3 ms·
As RISC-V moves into "high end" embedded (linux capable, raspy like), I suspect that a big obstacle to openness is going to be GPU IP. HW video decoding, displ
by obl 8y ago
As RISC-V moves into "high end" embedded (linux capable, raspy like), I suspect that a big obstacle to openness is going to be GPU IP.
HW video decoding, display protocol driving, 3D, ...
Is there any open hw initiative on that front that I'm not aware of ?
- sanxiyn 8y agoThe only thing I know about is http://nyuzi.org/ http://nyuzi.org/
- phkahler 8y agoFor video we can hope for free AV1 decode hardware possibly from Google. As for normal GPU functionality, I think the way Esperanto is going will be an option. They want to use thousands of RISC-V cores with vector instructions to do graphics. For general desktop compositing and simpler 3D this should be possible with fewer cores. However well that works or not, even that will be a while. In the mean time I think getting LLVM-pipe or the equivalent working well with RISC-V is important and may be the only viable open-source graphics solution at first.
- zokier 8y agoIsn't that all getting bit ahead of ourselves? Wouldn't the first order of business be creating framebuffer graphics and all the requisite HW needed to drive displays (HDMI/(e)DP/LVDS transceivers etc), and then maybe some basic 2D acceleration (bitblt etc)?
- snops 8y agoFor HDMI/DVI, there are a number of external transmitter ICs that take in 24 bit parallel video/I2S audio and spit out HDMI, e.g. http://www.analog.com/en/products/audio-video/hdmidvi-transmitters/adv7513.html http://www.analog.com/en/products/audio-video/hdmidvi-transm...