In-core memory requirements exceeding physical memory

AM_20700
AM_20700 Altair Community Member
edited July 2022 in Community Q&A

I'm running a frequency response (direct) analysis that includes pretension.

The model data is:

image

If I use the "-core IN" command, the solver fails with a warning of insufficient storage. When I use the command "-fixlen 200000" to allocate 200 GB of RAM (out of a total of 256 GB), the solver runs but I get the following warning:

image

Is this normal for a model of this size? Requiring 253.7 GB of RAM seems a bit much.

Tagged:

Answers

  • Prasad Reddi_20512
    Prasad Reddi_20512 Altair Community Member
    edited July 2022

    Hi,

    Sometimes memory requirement drastically shoots up depending upon the model contents also. Looking at the modal data image, your model is having lot of S2S contacts. Below things can be checked in your model:

    1. Any physical intersections in the model where you define contacts, if yes, try to remove penetration or issue ADJUST=NO card in the Contact card.

    2. Change all S2S contacts to N2S.

    3. Check for the outputs that you have requested. If there is any scope, restrict in requesting output to selective areas or increase the frequency of punching the results.

    Please post here, if you find any of the above points are working.

    Cheers,

    Prasad.

Categories