Johnson Liao
Subscriber

Dear Ashish from Johnson,

Thank you for your information, and after checking this help, it is no matter with the material issue in Mech GPU computing.

we tested different material model, their performance is very signicant for GPU, like elastic, plastic, visoelastic, anand model.

but for hyperleastic, its performance is almost zero.

so, let us curious whether the hyperelastic ( Mooney-Rivlin) is not supported in 2022R2 Mech GPU solver.

also investigate element_226 (structural-thermal-diffusion) , its result only 30% improvement. 

Rezoning only reduced 20% of using pure CPU solver.

regs,

Johnson