Install and run Anaconda, TensorFlow, PyTorch, and an AI benchmark with a single command. Enjoy NVIDIA RTX, CUDA, and CuDNN-powered performance 5 to 30 times greater than popular Unix® laptops. Develop with the industry-standard PyCharm IDE using the expertly packaged JetBrains Toolbox and desktop integration. This solution only applies to the Kubuntu Focus M1 and M2 models, as they have the required Nvidia hardware.
Run the app from
App Menu > Kubuntu Focus > Anaconda GPU AI Suite Installer, or open a terminal and enter
kfocus-conda (on 22.04 LTS) or
kfocus-001-conda (on 20.04 LTS). This installs an
Anaconda suite that is professionally vetted, comprehensive, and GPU accelerated. You will be prompted to optionally install the following environments:
Kfocus Conda will not overwrite or delete data unless you tell it to do so. You may run it again to add environments as needed. For example, you may install only
TensorFlow at first and then add
PyTorch later. Run the AI suite at the end to compare to the results shown in the Benchmarks section.
When you wish to build on these Conda environments, we recommend using Pycharm. You may install this with the JetBrains Toolbox installed on all Kubuntu Focus laptops.
You can change the GPU used for your displays using the GPU Widget found in the system tray. When you use
Nvidia GPU (Performance Mode), every display pixel consumes additional VRAM. Total display VRAM may exceed 2 GB. Run
watch nvidia-smi in a terminal to see VRAM use.
Sometimes you may want to use all VRAM for a pipeline so it can complete or run faster. The best solution is to click on
Switch to Intel GPU (Power Save Mode). This will free all VRAM by using the iGPU and system RAM for the display. Only the primary laptop display may be available, although some models support a second display with DisplayPort.
If you need to drive more displays, you can
Switch to Hybrid GPU (On-Demand Mode) as shown here. But So please remember to switch back when you are done!
You may run comprehensive TensorFlow benchmarks at the end of the Kfocus Conda script. Please compare to the official results. All tests are current as of June 2023.
|Results[1,2,3]||RTX 4090 M||RTX 4080 M||RTX 4070 M||RTX 3080 Ti M||RTX 4060 M||RTX 3070 Ti M||RTX 3060 M|
|Video RAM||16 GB||12 GB||8 GB||16 GB||8 GB||8 GB||6 GB|
 See official comparison chart
 All 40-series from Kubuntu Focus M2GEN 5
 All 30-series from Kubuntu Focus M2GEN 4
Q: Starting NVIDIA Visual Profiler results in an error. How can I fix this?
A: One must install Java 8 and then call
nvvp with the correct path.
Q: I booted my Kubuntu Focus, and it never got past the splash screen. How do I fix this?
NVIDIA Settings and other applications may ask you to save an
xorg.conf file. Doing so may disable your internal display from functioning with a GUI upon logout or restart, and you will not be able to sign-in. If this happens, we recommend you remove this file:
sudo /bin/rm /etc/X11/xorg.conf, type your password when prompted.
sudo systemctl restart sddm && exit.
You should be able to sign-in normally.
Q: I see some libray errors when installing Anaconda. What can I do?
A: The Focus M2 ships with OpenGL and Vulkan libraries installed. When you run the Kfocus Conda script, you may be prompted to download and install additional libraries such as CUDA and CuDNN. Please ensure you are connected to the Internet with good bandwidth, as these can be quite large.If you still receive these errors, you can cancel, run the FocusRx diagnostics shown below, and then try again.
This is a partial revision history. See the
git repository for all entries.
2023-11-18 e770b1c6Add search and help bar
2023-10-18 fac730e3Enhance VRAM usage advice
2023-06-19 a7af817bUpdate RTX 40-series benches, add image, revise VRAM text
2023-04-16 1bc15058Update display advice concerning iGPU Intel
2022-06-11 b304575dChange spelling from MaxQ to Max-Q
2023-05-18 0e693861Refine opening language
2023-05-18 1a4bd7dcAdded M2 GEN 4 details in benchmark
2023-05-18 1a4bd7dcMove focusrx to troubleshooting
2022-01-16 3ae944e8Fix HTML Entities; replace dist-upgrade with full-upgrade
2021-11-07 3e32a343Update table to current
2021-10-10 5728326eReformat to 2-column
2021-09-22 dc862884Update link and headline colors
2021-08-23 681261b4Review and update codeblocks
2021-08-19 b994dd79Add keywords for hybrid graphics mode
2021-03-07 e3c1285dRevise conda launch guide, update styles
2020-09-28 5d089610Update to M2 scores
2020-06-10 c4ed9299Restructure layout
2020-05-25 581c6523Expand benchmarks
2020-05-20 d00295d7Add Library check
2020-05-18 8c4b5b82Add GPU library fix
2020-05-15 0684cd5cAdd benchmark
2020-02-18 a4d437c5Add conda script
2020-01-31 70b4aa40First publication
We try hard to provide a useful solution validated by professionals. However, we cannot anticipate every situation, and therefore cannot guarantee this procedure will work for your needs. Always backup your data and test the solution to determine the correct procedure for you.
THIS SOLUTION IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOLUTION, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.