Jump to content

system performance on a flip simulation H15


Recommended Posts

Hello and greetings.

 

I have a question with a fluid simulation. I'm just simulating/cache particles to later make the mesh and whatever i want...(SOP write before fluidcompress node) and everything looks cool but the system usage is not as i expected. During the simulation(0.011 particle sep +-12M particles +-20min/frame) the system never use more than 35% CPU and just half of RAM i have (15.7GB of 32GB) and my question is how can i make to use 100% system to speed up the sim (if in some program settings is possible) or something is wrong.

 

Thanks and i'll appreciate every answer u can get me.

 

System spec i7 3820 3.6GHz , 32GB RAM Houdini 15.0.244.16 Indie on windows 7.

Link to comment
Share on other sites

I think that is normal. Especially if you have enabled OpenCL. If so then some calculation is being done by the GPU and does not show up in the CPU monitor. Not all tasks lend themselves to 100% CPU threading. Sometimes a computing task requires that one step be completed before another can begin thus adding a bottle neck to the calculation process that shows up as CPU not being used fully.

 

If you don't have a GPU monitor go ahead and pull one down. I use MSI Afterburner and HWMonitor, both are free.

 

If your motherboard supports it, you might consider overclocking your system. Houdini likes faster Ghz over number of cores when not rendering. A simple google of your CPU shows that others have successfully "upped" the running clock speed on that particular processor. A modest overclock might get you to 4.0Ghz.

Edited by Atom
Link to comment
Share on other sites

It's also important to remember that the i7 3820 has 8 logical cores presented to the system, but only 4 physical cores with SMT which allows 2 threads/core (hyperthreading). This often messes with the CPU%. If you run 8 threads, you'll have 2 threads/core competing for the limited resources of that core. If those threads continually stall one another while waiting on an execution unit, you won't get the full 25% usage for that core. SMT certainly doesn't give you double the performance (more like +10-20%) which is why CPU% based on 8 logical cores isn't very accurate.

 

Atom is also correct - if you're using OpenCL the CPU will spend time waiting for the GPU to finish. There's also some simulation tasks that are single-threaded, and synchronization spots, all of which cut away at full CPU usage.

Link to comment
Share on other sites

Ty for the answers guys. i've monitored the GPU but any process on this so i guess i don't have openCL enabled, but i can't confirm because simulation still working. and it's not the first simulation i've made on houdini with this cpu and another simulations used all threads with 100% usage. the only one difference between them is the version of houdini (from 13 to 15) and houdini 15 running with defaults settings. (any preferences modification)

 

To Atom: i can't overclock cpu because this version don't able me to overclock. i'm searchng some i7 overclockable LGA2011 betther than mine.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...