NVlabs / instant-ngp

Instant neural graphics primitives: lightning fast NeRF and more
https://nvlabs.github.io/instant-ngp
Other
15.88k stars 1.91k forks source link

GUI instant-ngp process uses 25% cpu (100% usage of 1 of the 4 cores ) while empty and doing nothing #1262

Open minch-dev opened 1 year ago

minch-dev commented 1 year ago

I used the latest version built from sources on Ubuntu 20.04. Built with TCNN_CUDA_ARCHITECTURES=61 parameter to fit my gpu. Software used for this build:

Ubuntu 20.04
cmake version 3.25.2
gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
Cuda compilation tools, release 12.0, V12.0.140
Python 3.8.10
Vulkan Instance Version: 1.3.239

I tried --no-train switch (Disables training on startup), didn't help The folder with sources and builds is at the mounted NTFS partition if it matters.

Tom94 commented 1 year ago

If you'd like to throttle the application, you can check the VSync checkbox in the GUI. Otherwise, it goes as fast as it can, which'll mean a very high frame rate at doing nothing if you've got nothing loaded.

minch-dev commented 1 year ago

If I check VSync it's still the same, only framerate drops from 80-something to 60.

Actually it does the same thing even if I untick the "Render" checkbox.