Openfast: The simulation is running too slowly

Created on 17 Mar 2019 · 6Comments · Source: OpenFAST/openfast

Bug description

When I run the OpenFAST using input file 5MW_Land_DLL_WTurb and WP_VSP_WTurb, I got an unbelievable "Total Real Time". The test for 5MW_Land_DLL_WTurb toke total run time about 120 seconds for 60 seconds' simulation, and the WP_VSP_WTurb toke total run time about 59 seconds for 40 seconds' simulation. Is this situation normal? I am new to OpenFAST, but according to my previous experience with FASTv8, this is unbelievable.

To Reproduce

Steps to reproduce the behavior:

Compile with 'gnu'
Run '5MW_Land_DLL_WTurb' case with 'DLL unused' settings, and run 'WP_VSP_WTurb' case with raw settings.
The problem is that I don't know if this is normal?

Expected behavior

I expect that this simulation can take less time, at least lower than the simulation time.

OpenFAST-v2.0.0-dirty

**************************************************************************************************
 OpenFAST

 Copyright (C)  National Renewable Energy Laboratory
 Copyright (C)  Envision Energy USA LTD

 This program is licensed under Apache License Version 2.0 and comes with ABSOLUTELY NO WARRANTY.
 See the "LICENSE" file distributed with this software for details.
 **************************************************************************************************

 OpenFAST-v2.0.0
 Compile Info:
  - Architecture: 64 bit
  - Precision: double
  - Date: Mar  6 2019
  - Time: 17:19:38
 Execution Info:
  - Date: 11/29/2018
  - Time: 09:28:55

System Information (please complete the following information):

OS: Ubuntu 16.04
Compiler: GFortran 5.4
Compiler settings: Release version and static library and any other things is follow raw setting

Bug

Source

Bing008

All 6 comments

@smilke 120 seconds wall time for a 60 second simulation does seem a bit long, but this could depend on multiple factors. OpenFAST has been configured to run nightly tests on one of NREL’s clusters. Last night the test reported a run time of about 72 seconds as can be seen here.

https://my.cdash.org/viewTest.php?buildid=1622580

ashesh2512 on 17 Mar 2019

👍1

Also keep in mind that double precision is much slower than single precision. Are you comparing the same numerical precision in fast 8 vs openfast?

rafmudaf on 17 Mar 2019

👍1

I have solved the problem by changing double precision to single precision. Thanks for your advise. But I am still confused about how to improve the running speed of FAST as much as possible

Bing008 on 17 Mar 2019

Dear @smilke,

Yes, I would encourage you to compile in single precision when running many simulations. Single precision should be sufficiently accurate and much faster than double.

We've also had a discussion on our forum about how to reduce the time necessary to run FAST / OpenFAST simulations--see: https://wind.nrel.gov/forum/wind/viewtopic.php?f=4&t=1416&p=12380.

Best regards,