Announcement

Collapse
No announcement yet.

AMD Radeon Open-Source Linux GPU Driver Performance: 2020 vs. 2021

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • qarium
    replied
    Originally posted by libv View Post
    I'll try to take this seriously when my OSHW laptop arrives.
    sure it looks like it is not seriously ... but believe me back-channel communication in the secret intel world...

    and also arrangments in the secret society world are in fact very serious.

    "So why did you and others fork RadeonHD then?"

    i think the reason why they did what they did is very simple

    because they need working DRM for Hollywood and this could not be done fully copensource

    also they need a way to restrict hardware features by firmware the Vega20 Radeon VII is the best example the chip does have 4096 shaders but the vega only has 3500 shaders... and only APPLE did sell the 4096 shader version...

    it is plain and simple not possible to restrict the hardware if the firmware is opensource...

    the last 10 years what happened to the true opensource(with opensource firmware) is similar to electric cars...

    what we have with amd today is not true opensource but it is also not closed source...

    same with electric cars what we have today with rechargable battery electric cars is not the true vision of Nicola Tesla...

    because the true vision was "Free Energy" cars means cares that generate their energy by zero point energy means they do not neet battery and they do not need to recharge...

    just forget the last 10 years its nothing more than bullshit show and you know it.

    just be sure of one thing: the next 10 years we will get true Free energy cars with unlimited range without recharge
    and also we will get true opensource hardware with true opensource drivers...

    hollywood is over. and also microsoft is over.

    Leave a comment:


  • libv
    replied
    Originally posted by libv View Post
    So why did you and others fork RadeonHD then?
    Surely Mr Bridgeman was not right when he said that you could not understand the structure of the RadeonHD driver.

    Leave a comment:


  • libv
    replied
    Originally posted by agd5f View Post

    It really depends on the team and the code base. Teams that mostly focus on windows tend to be more focused on closed source and packaged driver releases since that is how the process works on windows. Teams that focus on Linux understand the open source ecosystem. There was never any AMD/ATI struggle with regard to open source (at least as long as I've been at AMD). Early on there was not much funding for consumer Linux in general since the market was small and AMD was not in the best financial shape at the time.
    So why did you and others fork RadeonHD then?

    Leave a comment:


  • libv
    replied
    Originally posted by qarium View Post

    I am 100% sure we will soon have the funding but to depent on existing hardware to try to liberate existing hardware is clearly the wrong way. the FLOSS/Opensource community need their own hardware and CPU side is not that big problem you can buy a power9 system but GPU side is much more """dark""". and we will have some light as soon as LibreSOC hit the market.

    about funding in my experience to bring funding by selling hardware is much more easy than to fund developers to liberate hardware... this means as soon as we sell liberated hardware the money comes by selling the hardware.

    "hen why would anyone in his right mind even think about developing silicon from scratch?"

    soon opensource friendly people will take over ALL 210 country nations government in the world.

    and then we can spend billions of Gold backed money on this project.
    I'll try to take this seriously when my OSHW laptop arrives.

    Leave a comment:


  • agd5f
    replied
    Originally posted by Drago View Post

    Thanks for your effort and contributions, nevertheless!
    I wonder how much of this "open source" resistance is just shame of bad code, and fear of being sued of using something patented (intentionally or unintentionally)?
    It really depends on the team and the code base. Teams that mostly focus on windows tend to be more focused on closed source and packaged driver releases since that is how the process works on windows. Teams that focus on Linux understand the open source ecosystem. There was never any AMD/ATI struggle with regard to open source (at least as long as I've been at AMD). Early on there was not much funding for consumer Linux in general since the market was small and AMD was not in the best financial shape at the time.

    Leave a comment:


  • qarium
    replied
    Originally posted by libv View Post
    We cannot even get proper funding for the people who actually create open source support for existing hw.
    How on earth are we going to fund developing hw from scratch, and then getting said hw in the hands of consumers?
    Then why would anyone in his right mind even think about developing silicon from scratch?

    RadeonHD was the closest we ever got to getting solid fully open driver and firmware support for high end graphics hardware.

    Adding full native coreboot support was right on my radar at the time: i was contributing to coreboot before i joined suse and worked on freeing ATI. When suse threw out 20% of their employees here in Nuremberg, i was given time to continue the work that i halted, and i added the first native graphics support to coreboot, replacing the last binary blob on a consumer motherboard. The way i wrote RadeonHD allowed us to get any discrete graphics card supported with some limited information, and a replacement of ASICInit (~400loc with high degrees of code sharing between multiple boards).

    But this was not to be. AMD lost the internal struggle with ATI, and it took them the best part of a decade to recover. And then there were other companies not satisfied with suse or yours truly getting this prestigious project.

    Each time I go there, whether it is finding structure in display drivers and ridding the world of int10, whether it is open sourcing a big high end gpu vendor, whether it is opening up the whole ARM GPU world, the story always is the same.
    I am 100% sure we will soon have the funding but to depent on existing hardware to try to liberate existing hardware is clearly the wrong way. the FLOSS/Opensource community need their own hardware and CPU side is not that big problem you can buy a power9 system but GPU side is much more """dark""". and we will have some light as soon as LibreSOC hit the market.

    about funding in my experience to bring funding by selling hardware is much more easy than to fund developers to liberate hardware... this means as soon as we sell liberated hardware the money comes by selling the hardware.

    "hen why would anyone in his right mind even think about developing silicon from scratch?"

    soon opensource friendly people will take over ALL 210 country nations government in the world.

    and then we can spend billions of Gold backed money on this project.

    Leave a comment:


  • Drago
    replied
    Originally posted by libv View Post
    We cannot even get proper funding for the people who actually create open source support for existing hw.
    How on earth are we going to fund developing hw from scratch, and then getting said hw in the hands of consumers?
    Then why would anyone in his right mind even think about developing silicon from scratch?

    RadeonHD was the closest we ever got to getting solid fully open driver and firmware support for high end graphics hardware.

    Adding full native coreboot support was right on my radar at the time: i was contributing to coreboot before i joined suse and worked on freeing ATI. When suse threw out 20% of their employees here in Nuremberg, i was given time to continue the work that i halted, and i added the first native graphics support to coreboot, replacing the last binary blob on a consumer motherboard. The way i wrote RadeonHD allowed us to get any discrete graphics card supported with some limited information, and a replacement of ASICInit (~400loc with high degrees of code sharing between multiple boards).

    But this was not to be. AMD lost the internal struggle with ATI, and it took them the best part of a decade to recover. And then there were other companies not satisfied with suse or yours truly getting this prestigious project.

    Each time I go there, whether it is finding structure in display drivers and ridding the world of int10, whether it is open sourcing a big high end gpu vendor, whether it is opening up the whole ARM GPU world, the story always is the same.
    Thanks for your effort and contributions, nevertheless!
    I wonder how much of this "open source" resistance is just shame of bad code, and fear of being sued of using something patented (intentionally or unintentionally)?

    Leave a comment:


  • Azpegath
    replied
    Michael why have you stopped testing with Mad Max? Isn't that more demanding than some of the games you used in the tests? I also has a Vulkan and an OpenGL renderer.

    Leave a comment:


  • libv
    replied
    We cannot even get proper funding for the people who actually create open source support for existing hw.
    How on earth are we going to fund developing hw from scratch, and then getting said hw in the hands of consumers?
    Then why would anyone in his right mind even think about developing silicon from scratch?

    RadeonHD was the closest we ever got to getting solid fully open driver and firmware support for high end graphics hardware.

    Adding full native coreboot support was right on my radar at the time: i was contributing to coreboot before i joined suse and worked on freeing ATI. When suse threw out 20% of their employees here in Nuremberg, i was given time to continue the work that i halted, and i added the first native graphics support to coreboot, replacing the last binary blob on a consumer motherboard. The way i wrote RadeonHD allowed us to get any discrete graphics card supported with some limited information, and a replacement of ASICInit (~400loc with high degrees of code sharing between multiple boards).

    But this was not to be. AMD lost the internal struggle with ATI, and it took them the best part of a decade to recover. And then there were other companies not satisfied with suse or yours truly getting this prestigious project.

    Each time I go there, whether it is finding structure in display drivers and ridding the world of int10, whether it is open sourcing a big high end gpu vendor, whether it is opening up the whole ARM GPU world, the story always is the same.
    Last edited by libv; 07 January 2022, 06:41 PM.

    Leave a comment:


  • qarium
    replied
    Originally posted by libv View Post
    First off, no-one seems to care.
    i do care about closed source firmware... but honestly i am clueless how to fix this problem.

    i think right now LibreSOC is our best bet about this topic isn't it ?

    the FLOSS/Opensource community plain and simple need their own CPU and their own GPU ...

    we can ask 1 million times please nvidia open source the driver or 1 million time please amd opensource the firmware..
    this will never happen.

    so IBM power9/10 with LibreSOC gpu looks like the best future...

    Leave a comment:

Working...
X