mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
66ea76ec44
Modified the flag PYTORCH_TUNABLEOP_NUMERICAL_CHECK, so that it accepts the numerical tolerances in the format atol_rtol as compared to the previous 0 and 1. Retains previous functionality with default values as well. Pull Request resolved: https://github.com/pytorch/pytorch/pull/163079 Approved by: https://github.com/naromero77amd, https://github.com/jeffdaily
93 lines
1.2 KiB
Markdown
93 lines
1.2 KiB
Markdown
```{eval-rst}
|
|
.. currentmodule:: torch.cuda.tunable
|
|
```
|
|
|
|
# TunableOp
|
|
|
|
## Overview
|
|
|
|
```{eval-rst}
|
|
.. automodule:: torch.cuda.tunable
|
|
```
|
|
|
|
## API Reference
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: enable
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: is_enabled
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: tuning_enable
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: tuning_is_enabled
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: record_untuned_enable
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: record_untuned_is_enabled
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: set_max_tuning_duration
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: get_max_tuning_duration
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: set_max_tuning_iterations
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: get_max_tuning_iterations
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: set_filename
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: get_filename
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: get_results
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: get_validators
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: read_file
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: tune_gemm_in_file
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: mgpu_tune_gemm_in_file
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: set_rotating_buffer_size
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: get_rotating_buffer_size
|
|
```
|
|
|
|
```{eval-rst}
|
|
.. autofunction:: set_numerical_check_tolerances
|
|
``` |