Add TBE data configuration reporter to TBE forward (v3) (#4455) #4672

gchalump · 2025-08-11T20:21:56Z

Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/1516

Re-land attempt of D75462895

Add TBE data configuration reporter to TBE forward call.

The reporter reports TBE data configuration at the SplitTableBatchedEmbeddingBagsCodegen forward call. The output is a TBEDataConfig object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below.

Just Knobs for enablement

fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures)
- Default is set to False, enable this flag to enable reporter.
- To enable it locally use:
```
jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600
```

Environment Variables

The Reporter relies on several environment variables to control its behavior. Below is a description of each variable:

FBGEMM_REPORT_INPUT_PARAMS_INTERVAL:
- Description: Determines the interval at which reports are generated. This is specified in terms of the number of iterations.
- Example Value: 1 (report every iteration)
FBGEMM_REPORT_INPUT_PARAMS_ITER_START:
- *Description: Specifies the start of the iteration range to capture reports. Default 0.
- *Example Value: 0 (start reporting from the first iteration)

-   **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**:
  -   ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1.
  -   ***Example Value**: `-1` (report until the last iteration)

FBGEMM_REPORT_INPUT_PARAMS_BUCKET:
- Description: Specifies the name of the Manifold bucket where the report data will be saved.
- Example Value: tlparse_reports
FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX:
- Description: Defines the path prefix where the report files will be stored. Path will be created if not exist.
- Example Value: tree/tests/

Use Cases

FileStore
- General
  - Auto-create output directories if not exist.
- fb-internal:
  - Only export to manifold.
  - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist)
- OSS
  - Will use local FileStore to store the output

Example Usage

Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings:

FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2  FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3
FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2

Explanation

The above setting will report iter 3 and iter 5

FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2: The reporter will generate a report every 2 iterations.
FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0: The reporter will start generating reports from the first iteration.
FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default): The reporter will continue to generate reports until the last iteration interval.
FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports: The reports will be saved in the tlparse_reports bucket.
FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/: The reports will be stored with the path prefix tree/tests/. For Manifold make sure all folders within the path exist.

Note on Benchmark example

Note that with the --iters 2 option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0.

Other includes changes in this Diff:

Updates build dependency of tbe_data_config* files
Remove shutil and numpy.random lib as it cause uncompatiblity error.
Add non-OSS test, writing extracted config data json file to Manifold

Differential Revision: D79758603

netlify · 2025-08-11T20:22:01Z

✅ Deploy Preview for pytorch-fbgemm-docs ready!

Name	Link
🔨 Latest commit	`b631308`
🔍 Latest deploy log	https://app.netlify.com/projects/pytorch-fbgemm-docs/deploys/68b623f55d1c1100074aad19
😎 Deploy Preview	https://deploy-preview-4672--pytorch-fbgemm-docs.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

To edit notification comments on pull requests, go to your Netlify project configuration.

facebook-github-bot · 2025-08-11T20:22:04Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 X-link: facebookresearch/FBGEMM#1516 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

facebook-github-bot · 2025-08-12T23:24:18Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

facebook-github-bot · 2025-08-12T23:37:43Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 X-link: facebookresearch/FBGEMM#1516 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

facebook-github-bot · 2025-08-14T18:15:46Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

facebook-github-bot · 2025-08-14T18:19:50Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 X-link: facebookresearch/FBGEMM#1516 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

facebook-github-bot · 2025-08-18T15:57:58Z

This pull request was exported from Phabricator. Differential Revision: D79758603

facebook-github-bot · 2025-08-25T21:37:40Z

This pull request was exported from Phabricator. Differential Revision: D79758603

facebook-github-bot · 2025-08-25T21:53:19Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 X-link: facebookresearch/FBGEMM#1516 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Reviewed By: q10, spcyppt Differential Revision: D79758603

facebook-github-bot · 2025-08-27T19:40:07Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Reviewed By: q10, spcyppt Differential Revision: D79758603

facebook-github-bot · 2025-08-27T19:51:59Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Reviewed By: q10, spcyppt Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 X-link: facebookresearch/FBGEMM#1516 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Reviewed By: q10, spcyppt Differential Revision: D79758603

facebook-github-bot · 2025-09-01T22:50:00Z

This pull request was exported from Phabricator. Differential Revision: D79758603

Summary: X-link: facebookresearch/FBGEMM#1703 Pull Request resolved: pytorch#4672 X-link: facebookresearch/FBGEMM#1516 Pull Request resolved: pytorch#4455 Re-land attempt of D75462895 # Add TBE data configuration reporter to TBE forward call. The reporter reports TBE data configuration at the `SplitTableBatchedEmbeddingBagsCodegen` ***forward*** call. The output is a `TBEDataConfig` object, which is written to a JSON file(s). The configuration of its environment variables and an example of its usage is described below. ## Just Knobs for enablement - fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS is added for enablement of the reporter (https://www.internalfb.com/intern/justknobs/?name=fbgemm_gpu%2Ffeatures) - Default is set to `False`, enable this flag to enable reporter. - To enable it locally use: ``` jk canary set fbgemm_gpu/features:TBE_REPORT_INPUT_PARAMS --on --ttl 600 ``` ## Environment Variables --------------------- The Reporter relies on several environment variables to control its behavior. Below is a description of each variable: - **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL**: - **Description**: Determines the interval at which reports are generated. This is specified in terms of the number of iterations. - **Example Value**: `1` (report every iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_START**: - ***Description**: Specifies the start of the iteration range to capture reports. Default 0. - ***Example Value**: `0` (start reporting from the first iteration) - **FBGEMM_REPORT_INPUT_PARAMS_ITER_END**: - ***Description**: Specifies the end of the iteration range to capture reports. Use `-1` to report until the last iteration. Default -1. - ***Example Value**: `-1` (report until the last iteration) - **FBGEMM_REPORT_INPUT_PARAMS_BUCKET**: * **Description**: Specifies the name of the Manifold bucket where the report data will be saved. * **Example Value**: `tlparse_reports` - **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX**: - **Description**: Defines the path prefix where the report files will be stored. Path will be created if not exist. - **Example Value**: `tree/tests/` ## Use Cases - FileStore - General - Auto-create output directories if not exist. - fb-internal: - Only export to manifold. - Assert error, if the flag is set but failed to initialize manifold connection. (missing backend or manifold bucket is not exist) - OSS - Will use local FileStore to store the output ## Example Usage ------------- Below is an example command demonstrating how to use the FBGEMM Reporter with specific environment variable settings: ``` FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2 FBGEMM_REPORT_INPUT_PARAMS_ITER_START=3 FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/ buck2 run mode/opt //deeplearning/fbgemm/fbgemm_gpu/bench:split_table_batched_embeddings -- device --iters 2 ``` **Explanation** The above setting will report `iter 3` and `iter 5` * **FBGEMM_REPORT_INPUT_PARAMS_INTERVAL=2**: The reporter will generate a report every 2 iterations. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_START=0**: The reporter will start generating reports from the first iteration. * **FBGEMM_REPORT_INPUT_PARAMS_ITER_END=-1 (Default)**: The reporter will continue to generate reports until the last iteration interval. * **FBGEMM_REPORT_INPUT_PARAMS_BUCKET=tlparse_reports**: The reports will be saved in the `tlparse_reports` bucket. * **FBGEMM_REPORT_INPUT_PARAMS_PATH_PREFIX=tree/tests/**: The reports will be stored with the path prefix `tree/tests/`. For Manifold make sure all folders within the path exist. **Note on Benchmark example** Note that with the `--iters 2` option, the benchmark will execute 6 forward calls (2 iterations plus 1 warmup) for the forward benchmark and another 3 calls (2 iterations plus 1 warmup) for the backward benchmark. Iteration starts from 0. --- --- ## Other includes changes in this Diff: - Updates build dependency of tbe_data_config* files - Remove `shutil` and `numpy.random` lib as it cause uncompatiblity error. - Add non-OSS test, writing extracted config data json file to Manifold Reviewed By: q10, spcyppt Differential Revision: D79758603

facebook-github-bot · 2025-09-01T22:53:35Z

This pull request was exported from Phabricator. Differential Revision: D79758603

facebook-github-bot · 2025-09-02T14:43:04Z

This pull request has been merged in 46d3ae2.

meta-cla bot added the cla signed label Aug 11, 2025

facebook-github-bot added the fb-exported label Aug 11, 2025

gchalump force-pushed the export-D79758603 branch from 1105d5e to 84a53cc Compare August 12, 2025 23:21

gchalump force-pushed the export-D79758603 branch from 84a53cc to 6328c53 Compare August 12, 2025 23:22

gchalump force-pushed the export-D79758603 branch from 6328c53 to 387a16e Compare August 12, 2025 23:24

gchalump force-pushed the export-D79758603 branch from 387a16e to 712b9b5 Compare August 12, 2025 23:37

gchalump force-pushed the export-D79758603 branch from 712b9b5 to ed8a1a8 Compare August 14, 2025 18:09

gchalump force-pushed the export-D79758603 branch from ed8a1a8 to b1eae51 Compare August 14, 2025 18:10

gchalump force-pushed the export-D79758603 branch from b1eae51 to 7bbc999 Compare August 14, 2025 18:15

gchalump force-pushed the export-D79758603 branch from 7bbc999 to b913d90 Compare August 14, 2025 18:19

gchalump force-pushed the export-D79758603 branch from b913d90 to 657b504 Compare August 18, 2025 15:54

gchalump force-pushed the export-D79758603 branch from 657b504 to e25fe5b Compare August 18, 2025 15:55

gchalump force-pushed the export-D79758603 branch from e25fe5b to cc12771 Compare August 18, 2025 15:58

gchalump force-pushed the export-D79758603 branch from 74b84bb to 36240a3 Compare August 25, 2025 21:35

gchalump force-pushed the export-D79758603 branch from 36240a3 to def4adb Compare August 25, 2025 21:37

gchalump force-pushed the export-D79758603 branch 2 times, most recently from 9a45a11 to 6bdf1b1 Compare August 27, 2025 19:36

gchalump force-pushed the export-D79758603 branch from 6bdf1b1 to 974d301 Compare August 27, 2025 19:37

gchalump force-pushed the export-D79758603 branch from 974d301 to 49d635d Compare August 27, 2025 19:40

gchalump force-pushed the export-D79758603 branch 2 times, most recently from 92cdc2f to 218bfcd Compare September 1, 2025 22:43

gchalump force-pushed the export-D79758603 branch from 218bfcd to d92923d Compare September 1, 2025 22:44

gchalump force-pushed the export-D79758603 branch from d92923d to 4f98ca7 Compare September 1, 2025 22:50

gchalump force-pushed the export-D79758603 branch from 4f98ca7 to b631308 Compare September 1, 2025 22:53

facebook-github-bot closed this in 46d3ae2 Sep 2, 2025

facebook-github-bot added the Merged label Sep 2, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add TBE data configuration reporter to TBE forward (v3) (#4455) #4672

Add TBE data configuration reporter to TBE forward (v3) (#4455) #4672

Uh oh!

gchalump commented Aug 11, 2025

Uh oh!

netlify bot commented Aug 11, 2025 •

edited

Loading

Uh oh!

facebook-github-bot commented Aug 11, 2025

Uh oh!

facebook-github-bot commented Aug 12, 2025

Uh oh!

facebook-github-bot commented Aug 12, 2025

Uh oh!

facebook-github-bot commented Aug 14, 2025

Uh oh!

facebook-github-bot commented Aug 14, 2025

Uh oh!

facebook-github-bot commented Aug 18, 2025

Uh oh!

facebook-github-bot commented Aug 25, 2025

Uh oh!

facebook-github-bot commented Aug 25, 2025

Uh oh!

facebook-github-bot commented Aug 27, 2025

Uh oh!

facebook-github-bot commented Aug 27, 2025

Uh oh!

facebook-github-bot commented Sep 1, 2025

Uh oh!

facebook-github-bot commented Sep 1, 2025

Uh oh!

facebook-github-bot commented Sep 2, 2025

Uh oh!

Uh oh!

Add TBE data configuration reporter to TBE forward (v3) (#4455) #4672

Add TBE data configuration reporter to TBE forward (v3) (#4455) #4672

Uh oh!

Conversation

gchalump commented Aug 11, 2025

Add TBE data configuration reporter to TBE forward call.

Just Knobs for enablement

Environment Variables

Use Cases

Example Usage

Other includes changes in this Diff:

Uh oh!

netlify bot commented Aug 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ Deploy Preview for pytorch-fbgemm-docs ready!

Uh oh!

facebook-github-bot commented Aug 11, 2025

Uh oh!

facebook-github-bot commented Aug 12, 2025

Uh oh!

facebook-github-bot commented Aug 12, 2025

Uh oh!

facebook-github-bot commented Aug 14, 2025

Uh oh!

facebook-github-bot commented Aug 14, 2025

Uh oh!

facebook-github-bot commented Aug 18, 2025

Uh oh!

facebook-github-bot commented Aug 25, 2025

Uh oh!

facebook-github-bot commented Aug 25, 2025

Uh oh!

facebook-github-bot commented Aug 27, 2025

Uh oh!

facebook-github-bot commented Aug 27, 2025

Uh oh!

facebook-github-bot commented Sep 1, 2025

Uh oh!

facebook-github-bot commented Sep 1, 2025

Uh oh!

facebook-github-bot commented Sep 2, 2025

Uh oh!

Uh oh!

netlify bot commented Aug 11, 2025 •

edited

Loading