This plug-in is deprecated. Please use the Accelergy Library Plug-In.
An energy estimation plug-in for Accelergy framework. This plug-in provides an entry point for easy creation of various energy tables.
- Install Accelergy framework (if you have not yet).
- Run
pip install .
with the same options you used for installing Accelergy. - The installation installs a command
accelergyTables
to your system.
At least one row of example filled cells are provided in the templates.
Populate the cells with your own data.
Run pip install .
to reinstall the updated data files
Add the root to the provided set of tables by running accelergyTables
.
You should see the following in the printed log:
- The
table-based-plug-ins
is found by accelergy. - The
table-based-plug-ins
identifies a set of tables calledtest_tables
.
Note that most of the provided energy tables do not contain numerical data.
Add the root to your sets of tables to the Accelergy config file by running
accelergyTables -r $myRoot
You can also manually add the root in the ~/.config/accelergy/accelergy_config.yaml
file.
The plug-in will recursively search the subdirectories of the provided root to locate various sets of tables
A YAML identifier file that describes the basic information of your set of tables needs to be created.
set_of_table_templates
contains an example YAML identifier file test.table.yaml
.
The name of the YAML file must end with table.yaml
and following information needs to be provided in the file:
- The name of the set of tables
- The supported technology
- The accuracy of your set of tables
- The path to the directory that contains all your csv data files
You can create your csv tables anywhere in you file structure as long as the path pointer in its identifier file is correct.
set_of_table_templates/data
folder provides the necessary headers for all default primitive component classes.
counter.csv
and regfile.csv
are populated with random data to show what should be filled in the cells.
There are several rules for creating these data files:
- Data tables must be in csv format.
- The name of the file should be
<primitive component class name>.csv
, e.g., SRAM.csv for SRAM energy data. - The necessary hardware attribute names, action names, argument names are the headers(the first row) of the csv, and their corresponding values are filled in the cells.
- The energy column must have a header called "energy".