In the following lines, we list the frequently asked questions for which our users mostly try to find the answers. If no one matches your needs, please send an e-mail to cbia-supp@fi.muni.cz.
I need access to CBIA servers.
Ask the technical support cbia-supp@fi.muni.cz to create an account for selected servers.
What services are offered by CBIA servers?
The operating system installed on our servers is Linux Ubuntu (18.04 or 20.04). The most important and commonly required packages currently installed on our servers include:
Each user is allowed to install the environment that supports machine learning (e.g. Anaconda) into her/his user space.
The user can easily connect to these servers with:
The default graphical user interface (GUI) uses Xfce. We purposely use this desktop environment as it is a lightweight one, easily configurable, and it does not occupy CPU too much.
What are the available servers and what is their configuration?
We possess and manage 5 servers located in datacenter all connected to the power backup circuit.
More technical details can be found here.
I want to use GPU card. How can I list the existing GPU cards?
$> nvidia-smi
The command above also shows which card is currently occupied. Please, be aware that the occupiation does not correspond to the reservation! The card may look to be free as the person who reserved it is just preparing her/his computation.
Do I need to make a reservation before using GPU cards?
Yes, you have to. You can do it here. In the field "comment", please put the keyword that describes your task. Here are the possible keywords:
How can I use the GPU cards?
Before using GPU each user is required to make a reservation in the Information system. Please, be aware that the reservation is allowed only to those who have already an active account on our servers.
After making a reservation, the user runs her/his computation on the selected GPU card by using the system variable called CUDA_VISIBLE DEVICES.
For computing on the first GPU card, one should type:
$> export CUDA_VISIBLE_DEVICES=0; ./<your_application>
For computing on the second and the third GPU card, one should type:
$> export CUDA_VISIBLE_DEVICES=1,2; ./<your_application>
Please, do not forget to set the variable CUDA_VISIBLE_DEVICES, otherwise you may block the computation of other users!
How can I find out the occupacy of selected GPU card?
Download and install the script called nvitop. You needn't be logged as an administrator because the script is installed into user profile. Finally, run the following command:
$> nvitop --monitor
How can I find out the occupacy of CPUs?
Run the following command:
$> htop
If you find out that your processes tend to occupy almost all the CPUs, use system commands "taskset" and "renice" to limit your processes. See details below.
My process turned out to be very CPU intensive, but I don't want to kill it.
The system command "renice" changes the priority of your process. You cannot increase the priority but you can decrease the priority by assigning it numbers in the range 1-20. The higher number means lower priority. The following command decreases the priority od process with PID 1234 from the default (0) to the new level 10:
$> renice 10 -p 1234
The system commonad "taskset" can be used to assign your process to specific CPU cores. You can do it when starting the process but you can do it also during the life of your process without kill it. The following command will assign the process with PID 1234 together all its threads and subprocesses to CPU cores 10-20:
$> taskset -acp 10-20 1234
What is the expected size of my home? Is there some limit?
We currently do not apply any limit to user homes but the expected longterm size of data stored in your home is about 100GB (shortly you can use up to few TBs). Please, keep in mind that our GPU servers are dedicated to computation only. These are not fileservers. As soon as you finish your computation, please move your data to another location/server.
I need to install some package into the system but I do not have enough rights.
Send a request to the technical support cbia-supp@fi.muni.cz.
If you intend to use one of our automated microscopy systems, we can provide you with our assistant to help you configure the microscopy hardware and software appropriately. We help you optimize the image quality (from the optical as well as digital point of view) so that the image data is suitable for subsequent computer analysis and you can get the most from your data. We do not impose any limits on the duration of your experiment or on the final image data size. We can also help you design the whole experimental pipeline from sample to final evaluation of image analysis results. In order to inquire about suitable time for your experiment or to ask about further details, please contact our technical support cbia-supp@fi.muni.cz.
If you are a beginner when using one of the installed software packages, we can advice you where to find the required modules, functions or parameter settings. Again, please contact our technical support cbia-supp@fi.muni.cz.