Fluids

Fluids

Topics relate to Fluent, CFX, Turbogrid and more

No utilization of GPUs on dostributed memory in FLUENT 2022 R2

    • Ammar Ahmed
      Subscriber

      Hello everyone, I am trying to simulate flow through injector using VOF model with K-E RNG (swirl dominated flow) on FLUENT 2022 R2. I have distriburted my simulation on 3 PCs(Dell Precision Tower 7910) using distrubted memory option in FLUENT. Specifications of PC's are as following: PC1: MEMORY=64GB, GEFORCE RTX 3090 (24GB) 28 cores PC2: MEMORY=64GB, GEFORCE RTX 3090 (24GB) 28 cores PC2: MEMORY=32GB, GEFORCE RTX 1660 (6GB) 16 cores Fluent is showing that the GPUs are being enabled but during simulation the GPU is not being utilized but only the CPU and Memory. I have also installed CUDA Version 12 for proper intgration but no success so far. If anyone has any idea on how to resolve this issue, kindly let me know. I am also trying to introduce a new environment variable for proper utilization of GPUs on each device. Kindly tell if its the right approach. The mesh element number is 14million.

    • Kishan Konannavar
      Ansys Employee

      Hello Ammar

      Are you trying to employ the native GPU-solver?

    • Ammar Ahmed
      Subscriber

      I dont have that option when I open FLUENT R2 only the following options are available. I have attached the picture.

    • Ammar Ahmed
      Subscriber

      My simulations run completely on CPU and memory. The GPUs are not utalized not even a bit. Do we have to make an enviornment variable in system variable and assign it a value of 1 so that Cuda 12 forces GPU to be utilized?

       

Viewing 3 reply threads
  • You must be logged in to reply to this topic.