykumar Patel <[email protected]>
Signed-off-by: Nilaykumar Patel <[email protected]>
nkpatel-tt
force-pushed
the
nkpatel/upsample2d-generalization-remove-restrictions-input_width
branch
from
a6ede9f to
67eba82
Compare
Signed-off-by: Nilaykumar Patel <[email protected]>
…trictions-input_width Signed-off-by: Nilaykumar Patel <[email protected]>
ttnn/cpp/ttnn/operations/pool/upsample/device/upsample_program_factory_multicore.cpp
Show resolved
Hide resolved
ttnn/cpp/ttnn/operations/pool/upsample/device/upsample_program_factory_multicore.cpp
Show resolved
Hide resolved
Signed-off-by: Nilaykumar Patel <[email protected]>
nkpatel-tt
requested review from
ayerofieiev-tt,
dmakoviichuk-tt,
rfurko-tt,
cfjchu,
TT-BrianLiu,
razorback3,
dongjin-na,
bbradelTT,
mywoodstock,
shwetankTT,
sankarmanoj-tt and
pavlejosipovic
as code owners
…trictions-input_width
Suggested-by: Pavle Josipović <[email protected]> Signed-off-by: Nilaykumar Patel <[email protected]>
nkpatel-tt
force-pushed
the
nkpatel/upsample2d-generalization-remove-restrictions-input_width
branch
from
b677eea to
09b57cb
Compare
…trictions-input_width
bbradelTT
reviewed
Dec 19, 2024
ttnn/cpp/ttnn/operations/pool/upsample/device/upsample_program_factory_multicore.cpp
Outdated
Show resolved
Hide resolved
| writer_rt_args[5] = out_w; | ||
| writer_rt_args[6] = 0; // set for each core below | ||
| writer_rt_args[4] = input_nsticks_per_core; | ||
| writer_rt_args[5] = output_nsticks_per_core / 2; // half of the outputs are processed by each core |
There was a problem hiding this comment.
What happens if output_nsticks_per_core is odd?
There was a problem hiding this comment.
Moved logic to kernel where odd case is handled.
ttnn/cpp/ttnn/operations/pool/upsample/device/upsample_program_factory_multicore.cpp
Outdated
Show resolved
Hide resolved
bbradelTT
reviewed
Dec 19, 2024
ttnn/cpp/ttnn/operations/pool/upsample/device/upsample_program_factory_multicore.cpp
Show resolved
Hide resolved
mywoodstock
reviewed
Dec 20, 2024
...ttnn/operations/pool/upsample/device/kernels/dataflow/writer_upsample_multi_core_sharded.cpp
Outdated
Show resolved
Hide resolved
...ttnn/operations/pool/upsample/device/kernels/dataflow/writer_upsample_multi_core_sharded.cpp
Show resolved
Hide resolved
| constexpr uint32_t config_cb_id = get_compile_time_arg_val(3); | ||
|
|
||
| uint32_t reader_nsticks_per_core = (in_nsticks_per_core + is_reader) / 2; | ||
| uint32_t writer_nsticks_per_core = in_nsticks_per_core / 2; |
There was a problem hiding this comment.
I have removed this code and moved output sticks logic here and am deriving output sticks from reader_nsticks_per_core so that we handle even and odd case correctly.
...ttnn/operations/pool/upsample/device/kernels/dataflow/writer_upsample_multi_core_sharded.cpp
Outdated
Show resolved
Hide resolved
Signed-off-by: Nilaykumar Patel <[email protected]>
nkpatel-tt
changed the title
esmalTT
requested changes
Dec 20, 2024
Signed-off-by: Nilaykumar Patel <[email protected]>
This was referenced Dec 30, 2024
bbradelTT
approved these changes
Dec 30, 2024
esmalTT
approved these changes
Jan 4, 2025
There was a problem hiding this comment.
Approving to unblock, but please open an issue for this: #15205 (comment)
mywoodstock
approved these changes
Jan 8, 2025
|
please also run nightly pipeline to make sure there are no perf regressions. |
Done. |
nkpatel-tt
deleted the
nkpatel/upsample2d-generalization-remove-restrictions-input_width
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Ticket
Link
Problem description
Currently, whole input row is processed per core which inefficient since other cores could be idle
What's changed
Distribute work to all possible cores.
Checklist