Exceeding Memory Limits: Cannot Simulate a Soft-Body/flexcomp object with 36624 voxels about mujoco HOT 10 CLOSED

Did you try this?
https://mujoco.readthedocs.io/en/latest/XMLreference.html#size-memory
https://mujoco.readthedocs.io/en/latest/modeling.html#csize

from mujoco.

@yuvaltassa The mesh file that I'm currently trying to load is attached here (
MSH41_exceed_memory.zip). I tried the advice that you recommended in those above links. As you can see in the code below, I include the line <size memory = "8G"/> in my XML file which I thought would be a reasonable amount of allocated memory:

import mujoco
import mediapy as media

xml = r"""
<mujoco model="3DMBB">
  <size memory = "8G"/>
  <worldbody>
    <flexcomp name="3DMBB" type="gmsh" dim="3" file="converted_3d_mbb_ascii.stl_out_vol.msh">
    <edge equality="true"/>
    </flexcomp>
  </worldbody>
</mujoco>
"""

# Make model and data
model = mujoco.MjModel.from_xml_string(xml)
data = mujoco.MjData(model)

# Make renderer, render and show the pixels
renderer = mujoco.Renderer(model)
mujoco.mj_forward(model, data)
renderer.update_scene(data)

media.show_image(renderer.render())

However, I get this error message below:

ValueError: Error: engine error: mj_stackAlloc: insufficient memory: max = 8573357512, available = 107779936, requested = 8461964232 (ne = 57810, nf = 0, nefc = 57810, ncon = 0)

How do I load in .msh files with a large amount of nodes and elements (~10k nodes, ~40k elements) like this one? Do I need to try run the model on a different machine with more RAM (my machine has 64GB of memory)? Even when I try to load the model with the following setting <size memory = "20G"/> to increase the memory allocation I get the following output:

ValueError: XML Error: unsigned integer with an optional suffix {K,M,G,T,P,E} is expected in attribute 'memory' (or the size specified is too big)
Element 'size', line 3

This indicates that I can't allocate 20GB of memory even though I should be able to on my machine?

from mujoco.

Ah the latter seems like an actual bug in the parser, we should be using long instead of int somewhere...

from mujoco.

@yuvaltassa Thanks for the clarification - what about the first/former issue? Why do I get the error below when it clearly says I have both a max and available amount of memory greater than what was requested?

ValueError: Error: engine error: mj_stackAlloc: insufficient memory: max = 8573357512, available = 107779936, requested = 8461964232 (ne = 57810, nf = 0, nefc = 57810, ncon = 0)

Will I have to wait for the bug that you identified to be fixed before I can model this 3D mesh file in MuJoCo?

from mujoco.

Why do I get the error below when it clearly says I have both a max and available amount of memory greater than what was requested?

Count more carefully, available is smaller by an order of magnitude...

from mujoco.

So when I requested 8GB of memory in the above example, I actually only had 100MB available? Why is this the case? When I print out the available memory on my machine in the terminal, it says I currently have 39GB of memory available (see below):

Surely 39GB of available memory should be more than enough to simulate the mesh file with ~10k nodes, ~40k elements, no?

from mujoco.

Like I said, I think we have some bug related to using an int somewhere where we shouldn't. INT_MAX is 2147483647 which is just 2.14GB ...

from mujoco.

The main bug is now fixed at head. Build from source or wait for next release.

from mujoco.

When will the next release be?

from mujoco.

May 6th

from mujoco.