Author Topic: Open Source dash display  (Read 12551 times)

0 Members and 1 Guest are viewing this topic.

Offline darkspr1teTopic starter

  • Frequent Contributor
  • **
  • Posts: 322
  • Country: zm
Open Source dash display
« on: June 26, 2023, 07:09:39 pm »
Well as per the title I am starting a project on building a really suable LCD based dash replacement.
This has come about as two good friends have the same need , one has a car he got with zero dash (made from compressed card and ants ate it all :-p ) and the other wants to trail competition his late fathers car and needs a better visual feed back (rpm more important than speed, temps/engine status more important than rpm)
I've been dipping my hacking hands into a lot of ECU systems over the years from hacking suzuki/gm ecu's to dumping OBD readers that self destruct so I understand the tech in use on both side there and have also been re-wiring cars for years , Any they are my so called bona fides.


The reason for my post is i've hit a the proverbial compatibility wall , or at least my choice in H/W & S/W that i see for my use to remain open source is well a problem. Many of the graphical systems i want to use have issues ,


Some are open source only to a part and now more modern versions are closed source.
Some reply on specific hardware for the benefits (Beagle bone in built VR gfx with outdated blobs , Nvidia cuda - outdated blobs ) which means either old builds, NDA's and such
H/w must be able to provide a fast booting method or one should be easy to develop (removal of drivers/support for unneeded items , streamlining any u-boot/kernal cross overs like re-enabling enabled h/w like lcd's, usb etc)


project goal is write using something usable across h/w with 3.4 kernels up to 5 in order to re-use a bit of e-waste and reply on the device h/w for acceleration but via a more common API for later progression, try and avoid kernels not in main line(fail here already based on current e-waste)


Onto my Question  which am having a hard time answering for the linux world
what graphics engine should i use , I started with qt5 but further up it broke totally and a lot of re-writes need for 6 as well as logins and such kill that quick. I did think about just pure framebuffer use but i will loose any h/w speed provided and with some older h/w that is a must as they will struggle.


We dont want 3d graphics but cleardisplays and maybe later if the hard can handle some then have the choice to add it.
I see mesa, vulken, x11, wayland , qt(x) and more when i search on the hardware in hand or in general but it's not helping me answer what graphics engines cross over h/w type and across a lot of kernels and can have support for h/w based acceleration  (within reason of closed blobs updated support of kernels etc)




any one make a suggestion and why ?


 


darkspr1te

 

Offline Daixiwen

  • Frequent Contributor
  • **
  • Posts: 364
  • Country: no
Re: Open Source dash display
« Reply #1 on: June 27, 2023, 07:23:06 am »
Maybe look at the graphics part of SDL? It works on a lot of different targets and supports hardware acceleration AFAIK
 
The following users thanked this post: darkspr1te

Offline darkspr1teTopic starter

  • Frequent Contributor
  • **
  • Posts: 322
  • Country: zm
Re: Open Source dash display
« Reply #2 on: June 27, 2023, 07:44:10 am »
Ah, sdl was one of the one supported across most  device. thankyou, i will look into it .




darkspr1te

 

Offline Nominal Animal

  • Super Contributor
  • ***
  • Posts: 6516
  • Country: fi
    • My home page and email address
Re: Open Source dash display
« Reply #3 on: June 27, 2023, 04:22:58 pm »
Running SDL 2 with kmsdrm will let you run using a really minimal userspace: no X or Wayland needed at all, and you can strip off almost all services and implement your own static init for really fast bootup.
 
The following users thanked this post: darkspr1te


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf