- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi.
I use OpenVINO R3 and R4 versions .Both versions inference time are low for my tensorflow model but CPU usage is very high(nearly 100%).I use mobilenet model on video with 25 fps and the image dimension is 300*300. How can i reduce CPU usage?
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Mostafa,
Are you using -d CPU ? Then high CPU load is expected.
You could try using -d GPU or use a Movidius NCS2 or other inference device for lower CPU load.
Cheers,
Nikos
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Dear Mostafa, this is normal that the CPU usage is high if you do inference on CPU. Which behavior do you expect from your application? Low CPU usage?
As Nikos mentioned, if you want to want to not use the CPU for inference, use -d GPU or -d MYRIAD for doing the inference on either GPU or Movidius.
Best,
Severine
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I solve it by define OMP_NUM_THREADS = 1 envionment variable .
![](/skins/images/45C6C2D737ED71F4C51F0145C8CB1E9C/responsive_peak/images/icon_anonymous_message.png)
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page