With Kernel 5.14, the #Linux community celebrates 30 years since the birth of the biggest collaborative software project in the world! Here's a look at contributions made by Collaborans for this latest release: https://col.la/lk514 #HEVC #h265 #Hantro #MPEG2 #Embedded #OpenSource
Kernel 5.14: 30 years in the making and still improving

With the release of the 5.14 kernel, the Linux community celebrates 30 years since the birth of the biggest collaborative software project in the world. Here's a look at contributions made by Collabora's developers for this release.

@mntmn in case you're interested in #gstreamer as well: see https://source.puri.sm/pureos/packages/gst-plugins-bad1.0
fand https://source.puri.sm/guido.gunther/livi for a minimal player that I'm using to validate #hantro/#imx8 accelerated decoding with #gtk4 .
PureOS / Packages / Gst Plugins Bad1.0

Pureos fork of https://salsa.debian.org/gstreamer-team/gst-plugins-bad1.0.git

Stitched together a small (µ) video player using #gstreamer and #gtk4 to test accelerated #h264 video decoding on the #librem5 using it's #hantro VPU. It gets along with 20% CPU usage when rendering the 15 Mbps jellyfish demo to a #GdkPaintable which is twice as much as the raw `gst-launch-1.0` pipeline (https://social.librem.one/web/statuses/105799071793016596) but not too bad. It also suspends rendering on screen blank and prevent screen lock during playback so just enough for 'on the road battery drain testing'.
Guido Günther (@[email protected])

Attached: 1 image This is the jellyfish h264 demo on a #librem5 using the #imx8mq's #hantro VPU. Using the CPU we take 300% of CPU time, using the VPU instead we take 10% (and even that can be optimized further). Using the VPU also saves ~1.5W of power. Thanks go to the #gstreamer and #linux kernel folks for making this possible!

New blog post! Mainline Linux gains accelerated video decoding for Microchip's SAMA5D4 https://col.la/avdms #Hantro #V4L2 #Embedded #Linux #OpenSource
Mainline Linux gains accelerated video decoding for Microchip's SAMA5D4

Stateless hardware accelerators provide more control over the codec pipeline resulting in improved power savings among other benefits.

The power difference with #webkit / #epiphany watching the Big Buck Bunny demo with #hantro VPU enabled is about 500mA less power at @1920 x1080 and it renders smoothly while it stutters badly in software.

This is the jellyfish h264 demo on a #librem5 using the #imx8mq's #hantro VPU.

Using the CPU we take 300% of CPU time, using the VPU instead we take 10% (and even that can be optimized further). Using the VPU also saves ~1.5W of power. Thanks go to the #gstreamer and #linux kernel folks for making this possible!

@mntmn Since I know the #imx8mq's performance from the #librem5 (and it's devkit) i'm not worried about Reform' s performance at all for daily / on the road usage. There's also so many improvements coming that can be done on the software side like #hantro video decoding.