Install a comprehensive Python and GPU powered deep learning suite. The Nvidia RTX tensor cores can accelerate deep learning tasks by up to 25x (2,500%) versus other popular developer laptops.
The Focus ships with CUDA, TensorRT, cuDNN, OpenGL, and Vulkan libraries pre-installed. This enables sophisticated compute and gaming out of the box. However, adding repositories or packages can alter and break the libraries. You may check and reset them as shown below:
Use kfocus-001-conda to install an
Anaconda suite. This is professionally vetted, comprehensive, and GPU accelerated. Open a terminal and enter the following:
Follow the prompts. You will be asked to install the following environments:
The script 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.
When running Deep Learning workloads, you may notice that the software reports less available RAM than the GPU actually has. Part of this is because of driver overhead, but also because the GPU uses some memory for your display. One can instead use the Intel GPU for the display and use the Nvidia GPU for only Deep Learning or other computing tasks. To do this, start the
Nvidia Settings app and then select
Prime Profiles > Nvidia On-Demand and reboot. This mode does not significantly improve battery life, so we recommend using
Nvidia (Performance mode) or
Intel (Power Saving Mode) as it suits your needs at other times.
You may run comprehensive TensorFlow benchmarks at the end of the
kfocus-001-conda script. Please compare to the official results. All tests are current as of November 2020.
|RTX 2080 Super||RTX 2070||RTX 2060||Titan X||Titan XP W10||Titan XP Ubuntu|
 Kubuntu Focus 2.2Kg, 20mm thick
 Unconstrained Desktop GPU
This is a partial revision history. See the
git repository for all entries.
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 70b4aa40Initial document
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.