Hi List, I'm new to both gstreamer and Raspberry Pi, so please bear with me. Here are a few simple questions you can probably answer from the top of your head, and thanks in advance!
My config is Raspberry Pi B+ (single core 700MHz) and PlayStation Eye (USB camera + microphone), running gstreamer-0.10. Basically I ran 3 tests with the video capture with the following command to compare CPU usage: $ watch -d -n 2 ps -C gst-launch-0.10 -o %cpu,%mem,cmd Test 1 (converting to RGB and sample at 30 fps): $ gst-launch-0.10 v4l2src ! autovideoconvert ! videorate ! video/x-raw-rgb, format=RGB3, depth=24, framerate=*30/1*, width=320, height=240 ! autovideoconvert ! fakesink CPU was at 25%. Then I wondered whether I can improve over this 25%, one nature thought was to lower the sample rate from 30/1 to 5/1: Test 2 (sampling at 5/1 instead of 30/1): $ gst-launch-0.10 v4l2src ! autovideoconvert ! videorate ! video/x-raw-rgb, format=RGB3, depth=24, framerate=*5/1*, width=320, height=240 ! autovideoconvert ! fakesink CPU usage jumped to 70%!! Note that this camera is able to do 120 fps at 320x240 (according to Wikipedia <http://en.wikipedia.org/wiki/PlayStation_Eye>), or 60 fps at 320x240 according to v4l2-ctl output (appended below), so I ran test 3: Test 3 (sampling at 60/1 instead of 30/1): $ gst-launch-0.10 v4l2src ! autovideoconvert ! videorate ! video/x-raw-rgb, format=RGB, depth=24, framerate=60/1, width=320, height=240 ! autovideoconvert ! fakesink CPU was around 50%. Then here are 3 questions: 1. Could someone explain to me why this CPU usage: 5fps: 70% 30fps: 25% 60fps: 50% 2. My application needed RGB, that's why I was converting to it (x-raw-rgb). To do computation on the *appsink* end, currently I do this: still sample at 30 fps (because CPU is lowest at 25%), but in my application for every 6 frames I drop 5, thus I'm in fact processing at a speed of 5 fps. This leaves enough CPU power to my appsink. Does this make sense? 3. Is there any other way to get lower than 25% CPU? 4. This deals with audio, that has been bugging me a bit. If I print the duration of each audio buffer, like the following in python: print appsink.emit('pull-buffer').duration/1000000.0 For "alsasrc" I get a constant 10ms, but for "pulsesrc" I've seen 26ms, 36ms, and 46ms in different runs! Any idea? Thank you! Xuchen P.S., The camera looks like this: $ v4l2-ctl --all Driver Info (not using libv4l2): Driver name : ov534 Card type : USB Camera-B4.09.24.1 Bus info : usb-bcm2708_usb-1.2 Driver version: 3.12.22 Capabilities : 0x85000001 Video Capture Read/Write Streaming Device Capabilities Device Caps : 0x05000001 Video Capture Read/Write Streaming Priority: 2 Video input : 0 (ov534: ok) Format Video Capture: Width/Height : 640/480 Pixel Format : 'YUYV' Field : None Bytes per Line: 1280 Size Image : 614400 Colorspace : SRGB Streaming Parameters Video Capture: Capabilities : timeperframe Frames per second: 60.000 (60/1) Read buffers : 2 User Controls brightness (int) : min=0 max=255 step=1 default=0 value=0 flags=slider contrast (int) : min=0 max=255 step=1 default=32 value=32 flags=slider saturation (int) : min=0 max=255 step=1 default=64 value=64 flags=slider hue (int) : min=-90 max=90 step=1 default=0 value=0 flags=slider white_balance_automatic (bool) : default=1 value=1 exposure (int) : min=0 max=255 step=1 default=120 value=253 flags=inactive, volatile gain_automatic (bool) : default=1 value=1 flags=update gain (int) : min=0 max=63 step=1 default=20 value=255 flags=inactive, volatile horizontal_flip (bool) : default=0 value=0 vertical_flip (bool) : default=0 value=0 power_line_frequency (menu) : min=0 max=1 default=0 value=0 sharpness (int) : min=0 max=63 step=1 default=0 value=0 flags=slider Camera Controls auto_exposure (menu) : min=0 max=1 default=0 value=0 flags=update
_______________________________________________ gstreamer-embedded mailing list gstreamer-embedded@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/gstreamer-embedded
