This is always a good reminder to remember that if you have a newer system and older card that it may not be possible to work or to be vigilant before upgrading/plan ahead. In general if you are on a less than 10 year old system it probably has EFIonly or maybe CSM/Legacy Mode is possible around that time. If your system supports EFIonly, then the card you install MUST have an EFIROM installed on it. Many cards, even older ones are possible to flash and u........

Be careful with this and make sure you know what you're doing. In this example we took an Old AMD Instinct Mi25 GPU and flashed to WX9100, because both cards are essentially the same hardware (gfx900) but the WX9100 firmware will enable the miniDP port and also higher power and performance (Mi25 is capped at 110W vs WX9100 170W and of course makes use of the display). Most guides on the internet for some reason have people using a physical ROMprogrammer, which works but is........

compiling ollama from source by using a docker container with a prebuilt rocm 6.1 dev environment: rocm/dev-ubuntu-22.04:6.1-complete Install these packages first apt install cmake git apt install software-properties-common add-apt-repository ppa:longsleep/golang-backports apt install golang-go Clone the repo git clone https://github.com/ollama/ollama /rtt/o........

The latest Virtio drivers virtio-win-0.1.285-1 seem to have a bug where you can select several resolutions but not 1080P (1920x1080). If you need 1080P go back to an earlier one from about 2023 such as virtio-win-0.1.240-1 and remember to uninstall the newer version first. After this should you be able to see 1080P again.........

This has been an ongoing issue with X11 compositers like marco, unity and cinnamon etc... Ifind especially with an Nvidia GPU that as the system runs longer after a number of weeks or a number of months the GUI just gets slower and slower. It's not something you'll instantly notice but just a hint of lag here or there that gets worse. One hint you'll find is by watching nvidia-smi and checking the utilization you will start seeing random spikes of 80-100% even without maxi........

nvidia-smi Tue Apr 7 15:48:11 2026 +-----------------------------------------------------------------------------------------+ | NVIDIA-SMI 570.211.01 Driver Version: 570.211.01 CUDA Version: 12.8 | |-----------------------------------------+------------------------+----------------------+ | GPU Name &nb........

This is not scientific but Itook an old Xeon V4 workstation and an old Quadro GPU and compared the Whisper times for the same approximately 4 minute audio, transcribing in English. Here are the times which is slightly surpising. I used the small model with this command: whisper vid.mp4 --model small --output_format srt --language en Of course to test the GPU I added the --device cuda flag and confirmed GPU was being use........

The default version is CPU only, which usually is not what we want if we have a GPU: pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121 Now cuda enabled pytorch apps like whisper etc... can take advantage of CUDA finally! Check it is enabled False means it is not working, or at least your cuda/nvidia driver setup is installed and True means good. python3 -c........

I find this this very handy for video projects which involve nvidia cards. Imagine if you need to know if the card support a specific codec, resolution or even hardware encoding or decoding, this is a huge timesaver. One caveat is to double check forums and other reports/driver info as sometimes support is listed but may not be present in specific driver versions or OS's (eg. sometimes h265 is supported in the Windows driver but not Linux). This is one area that Nvidia i........

In a lot of distros, vdpau which accelerates video through the GPUis often not installed even in recent distros. This will result in Firefox with Youtube dropping frames when you check the "Stats for Nerds" and to the eye as jerky or slower video playback due to the dropped frames. Some sites will say that because you are watching VP09 that there is no GPU support, but with vdpau installed and enabled, many GPUs will still be able to use hardware decoding for 4K........

This seems to happen in many different drivers but it happened more often in newer versions such as 530 vs 525. Then nvidia-modeset goes to 100% There are many reports of this appearing since driver 4.70 and I can confirm I've seen this in various machines. https://forums.de........

The issue is that Docker images are stripped down, so many tools and even python3 is missing, so you'll have to build or update the actual image yourself. I assume you have started an image with something like this and that you have the Nvidia Toolkit installed (assuming you are using GPUs). If you're not using nvidia just remove --runtime=nvidia --gpus all. docker run -it --runtime=nvidia --gpus all ubuntu bash These works for most images li........

Welcome to our in-depth guide on configuring NVIDIA GPUs with Docker on Ubuntu. This post is tailored for developers, data scientists, and IT professionals who are looking to leverage the power of NVIDIA's GPU acceleration within Docker containers. Whether you're working on machine lea........

Do you have access to some old Tesla GPUs and want to try them at home in your Desktop or old Server? Some people have wanted to try these units for gaming but keep in mind they have no video out port, they were only meant for AI applications such as Deep Learning. The easiest way by far is to choose an AIservice that has everything ready to go, perhaps with a bunch of Docker or Kubernetes containers. This can be done with Cloud services like Google, Amazon and many........

It may appear to be an Xorg or lightdm/gdm/mdm error but in reality for many users with this issue, it's a driver conflict and issue. I had a system that had two GPUs, an Intel and Nvidia GPU. The only thing that got it working was to remove the nouveau driver and blacklist it so it never came back, then the Intel GPU works fine without these issues. Solution sudo rmmod nouveau add nouveau/other driver to blacklist edit th........

Here is how Ifixed it on a Mint/Ubuntu install 1.) First download the latest AMDGPU-Pro driver from here: https://www.amd.com/en/support Navigate to your relevant video card: 2.) Download the installer One issue is that by default they give you a version for the latest version of........

This error implies that there may be an issue with Xorg or maybe your NVIDIA GPU cannot start or initialize: 35 laptop kernel: [ 2031.857704] nvidia: loading out-of-tree module taints kernel. 35 laptop kernel: [ 2031.857724] nvidia: module license 'NVIDIA' taints kernel. 35 laptop kernel: [ 2031.857725] Disabling lock debugging due to kernel taint 35 laptop kernel: [ 2031.873280] nvidia: module verification failed: signature a........

(firefox:9562): LIBDBUSMENU-GLIB-WARNING **: Unable to get session bus: Failed to execute child process "dbus-launch" (No such file or directory) ExceptionHandler::GenerateDump cloned child 9743 ExceptionHandler::WaitForContinueSignal waiting for continue signal... ExceptionHandler::SendContinueSignalToChild sent continue signal to child [Parent 9562, Gecko_IOThread] WARNING: pipe error (40): Connection reset by peer: file /build/firefox-EymEXX/fire........

Just install these packages and restart the VM: 1.) Enable guest-utils on the host side: sudo apt install virtualbox-guest-utils virtualbox-guest-x11 2.) Enable guest editions on the VMside This must be done for each VM that you want to have the guest additons for accelerated GPU performance and for drag and drop/clipboard sharing First insert the Guest Addtions CD image........

I used to believe that for Desktops especially that the "ondemand" CPUfrequency changing that kernels included with Ubuntu and Debian based distros have would be sufficient for snappy performance. However, you can feel the lack of performance on the fastest computer if you have ondemand. A lot of times even under high load 100% of your CPUfrequency in MHz will not be used. For example a 2.8Ghz CPUmay only run at 1.8MHz or even .9GHz. Now........

radeon_dp_aux_transfer_native: 158 callbacks suppressed The simple answer is that radeon driver sucks and is a remnant of typical AMD/ATI issues. Use AMDGPUif you support it:........

Occasionally my whole screen locks up and I cannot even swith to the console and I find this in my syslog: *-display description: VGA compatible controller product: Mullins [Radeon R3 Graphics] vendor: Advanced Micro Devices, Inc. [AMD/ATI]  ........

To enable amdgpu we have to set special kernel boot parameters. The easiest way is to make it permanent and apply to all kernels (no messing around with grub.cfg) so we'll edit those defaults in /etc/default/grub by changing the GRUB_CMDLINE_LINUX_DEFAULT parameter. After that don't forget to run "update-grub"to apply it (otherwise amdgpu will never be enabled). The opensource driver for AMD is really quite impressive these days, as we show it works with ve........

The easiest way to know if your videos are playing with GPU acceleration are to watch the process of xplayer, mpv or whatever you are playing. The CPU usage should be no more than 10% for that process/program if it is using acceleration. Let's manually play with vdpau to make sure it works before we make it permanent: First make sure you have libvdpau installed: sudo apt install vdpau-driver-all If yo........

I find that the default settings for the radeon driver that is applied to most AMD cards is horrible. For example by default TearFree is not enabled and it causes videos to have some kind of square artifacts. Here are the settings I have found most suitable for AMD cards: You need to create file in the following path and restart Xorg or your computer to apply it: *Beware that making a mistake here will possibly make your computer........

You can do other things but this particular script is just to set all AMD cards to 80% fan speed (remember this script needs to applied everytime you reboot). You could set it is a cron or just throw it into /etc/rc.local basepath=/sys/class/drm for hwmon in `ls -1 /sys/class/drm|grep card[0-99]$`; do echo card=$hwmon; hwmonname=`ls $basepath/$hwmon/device/hwmon|grep hwmon[0-99]` hwmonpath=$basepath/$hwmon/device/hwmon/$hwmonname echo "ec........

sudo apt-get install hwloc-nox Reading package lists... Done Building dependency tree Reading state information... Done The following NEW packages will be installed: hwloc-nox 0 upgraded, 1 newly installed, 0 to remove and 530 not upgraded. Need to get 151 kB of archives. After this operation, 453 kB of additional disk space will be used. Get:1 http://archive.ubuntu.com/ubunt........

[root@localhost:~] BootModuleConfig.sh echo host-ind nfcd........

sudo apt-get install libcurl4-openssl-dev git build-essential autotools-dev autoconf libcurl3 sudo apt-get install libcurl4-gnutls-dev git clone https://github.com/wolf9466/cpuminer-multi sudo apt-get install cmake libpthread-* libmicrohttpd-dev libssl-dev libhwloc-dev git clone https://github.com/fireice-uk/xmr-stak-cpu.git make install cd bin chmod +x xmr-stak-cpu ./xmr-stak -O xmr........

./nsgpucnminer: error while loading shared libraries: libOpenCL.so.1w: cannot open shared object file: No such file or directory ........

You can search for this bug and it seems like it may be related to ecryptfs and is many years old. The symptoms are that you return to the computer and the screensaver was active or the screen was asleep/black and it doesn't seem to come back. But you check by SSH the computer is running fine and are frustrated you'll lose your running programs and have to reboot. There is a simple solution: Ctrl + Alt + F1 Ctrl +Alt + F8 Ba........

Normally lspci will show you just like this and would suggest they are exactly the same card: 1a:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Ellesmere [Radeon RX 470/480/570/580] (rev e7) 1c:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Ellesmere [Radeon RX 470/480/570/580] (rev e7) lspci -vnn is the answer As we can see one is a Gigabyte and the other is an MSI card. Wha........

When using the .sh script the rendering doesn't work after an upgrade of related packages to kdenlive. kdenlive (kdenlive:amd64 (4:17.04.1+git201705191233~ubuntu16.04.1)) with affine or composite transitions was fine but is now broken during the time of transition it is just a white screen. The previous version was fine: kdenlive:amd64 (4:17.04.1+git201705191233~ubuntu16.04.1 But now I can't find it or install it:........

I am using a GTX 1060 but replace the download for the driver with the correct/current version for your particular card by visiting: http://www.nvidia.com/Download/index.aspx?lang=en-us yum install automake curl openssl-devel libcurl-devel gcc gcc-c++ yum -y install kernel-devel-`uname -r` yum -y install unzip #the........

When running cudaminer once it tries to initialize the card the entire screen freezes. The computer itself is still running but the Xorg is done for, you cannot even switch to another console window and must reboot (even an mdm or Xorg restart does not help). At first cudaminer will give you these errors: stratrum_recv_line failed ...retry after 15 seconds GPU #0: Geforce 210 with compute ca........

At first my BIOS said the card may not work right because there is no more option ROM space. I disabled the Option ROM for both LSI 1068 and 2008 chipsets, Network Boot ROM and most other PCI slots, Serial Port, etc... and the message went away but the card still does not work properly. But it still cannot initialize the card properly(does not work): [ 33.943272] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:........

For years I've always built cheap systems believing that there is little difference in more expensive components when it comes to reliability and quality, I generally believe this still except for Power Supplies. I've always bought cheap cases with nice sounding 350-550W stock/cheap/crap power supplies and haven't had any issues for the most part until recently. One such case is an NGEAR case with a 550W Optimax power supply, I always read that these supplies don't produce the........

I'm running Ubuntu 10.10 with Asus NVIDIA GT430 2.6.35-32-generic-pae #65-Ubuntu SMP Tue Jan 24 14:06:16 UTC 2012 i686 GNU/Linux Nvidia binary driver: 260.19.06 The screen sometimes completely freezes and locks up or everything starts going very slow: [1903398.100007] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context [1903402.373210] NVRM: Xid (0001:00): 53, CMDre 00000000 00000080 00000000 00000005 00........

www.antec.outervision.com/index.jsp I found that wattage/power consumption tool to be very useful, especially when you're worried that you're hitting the Power Supply's limit. You'll see very quickly that different CPU's, Graphics, Cards, HDDS use a wide range of energy. That tool can help you choose a build that won't stress or go over the maximum limits of your Power Supply.........

Do you have an LGA1156 board and wonder why your onboard video isn't working? It's probably because there is no GPU integrated on the board itself, it is supposed to be on the CPU and new CPU's from Intel don't come with it! I was shocked to learn that you can buy a motherboard with onboard video which doesn't work. Inever knew it was being done or possible, but to make it worse the latest Core i5's don't have video on the die of the CPU. This means having on........

I've had an old Blackberry for a few years and I was ready to spend some serious cash on a nice Nokia. I love the idea of the Linux based Symbian S60 OS, so I was looking at phones like the E63, E71 (both are basically the same, at least CPU/screensize although E71 is thinner, has a 3MP camera, has GPS, and has dedicated buttons to adjust the volume). Then Iheard about the N97 and what a revolution it was supposed to be and it was even compared to iPhone. To........