How to GPU accelerate in LS-DYNA

- September 2, 2021 at 10:29 am
  
  water5726water5726
  Subscriber
  
  Hi everyone,
  I use ANSYS-DYNA 2021R2 to simulate a drop analysis.
  I opened the function of GPU acceleration, but the GPU cannot be used in calculation.
  Could anyone guide me on how to enable GPU acceleration in ANSYS-DYNA?
  Please help. Thank you!
- September 7, 2021 at 4:13 pm
  
  Ram Gopisetti
  Ansys Employee
  
  what's your graphic card series, the supported NVIDIA are QUADRO series and only for the Mechanical applications.
  GPU is not taken into consideration for LSDYNA-WB in solving, you need to move to LSRUN along with liblstc_gpu library included from the LST-FTP to run the models on GPU .
  Moreover, GPU is not supported on explicit but for implicit you can run and you wont see much of significant boost in solve time by running dyna on GPU. Why not try MPP version of the dyna for faster solve times.
  Cheers, Ram
- September 13, 2021 at 4:11 am
  
  water5726water5726
  Subscriber
  
  Hi Ram Thank you for your answer.
  
  I have searched the article is talking about GPU accelerating in LS-DYNA, it seems to have good performance in implicit solving.
  The article is published in 2011.
  https://www.dynamore.de/de/download/papers/konferenz11/papers/session6-paper1.pdf
  
  I am evaluating which method can effectively speed up the calculation.
  Regarding your description, GPU won't see much of significant boost so using MPP version is better than GPU, right?
  
  Thanks!