Announcement

Collapse
No announcement yet.

NVIDIA 495.44 Linux Driver Released With GBM Support

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

  • timrichardson
    replied
    good news for the climate, since Hell has now frozen-over.

    Leave a comment:


  • Gonk
    replied
    Originally posted by Myownfriend View Post
    DMA-buf support was added in their 470.x drivers back in June.
    Do you recall the specific version? I'm trying to find something about that in past release notes and haven't had any success so far.

    Leave a comment:


  • Alexmitter
    replied
    Originally posted by drakonas777 View Post
    Multiple users: I was experiencing problems with NVIDIA GPU.
    birdie: I was not, so you are wrong or lying.

    This whole thread basically LOL
    every birdie comment ever basically. And can you blame him, look how hot jensen is in his leather jacket, such a great company.

    Leave a comment:


  • oiaohm
    replied
    Originally posted by Linuxxx View Post

    Curious to know: which problems?

    From my years-long experience of using nVidia GPUs in my HTPCs, I've never encountered any problems whatsoever.
    On the contrary, 'nvidia-settings' allows me to easily set the color range to 'limited' [studio levels], so that blacks are not crushed & whites not overblown.

    And just to put this into perspective:
    AMDGPU still doesn't have this very basic support after all these years:


    Anyone left over at AMD that still cares about the above issue?
    The problems described in that bug turn up with Nvidia when used with particular monitors. Strange EDID information and other data being messed up. Out of specification HDMI data transfer makes all kinds of messes.

    Yes using different adapters result in different voltages and other things that can at times over come the issues temporary.


    There are a stack of different bug reports about Nvidia drivers and TVs some are cables, some are the TV being jackass not to HDMi specification.

    Linuxxx this is one of these case of works for me can happen a lot. Yes people suffering from the problem want to blame AMD or Nvidia alone for the problem without considering that the problem is part that stuff is outside specification. Yes different devices have different tolerances in different areas for out of specification. Yes the tolerances can alter with the version of driver you are using as well.


    Yes in AMD bug "If I use an adapter and plug it in via DVI instead of HDMI, it looks fine." this here should straight up have you going this is trouble. DVI to HDMI adapters have a lot higher tolerance to out of specification HDMI signals. The data to calibrate and setup the monitor AMD drivers use is exactly the same between DVI to HDMI and using the HDMI port if there is a difference you have a data loss problem caused by something being out side HDMI specification. In the amd cases with colour space setup if the communication is good on all ports the results should be absolutely identical when they are not you have a communication problem.

    Yes you find the same thing with Nvidia at times that Display port to HDMI works yet HDMI port on the Nvidia does not work again a case like this the Display port to HDMI part just has a higher tolerance to out of specification. Of course then it come why it out specification bad cable, bad monitor, damaged port on graphic card(yes cracked solder poor joints on the HDMi port).

    If you own good to HDMI specification hardware with good HDMI specification cables you basically don't suffer from a complete set of problems. But there are a lot of users out there Linuxxx suffering from these problems.

    Please note that bug report is different to lack of easy means to set color range. The AMD bug you quotes where the graphics card sets up the wrong colour range by default but then when you use a different port with adapter it magically gets it right this is communication problem/out of specification problem.

    Yes out of specification problem the results can be fun thing with Nvidia at times where you set that X monitor should have Y colour space and the next time you start the computer up it does not apply that colour space why because the monitor ID in the EDID is mangled so now it thinks it is a different monitor.

    Linuxxx really if you are after simpler way to set colour space on a monitor you need to find a different bug report to 476 because that not the problem there.

    Leave a comment:


  • WonkoTheSaneUK
    replied
    Originally posted by bosjc View Post

    So you now have a 10 year old NVIDIA potatoe that you can continue to use with X.
    You are Dan Quayle and I claim my $5.

    Leave a comment:


  • Linuxxx
    replied
    Originally posted by Goddard View Post
    Can anyone tell me if this solves the problem with Nvidia and HDMI when connected to TVs?
    Curious to know: which problems?

    From my years-long experience of using nVidia GPUs in my HTPCs, I've never encountered any problems whatsoever.
    On the contrary, 'nvidia-settings' allows me to easily set the color range to 'limited' [studio levels], so that blacks are not crushed & whites not overblown.

    And just to put this into perspective:
    AMDGPU still doesn't have this very basic support after all these years:



    bridgman
    Anyone left over at AMD that still cares about the above issue?

    Leave a comment:


  • drakonas777
    replied
    Multiple users: I was experiencing problems with NVIDIA GPU.
    birdie: I was not, so you are wrong or lying.

    This whole thread basically LOL

    Leave a comment:


  • oiaohm
    replied
    Originally posted by piotrj3 View Post
    Problem is, problem is way bigger then just that. Eg. you want optimus mostly to save energy, but most laptops on battery aren't so energy efficient on Linux as they are on windows, eg. my Haswell (4712mq with gtx850m from MSI) laptop during optimus comfortably could work 5 hours on battery on windows, but couldn't do 3 hours in linux with entire nvidia gpu disabled (only intel igpu was working). Intel power managment is diffrent on linux and windows, and it is bettter on windows a ton, inherently Optimus aims to do same thing, just everybody says "nvidia fuck you" while nobody says "intel fuck you" despite that outcome for end user (me) was the same.

    And if you still claim it is nvidia fault, well later i had one more laptop intel only without nvidia gpu and i saw similar outcome.
    It really not that simple either.
    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

    No the power management is not always better on Windows by a ton over Linux.



    Choice of distribution and desktop environment both are big factors under Linux on battery life.

    Intel did a white paper on why Linux on some laptops are so bad in power usage. ODM/OEM can be to blame. Intel did cover a stack of setting that can be set in the EFI/BIOS that can result in Linux using more power yet windows being fine after you had installed your ODM/OEM drivers as well.

    That paper said if windows performance on laptop was not decent without installing the vendors own drivers you had something that was screwed at the firmware level. Yes cheaper to make a windows driver than have to reflash the bios/EFI. MSI ones are quite common that if you put a stock windows them and don't add MSI drivers battery life is shorter as well due to EFI/BIOS issue. Yes this not being able to work with generic windows can be problem when attempting data recovery out of device using a Windows portable execution environment as well so not a Linux only problem caused by ODM/OEM taking a short cut.

    When it comes to bad battery life there are quite a few people who need the F you message in your case one of the peoples who might be due a F you message is MSI, the Linux distribution, Desktop environment and Intel.

    I wish this was a simple problem with a single party to blame with the battery life problem most cases it not.

    Leave a comment:


  • X_m7
    replied
    Originally posted by piotrj3 View Post

    Problem is, problem is way bigger then just that. Eg. you want optimus mostly to save energy, but most laptops on battery aren't so energy efficient on Linux as they are on windows, eg. my Haswell (4712mq with gtx850m from MSI) laptop during optimus comfortably could work 5 hours on battery on windows, but couldn't do 3 hours in linux with entire nvidia gpu disabled (only intel igpu was working). Intel power managment is diffrent on linux and windows, and it is bettter on windows a ton, inherently Optimus aims to do same thing, just everybody says "nvidia fuck you" while nobody says "intel fuck you" despite that outcome for end user (me) was the same.

    And if you still claim it is nvidia fault, well later i had one more laptop intel only without nvidia gpu and i saw similar outcome.
    Interesting, I also happen to have a MSI laptop with Optimus (Skylake 6700HQ and GTX 960M), and back when I had Windows on it still it only lasted about 3 hours, while on Linux I could do 4. Now that my battery has worn down somewhat it's down to like 3 hours on Linux too. Of course, other stuff changing probably had an effect, for example I used to use a basic window manager only setup, and now I'm back on full KDE.

    Leave a comment:


  • piotrj3
    replied
    Originally posted by Leopard View Post

    Nvidia literally owns laptop market when it comes to dgpu ones.

    Nvidia at that time supported both Linux and Windows.

    Nvidia at that time only supported Optimus on Windows, not Linux.

    All of those decisions made Linux inferior compared to Windows at that time? Yes

    When proper Optimus support landed on Linux? At 2019 August

    Linus's "NV fuck you" reaction happened at 2012.

    So it should be obvious why that "fuck you" was well deserved.
    Problem is, problem is way bigger then just that. Eg. you want optimus mostly to save energy, but most laptops on battery aren't so energy efficient on Linux as they are on windows, eg. my Haswell (4712mq with gtx850m from MSI) laptop during optimus comfortably could work 5 hours on battery on windows, but couldn't do 3 hours in linux with entire nvidia gpu disabled (only intel igpu was working). Intel power managment is diffrent on linux and windows, and it is bettter on windows a ton, inherently Optimus aims to do same thing, just everybody says "nvidia fuck you" while nobody says "intel fuck you" despite that outcome for end user (me) was the same.

    And if you still claim it is nvidia fault, well later i had one more laptop intel only without nvidia gpu and i saw similar outcome.

    Leave a comment:

Working...
X