TUFLOW Runner: Difference between revisions
Content deleted Content added
No edit summary |
mNo edit summary |
||
| (9 intermediate revisions by 3 users not shown) | |||
Line 1:
==Introduction==
The TUFLOW Runner provides a simple GUI for creating a queue of TUFLOW and TUFLOW FV simulations which will be run depending on available licenses and resources. The runner is available from the [https://gitlab.com/tuflow-user-group/tuflow/simulation-management/tuflow-runner TUFLOW GitLAB User Group] and is also bundled with the TUFLOW QGIS plugin. The TUFLOW Runner can replace the need for batch files.
Line 30 ⟶ 29:
==Setting up the TUFLOW Runner==
Launch the runner.
The tool can be found via the QGIS menu '''Plugins >> TUFLOW >> RUN >> Launch TUFLOW Runner'''. To set up:▼
* If running from python, launch the "main.py" script.
▲
Before using the runner, the computer settings need to be specified. You can do this by Choosing '''Settings''' option from the Menu.
[[File:Setting Runner 1.jpg|550px|]]<br>
[[File:Setting Runner 2.jpg|550px|]]<br>▼
The resources tab at the top of the dialog is to specify the machine limits for the number of GPU cards and CPU threads. These will generally only be limiting if you are running both TUFLOW and TUFLOW FV at the same time. Otherwise, model specific limitations will prevent exceeding these limits.
To discover a computer's NVIDIA GPU hardware, see <u>[[Console_Window_GPU_Usage | NVIDIA GPU Hardware and Usage]]</u>.
TUFLOW and TUFLOW FV each have their own tab to specify executable locations and configuration parameters. For each configuration, you can choose whether to run with one or more GPUs and how many CPU threads to make available. These settings will override settings specified in the input files (for TUFLOW this requires that the "write hardware flags" is specified and you are using 2023-03-AB release or later). Because TUFLOW may use multiple threads during the initial setup phase but a GPU card during the hydraulic calculations when doing GPU simulations, it is recommended to set the number of CPU threads to the number of cores available on the machine and the reserve value to 0. CPU simulations will generally use the same number of threads for reserved and the primary CPU thread setting. Set the configurations to not exceed available licenses.
If you have 2 GPUs and 24 threads available with licenses to support them, the configurations could be setup multiple ways to maximum the complete resources:
*Option 1 - Three configurations where two have a GPU, 12 threads, and 0 threads reserved and a configuration with 0 GPUs with 24 threads and 24 threads reserved.
*Option 2 - Two configurations where one has 2 GPUs, 12 threads, and 0 threads reserved and a configuration with 0 GPUs with 24 threads and 24 threads reserved.
For either of these options, simulations would try to run on the first specified configuration first so you would want the GPU configurations listed first to take advantage of these resources first.
Even if your licensing and resources allow, it may not be worthwhile to run CPU configurations because of the speed advantages of the GPU hardware.
▲[[File:Setting Runner 2.jpg|550px|]]<br>
==Running the tool==
Line 44 ⟶ 59:
*Identify the folder that contains the simulation files ('''Base folder''') and select the simulation file. This must be .tcf for TUFLOW Classic/HPC or .fvc for TUFLOW FV.
[[File:Running the tool 1.jpg|550px|]]<br>
*Click on the '''Add''' button to add a simulation to the queue. If a TUFLOW model includes Events (~e~) and/or Scenarios (~s~) by clicking on the '''Add''' button a '''dialog''' (please see image below) will come up to add model’s Scenarios and/or Events. '''
[[File:Running the tool 2.jpg|550px|]]<br>
'''Please note:''' For TUFLOW model which includes Scenario and/Events, the (~s~) and/or (~e~) must be included in the simulation filename.
* Then, toggle on '''Run''' to run the simulation.
[[File:Running the tool 3.jpg|750px]]
==Running Multiple Simulation==
The runner will launch as many simulations as available in the queue depending on computer resources and model configurations. As model simulations finish, new simulations will be launched as resources become available.
==Troubleshooting==
| |||