Announcement

Collapse
No announcement yet.

Mesa RADV vs. AMDVLK Radeon Vulkan Performance For July 2021

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

  • birdie
    replied
    Originally posted by damentz View Post

    No, Nvidia is much worse. I just recently switched from a 980 TI to a 6700xt and here's the list of upgrades (besides just overall being faster):
    1. Xorg doesn't use 100% CPU when rendering two applications at once
    2. KDE notifications don't stall the render pipe line and cause dropped frames across the board
    3. Multiple displays don't cause screen tearing on mixed refresh rate (Using TearFree in xorg.conf)
    4. GPU acceleration in firefox doesn't cause other windows to render at less than 30fps
    5. Displays are detected properly and limited RGB and interlaced modes are ignored (yes, nvidia loves randomly picking limited rgb and interlaced modes when available)
    6. Video acceleration is available using VA-API in Chromium + Firefox
    7. Wayland
    If you use Nvidia to get paid using CUDA, then good for you. If you don't, Nvidia is a steaming pile of shit on Linux. Maybe if you configure it for an HTPC that can do only one thing at a time, but the GPU scheduler for nvidia cards is so horrible that you literally can only do one thing at a time unless you suffer massive frame drops and other inconveniences.

    If you're all for checking boxes on a spreadsheet, buy nvidia. If you actually use your system and want to not pull your hair out, use AMD.
    Haven't observed the first five issues. It would be great if you explained how they can be reproduced as NVIDIA is not dumb and deaf and given enough pressure they solve nearly everything. I don't understand the "Displays are detected properly and limited RGB and interlaced modes are ignored" it sounds like NVIDIA does exactly right, why would anyone want limited RGB/interlaced mode? - is this still an issue with NVIDIA drivers 470.74?

    I won't argue with missing HW video acceleration - but this is a general Linux issue, not NVIDIA's, and I don't use Wayland, not intend to in the next five years. I like feature-complete projects.

    Leave a comment:


  • damentz
    replied
    Originally posted by Leopard View Post

    Nvidia does the opposite then. Kudos to Nvidia.

    One driver that does the job right.

    Currently AMD situation on Linux is a mess.

    On NV you get CUDA,Optix, RT, DLSS goodness in one big good package. They even contribute to Proton to make DLSS Windows titles to work.

    While on AMD; you need prop package for OpenCL to work, you need prop package to have RT but prop package has disastrous OpenGL portion compared to RadeonSI and vkd3d workloads doesn't work properly on it, while it does on RADV.

    To me it seems like BS excuses from AMD, Nvidia does what AMD can't just fine.
    No, Nvidia is much worse. I just recently switched from a 980 TI to a 6700xt and here's the list of upgrades (besides just overall being faster):
    1. Xorg doesn't use 100% CPU when rendering two applications at once
    2. KDE notifications don't stall the render pipe line and cause dropped frames across the board
    3. Multiple displays don't cause screen tearing on mixed refresh rate (Using TearFree in xorg.conf)
    4. GPU acceleration in firefox doesn't cause other windows to render at less than 30fps
    5. Displays are detected properly and limited RGB and interlaced modes are ignored (yes, nvidia loves randomly picking limited rgb and interlaced modes when available)
    6. Video acceleration is available using VA-API in Chromium + Firefox
    7. Wayland
    If you use Nvidia to get paid using CUDA, then good for you. If you don't, Nvidia is a steaming pile of shit on Linux. Maybe if you configure it for an HTPC that can do only one thing at a time, but the GPU scheduler for nvidia cards is so horrible that you literally can only do one thing at a time unless you suffer massive frame drops and other inconveniences.

    If you're all for checking boxes on a spreadsheet, buy nvidia. If you actually use your system and want to not pull your hair out, use AMD.

    Leave a comment:


  • pal666
    replied
    Originally posted by Leopard View Post
    Nvidia does the opposite then. Kudos to Nvidia.
    novideo sells closed windows driver in linux shim to imbeciles like you

    Leave a comment:


  • Leopard
    replied
    Originally posted by pal666 View Post
    it's important. just when you pay for 1% of sales, you can expect 1% of developer time
    not all, not even deserved 99%, but most of it
    Nvidia does the opposite then. Kudos to Nvidia.

    One driver that does the job right.

    Currently AMD situation on Linux is a mess.

    On NV you get CUDA,Optix, RT, DLSS goodness in one big good package. They even contribute to Proton to make DLSS Windows titles to work.

    While on AMD; you need prop package for OpenCL to work, you need prop package to have RT but prop package has disastrous OpenGL portion compared to RadeonSI and vkd3d workloads doesn't work properly on it, while it does on RADV.

    To me it seems like BS excuses from AMD, Nvidia does what AMD can't just fine.

    Leave a comment:


  • pal666
    replied
    Originally posted by user1 View Post
    But I was referring to customers that are not regular users like us, for example Stadia, or maybe some enterprise users.
    it's safe to assume that open driver was a requirement for stadia

    Leave a comment:


  • pal666
    replied
    Originally posted by user1 View Post
    Out of curiosity, are there any customers that benefit from the fact that AMDVLK is open source?
    kernel drivers require open userspace. also radeonsi pulls code from amdvlk(and radv pulls code from radeonsi). i.e. you are that benefitting customer

    Leave a comment:


  • pal666
    replied
    Originally posted by jrch2k8 View Post
    i meant clearly FOSS AMDVLK
    me too
    Originally posted by jrch2k8 View Post
    Also purely speaking about Vulkan i think RADV has way bigger userbase than closed AMDVLK since Vulkan hasn't been adapted too widely on enterprise software yet.
    radv is linux only, it can have some chances against amdvlk code only when/if samsung will use it in smartphones
    Last edited by pal666; 26 July 2021, 05:54 PM.

    Leave a comment:


  • pal666
    replied
    Originally posted by Leopard View Post
    One thing i don't understand is why AMDVLK and respectively Pro are so late about supporting features. Supporting features in a timely manner is not important because Linux is a lesser platform?
    it's important. just when you pay for 1% of sales, you can expect 1% of developer time
    Originally posted by Leopard View Post
    Does all man power goes to Windows driver and D3D?
    not all, not even deserved 99%, but most of it

    Leave a comment:


  • pal666
    replied
    Originally posted by user1 View Post
    But aren't we talking about AMDVLK-open which is exclusively for Linux? I think that's what he meant.
    it's built for linux, but it's built from cross-platform code

    Leave a comment:


  • smitty3268
    replied
    Originally posted by Venemo View Post
    I don't wanna speculate, but I think maybe they simply haven't had time to implement all of it yet in LLVM.
    You're right, that's very possible as well. If that's true, I believe it shows that they have zero interest in the driver for individual consumer purposes, though. And basically LLVM RT support will come whenever their ROCm compute team decides they need to support it, and not before.

    Given how well their ROCm support has gone so far, that's deeply troubling. But certainly better than the alternative, if true.

    Leave a comment:

Working...
X