| ... | ... | @@ -9,10 +9,12 @@ export EAR_MAX_P_STATE=6 |
|
|
|
export EAR_TIMES=1
|
|
|
|
```
|
|
|
|
|
|
|
|
To complete the learning phase all the kernels will have to be launched in every node of your cluster in the different selected frequencies. You can also maximize the precision of the gathered data by repeating the execution more than 1 time.
|
|
|
|
All the kernels have to be launched in every node of your cluster in the different selected frequencies. You can also maximize the precision of the gathered data by repeating the execution more than 1 time.
|
|
|
|
|
|
|
|
Kernels launch
|
|
|
|
--------------
|
|
|
|
Run the kernels `scripts/learning/kernels_test.sh hostlist` for each island hostlist.
|
|
|
|
|
|
|
|
It is required to use the same name for a kernel launched at different P_STATEs, because the binary which calculates the coefficients employs the name to classify the kernels. The provided script does this automatically. |
|
|
\ No newline at end of file |
|
|
|
It is required to use the same name for a kernel launched at different P_STATEs, because the binary which calculates the coefficients employs the name to classify the kernels. The provided script does this automatically.
|
|
|
|
|
|
|
|
Continue the learning phase following the [next section](5-˗-Coefficients-compute). |
|
|
\ No newline at end of file |
