---

tags: user support coffee break
title: archive5

---

# Archive of Weekly CSC user coffee break Q&As (vol5)

## Older archives
Vol4: https://hackmd.io/N91KGM4KQEekkIQCQ10HmQ?view
Vol3: https://hackmd.io/SggRjgEYToWq4zA7D6ecbw?view
Vol2: https://hackmd.io/uVI5gLKDQoWZgAVcufYg2A?view
Vol1: https://hackmd.io/9QsIbJ03T1SaNDV_xbT8Dw?view

# Q&A for Weekly CSC user coffee break

*CSC's open support session organised every **Wednesday at 14:00** in [Zoom](https://cscfi.zoom.us/j/65059161807).*

:::info
**Zoom:** <https://cscfi.zoom.us/j/65059161807>
**Q&A:** <https://siili.rahtiapp.fi/weekly-user-zoom>
**Events page (slides and recordings):** <https://csc.fi/koulutuskalenteri/csc-research-support-coffee-every-wednesday-at-1400-finnish-time-2/>
**Feedback:** <https://link.webropolsurveys.com/S/94AB9F77D8EFF054>
**Q&A archive:** <https://hackmd.io/@CSC-research-support/Hy5JgiNQ1e>
**Short talk archive:** <https://hackmd.io/1xCtrll9SN2wqGYjgZEqhw?view>
:::

:::warning
:bulb: **Useful links & further resources**
* [CSC docs](https://docs.csc.fi/)
* [Getting started with supercomputing at CSC](https://docs.csc.fi/support/tutorials/hpc-quick/)
* [CSC Computing Environment course materials](https://csc-training.github.io/csc-env-eff/)
- [CSC's courses](https://www.csc.fi/training)
:::

## 2026-01-28 session
**Short talk:** Processes, threads and cores: how to ensure efficient usage of CSC supercomputers
([Docs documentation](https://docs.csc.fi/computing/running/creating-job-scripts-mahti/#undersubscribing-nodes), [slides](https://kannu.csc.fi/s/smFdbmaDCnBRwjW) from previous time, [recording](https://video.csc.fi/media/t/0_uoq5c9bi) of previous time)

**Upcoming short talks** (initial plan)
- 4.2. Roihu schedule update
- 11.2. Understanding parallel scalability
- 25.2. SSH CA in Roihu

#### Course & Webinar Advertisements

- Course: [Hands-on Data Protection workshop](https://csc.fi/en/training-calendar/hands-on-data-protection-workshop/) 27.-29.1.2026
- Course: [Single-cell RNA-seq data analysis using Chipster](https://csc.fi/en/training-calendar/single-cell-rna-seq-data-analysis-using-chipster-3/) 3.-4.3.2026
- Course: [CodeRefinery workshop](https://csc.fi/en/training-calendar/coderefinery-workshop-4/) 17.-19.3. & 24.-26.3.2026
- School: [CSC Spring School on Computational Chemistry](https://csc.fi/koulutuskalenteri/csc-spring-school-on-computational-chemistry-2026/) 8.-10.4.2026
- School: [BioExcel Summer School](https://csc.fi/koulutuskalenteri/bioexcel-summer-school-2026/) 7.-12.6.2026
- School: [CSC Summer School in High-Performance Computing 2026](https://csc.fi/en/training-calendar/csc-summer-school-in-high-performance-computing-2026/) 23.6.-2.7.2026

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - Robot account in Mahti
Room 4 - MetaHipMer2
Room 5 - BDRhapsody pipeline
Room 6 -
Room 7 -
Room 8 -
Room 9 -
Room 10 -

---

### Questions

- **Q523** Puhti and Mahti are to be disassembled in few months time frame. I've expressed interest to host few surplus CPU/GPU nodes for our research group (Future Computing Group in UOulu), but I cannot find any mechanical and electric specifications to the nodes from CSC pages or from Bull -- i.e. what are the power requirements and are they using schuko, noise levels etc. as I've got only limited room, cooling and power available. /Jani Launonen

- Please email directly to Simon Westersund: [e-mail]
- https://docs.csc.fi/support/faq/roihu/#3-i-heard-that-puhti-will-be-donated-for-continued-use-can-i-apply-for-hardware
- C13 power plugs, dual power supplies per chassis. Depending on the node type, the power needs are different. Noise levels are unknown, but expect them to be high. Normal air-cooled rack-mounted server levels.

- **Q524** Robot account in Mahti - practicalities. Gleb Tikhonov

- breakout room number 3

- **Q525** Installation of the MetaHipMer2 tool for co-assembly of metagenomic reads on Mahti - best practices? + Matti Ruuskanen

- breakout room number 4

- **Q526** Running BDRhapsody pipeline for single cell RNAseq samples in Puhti + Heather

- Break out room 5
- Customer is running a (container?) pipeline in interactive node, but hits the time limit
- Difficult to estimate the memory and time needed
- 1-2 TB data, 4 samples
- Can test with 1 sample, to get some estimates of the resources needed
- Recommended to use a batch job, since the pipeline is doing "all at once", and no interactive use is needed
- Our free CSC Computing environment course: https://csc.fi/en/training-calendar/csc-computing-environment-self-learning/
- and related material: https://csc-training.github.io/csc-env-eff/
- More information regarding batch jobs: https://csc-training.github.io/csc-env-eff/part-1/batch-jobs/
- More info on estimating the resources: https://csc-training.github.io/csc-env-eff/part-1/batch-resources/
- "Quick start guide": https://docs.csc.fi/computing/running/getting-started/

## 2026-01-21 session
No short talk this time.

#### Course & Webinar Advertisements

- Course: [Hands-on Data Protection workshop](https://csc.fi/en/training-calendar/hands-on-data-protection-workshop/) 27.-29.1.2026
- Course: [Single-cell RNA-seq data analysis using Chipster](https://csc.fi/en/training-calendar/single-cell-rna-seq-data-analysis-using-chipster-3/) 3.-4.3.2026
- School: [CSC Spring School on Computational Chemistry](https://csc.fi/koulutuskalenteri/csc-spring-school-on-computational-chemistry-2026/) 8.-10.4.2026
- School: [BioExcel Summer School](https://csc.fi/koulutuskalenteri/bioexcel-summer-school-2026/) 7.-12.6.2026

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started (Dean)
Room 2 - KimmoM + Mikko
Room 3 - Long term storage (Johan)
Room 4 - Getting started 2 (Mats)

---

### Questions

- **Q519** Current status of Lumi Francesco

- LUMI is offline atm, email / info to follow

- **Q520** Long term archival grade preservation of data, any recommendations by CSC:) + Shaheryar

- Answer / breakout room number
- https://digitalpreservation.fi/

- **Q521** Is it allowed to submit slurm jobs from within a slurm script? - Devon

- Yes, but there can be issues.

- **Q522** Roihu schedule / Alexander

- Most updated info: https://docs.csc.fi/computing/systems-roihu/

## 2026-01-14 session

#### Course & Webinar Advertisements

- Course: [Hands-on Data Protection workshop](https://csc.fi/en/training-calendar/hands-on-data-protection-workshop/) 27.-29.1.2026
- Hackathon: [Code Kickstart Hackathon](https://csc.fi/en/training-calendar/code-kickstart-hackathon/) 25.-27.2.2026. **Register by 16.1.2026!**
- Course: [Single-cell RNA-seq data analysis using Chipster](https://csc.fi/en/training-calendar/single-cell-rna-seq-data-analysis-using-chipster-3/) 3.-4.3.2026
- School: [CSC Spring School on Computational Chemistry](https://csc.fi/koulutuskalenteri/csc-spring-school-on-computational-chemistry-2026/) 8.-10.4.2026
- School: [BioExcel Summer School](https://csc.fi/koulutuskalenteri/bioexcel-summer-school-2026/) 7.-12.6.2026

### Questions

- **Q515** Hi, I tried to install Earl Grey using tykky (it's installed via conda), but I have to configure Dfam and Earl Grey directs me to a location I cannot access:
"Before using Earl Grey, you MUST download the required partitions from Dfam (https://dfam.org/releases/current/families/FamDB/). These must be located at /PUHTI_TYKKY_DJjLbp5/miniforge/envs/env1/share/RepeatMasker/Libraries/famdb/." + Jana

- Since the Dfam databases are somewhat large it is best not to include them in the container image, but rather bind them at runtime. The databases also need to be indexed an this is easiest to do when building the container image.
1. Download and uncompress all the necessary dfam partitions (including partition 0) to a directory in /scratch, e.g. "/scratch/project_12345/famdb" Use correct directory instead of this example. Adjust following commands accordingly.
2. Prepare a definition file "earlgrey.def" with following content (Note: perl command all on one line):

      Bootstrap: docker
      From: docker.io/tobybaril/earlgrey:latest-nodfam

      %post
      # Rerun RepeatMasker configuration
      cd /usr/local/share/RepeatMasker/
      perl ./configure -libdir /usr/local/share/RepeatMasker/   Libraries/ -trf_prgm /usr/local/bin/trf -rmblast_dir /usr/local/bin -hmmer_dir /usr/local/bin -abblast_dir /usr/local/bin -crossmatch_dir /usr/local/bin -default_search_engine rmblast

3. Build a container with the new indexes:

            apptainer build --bind /scratch/project_12345/famdb:/usr/local/share/RepeatMasker/Libraries/famdb earlgrey.sif earlgrey.def

4. Do the same --bind when running:

          apptainer exec --bind /scratch/project_12345/famdb:/usr/local/share/RepeatMasker/Libraries/famdb earlgrey.sif earlGrey <options>

- **Q516** Schrödinger(Maestro) MATERIALSCIENCE_BILAYER token missing in GUI step
- I would like to set up simulations locally then wirte the GUI-generated input files to run in Puhti, but got the error “No MATERIALSCIENCE_BILAYER license token is available” in the Membrane Analysis / bilayer preparation step on my local Maestro GUI, so I cannot generate the required .cms/.maegz input files. Does this mean that bilayer/membrane tools are not included in CSC’s Schrödinger license? + **Qianqian**

- That's correct, we do not have the MATERIALSCIENCE_BILAYER license available. Sorry about this :(
- These are the MS modules we have licenses for:

       Users of MATERIALSCIENCE_CG: (Total of 1000 licenses issued; Total of 43 licenses in use; Total of 957 licenses available)
       Users of MATERIALSCIENCE_DIFFUSION: (Total of 1000 licenses issued; Total of 43 licenses in use; Total of 957 licenses available)
       Users of MATERIALSCIENCE_GA: (Total of 1000 licenses issued; Total of 0 licenses in use; Total of 1000 licenses available)
       Users of MATERIALSCIENCE_MAIN: (Total of 100 licenses issued; Total of 0 licenses in use; Total of 100 licenses available)
       Users of MATERIALSCIENCE_PENETRANT: (Total of 1000 licenses issued; Total of 43 licenses in use; Total of 957 licenses available)
       Users of MATERIALSCIENCE_QE: (Total of 1000 licenses issued; Total of 43 licenses in use; Total of 957 licenses available)

- You can check the full list yourself using `$SCHRODINGER/run lictool status`

- **Q517** installing pip install flash-attn on Mahti (14.01.2026)

- container:

    module purge
    module load tykky
    cd /projappl/project_XXXXXXX
    vim packages.sh

* Extract cluster name from hostname

    CLUSTER_NAME=$(echo $HOSTNAME | grep -oE '(mahti|puhti|narvi)' | head -1)

    echo "Loading latest CUDA module for cluster: $CLUSTER_NAME"

    if [ "$CLUSTER_NAME" = "mahti" ]; then
	  module load gcc/10.4
    else
	  module load gcc/13.2
    fi
    module load cuda/12.6

    echo "NVCC: $(nvcc --version)"
    echo "NVCC: $(which nvcc)"
    echo "Installing torch..."
    pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126
    echo "--- Compiling Flash Attention ---"
    pip install flash-attn --no-build-isolation --no-cache-dir

- Can you use the pre-installed PyTorch which also has flash-attention included? https://docs.csc.fi/apps/pytorch/

- **Q518** Are their any resources to automate Deep Learning training tasks for a Anova analysis using different random starting seeds ? Name: BV Pradeep Uwasa.

- Array jobs (https://docs.csc.fi/computing/running/array-jobs/)

## 2025-12-10 session
~~Short talk: "Understanding parallel scalability" (Jussi E)~~ -> cancelled/moved to early 2026

#### Course & Webinar Advertisements

- Course: [Machine learning with spatial data](https://csc.fi/en/training-calendar/practical-machine-learning-with-spatial-data-2/) 10.-12.12.2025
- School: [CSC Spring School on Computational Chemistry](https://csc.fi/koulutuskalenteri/csc-spring-school-on-computational-chemistry-2026/) 8.-10.4.2026
- School: [BioExcel Summer School](https://csc.fi/koulutuskalenteri/bioexcel-summer-school-2026/) 7.-12.6.2026

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Beginners /Ari-Matti
Room 2 - Ambassadors
Room 3 - Chipster
Room 4 - Oskar / pytorch
Room 5 - R / Heli
Room 6 - SD / Kimmo
Room 7 - Allas/ JuhaL

---

### Questions

- **Q510** I try to do multiqc analysis and RNA-seq quality metrics with RSeQC in Chipster, but they don´t work and error message says expired waiting + Jasmin Robert Jackson

- Related to Chipster
- breakout room number 3
- Chipster / Rahti / Pouta was experiencing some unfortunate hiccups, but the issue was now solved, and it was ok to continue the analysis.
- As the data is big, some tools too take a long while to run. Several jobs can be queued at once.
- RNAseq analysis workflow was checked

- **Q511** Use of shared libraries in python, pytorch apptainer, Volodymyr Beimuk

- Answer / breakout room number 4
- To run something inside the container with a virtual environment activated:

    export SING_FLAGS="-B /usr/lib/example.so"
    apptainer_wrapper exec bash -c "source $VENV/bin/activate && python $PYSCRIPT"

- alternatively you can use the python inside the venv directly, if you don't need the other things the activate script does.
- explicitely only bind the files that are needed from `/usr/lib` instead of the entire directory
- Update: the links in the venv point to the wrapper intentionally. It should work, when you are using `singularity_wrapper` instead of calling `apptainer` directly. https://docs.csc.fi/computing/containers/overview/#using-apptainer-wrapper
- This in principle also works:

    srun singularity exec -B "$TOBIND" "$SIF" bash -c "source $VENV/bin/activate && python $PYSCRIPT"

- But then you have to make sure the links (e.g. python) inside the venv point at the python inside the container e.g. `/usr/bin/python` instead of the wrapper.

- **Q512** How to set up parallelisation with brms in R, Moritz Oberndorfer

- Answer / breakout room number 5

- **Q513** New to csc and having issues with making R and python combined environment Buddika Jayasingha

- Answer / breakout room number 1

- **Q514** 2. In the interactive Allas interface, any user seems able to delete buckets. Is there a way to set restrictions so that certain users can only view/read buckets, without having permissions to edit or delete them? Buddika Jayasingha

- All project members have full permissions to the buckets of the project. The workaround is to create a separate project to which only the data maintainers/curators belong, and then make the data in that project public, or available to other projects, https://docs.csc.fi/data/Allas/using_allas/common_use_cases/ , https://docs.csc.fi/data/Allas/allas_project_example/ . / breakout room 7

## 2025-12-03 session

#### Course & Webinar Advertisements

- Course: [Machine learning with spatial data](https://csc.fi/en/training-calendar/practical-machine-learning-with-spatial-data-2/) 10.-12.12.2025
- School: [CSC Spring School on Computational Chemistry](https://csc.fi/koulutuskalenteri/csc-spring-school-on-computational-chemistry-2026/) 8.-10.4.2026
- School: [BioExcel Summer School](https://csc.fi/koulutuskalenteri/bioexcel-summer-school-2026/) 7.-12.6.2026

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Beginners + Q507 (Oskar, Iida)
Room 2 - Allas Q504 (Mats)
Room 3 - RStudio Q508, Q509 (Heli)
Room 4 - Q505 (Laxman)

---

### Questions

- **Q504** Connecting Puhti to Allas: -Aryan

- Answer / Mats in breakout room number 2

- **Q505** Help with scaling up resource intensive freebayes job using snakemake and HyperQueue - Natalie van Dis
NB: I have a working snakemake / HyperQueue script but it's so resource intensive that I don't know on which partition to run it and what resources to specify

- Answer / breakout room number 4 (Laxmana)

- **Q506** Problem with remote connection to puhti/mahti when using local VSCode. I believe the problem is due to lack of memory when trying to download newest vscode-server and extensions - Felipe

- https://docs.csc.fi/support/faq/vscode-out-of-memory/

- **Q507** Help with using batch job for Puhti - Yuan

- Breakout room number 1

- **Q508** Running a SLURM RStudio parallel job in puhti + Aino-Maija

- Answer / Heli in breakout room number 3

- **Q509** R INLA model failure, advise on resource allocation - Natalie van Dis
- Room 3

## 2025-11-19 session

**Short talk:** Processes, threads and cores: how to ensure efficient mapping in CSC supercomputers (JussiE)
- [Docs documentation](https://docs.csc.fi/computing/running/creating-job-scripts-mahti/#undersubscribing-nodes)
- [Slides](https://kannu.csc.fi/s/smFdbmaDCnBRwjW)

#### Course & Webinar Advertisements

- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- Course: [Python for Scientific Computing](https://csc.fi/en/training-calendar/python-for-scientific-computing/) 25.-27.11.2025
- Course: [Schrödinger Suite Workshop](https://csc.fi/en/training-calendar/schrodinger-suite-workshop-2025/) 26.-27.11.2025
- Course: [Machine learning with spatial data](https://csc.fi/en/training-calendar/practical-machine-learning-with-spatial-data-2/) 10.-12.12.2025

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - SD Services / Q498 (Meeri)
Room 4 - AI + machine learning / Q499 (Mats)
Room 5 - Short talk
Room 6 - Getting started B
Room 7 - Q497 /Alvaro & Tiia
Room 8 - MATLAB / Jaan
Room 9 -
Room 10 -

---

### Questions

- **Q497** Hi, I am not able to cut and paste text into the terminal in Puhti. This would be useful when dealing with complicated directory and file names. Is there some way to paste text into the terminal? I have a windows system and I am using Puhti through the web interface. Tia Korhonen

- Answer / breakout room number
- Which operating system you have (on your own computer)?
- Are using an ssh client or the Puhti web user interface (https://www.puhti.csc.fi/)?
- I have a windows system
- and I am using Puhti through the web interface
- Room 7 -Alvaro

- **Q498** I would like to test the SD Desktop, but I cannot see it as an available service for me. I wonder if it is because of that I have the status of visiting researcher in our university? I have created a test project (Academic). -Nina Hänninen

- This issue was due to the HAKA attributes we get from UEF authentication, discussed in Breakout room number 3
- Let's discuss, the test project is (probably) not secondary use, but the actual project would be secondary use...
- Yes I also noticed somewhere the mention of student project, but the funny thing is, that I am post-doc researcher, but have also a study right to MSc level on another field... And actually the same problem is with my supervisor, who is as a visiting researcher status, he couldn't access SD Desktop either.

- **Q499** I have my python code running Qwen3 VLM model (`Qwen/Qwen3-VL-8B-Instruct`) which runs on both Puhti and Mahti (same python script with two sbatch scripts for gpusmall partition in Mahti and gpu in Puhti). In Mahti, it takes something around ~16 hours whereas in Puhti, it never accomplishes the tasks and always exceeds the 3day limit and it fails. I noticed the following:
- Puhti:

Python 3.10.18 | packaged by conda-forge | (main, Jun  4 2025, 14:45:41) [GCC 13.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> cuda = torch.cuda.current_device()
>>> torch.cuda.get_device_capability(cuda)
(7, 0)

- Mahti:

Python 3.10.18 | packaged by conda-forge | (main, Jun  4 2025, 14:45:41) [GCC 13.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> cuda = torch.cuda.current_device()
>>> torch.cuda.get_device_capability(cuda)
(8, 0)

Could it be such that it will never be possible to run new VLMs in Puhti given the gpu compatibility issues of V100?
The situtation is pretty much the same using the quantized models.

- Let's discuss in breakout room 4 after the short talk.

- **Q500** Is a shell pipe one or several tasks, what should the --ntasks parameter be? + Heli

- Room 5

- **Q501** If my seff output gives me over 100% memory efficiency, is this a sign I should spread out the CPUs? E.g. Nodes: 1, Cores per node: 2, Memory utilized: 3.72 GB, Memory Efficiency: 185.80% of 2GB (of 1.00 GB/core) + Joonas L

- Room 5

## 2025-11-12 session

**Short talk:** No short talk

#### Course & Webinar Advertisements

- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025. Still possible to register today 12.11.!
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- Course: [Python for Scientific Computing](https://csc.fi/en/training-calendar/python-for-scientific-computing/) 25.-27.11.2025
- Course: [Schrödinger Suite Workshop](https://csc.fi/en/training-calendar/schrodinger-suite-workshop-2025/) 26.-27.11.2025

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - CSC Computing Environment course
Room 4 - SD (Kimmo)
Room 5 - Uploading images to HPC environment (Heli, Iida)

---

### Questions

- **Q495** After I added the ssh key to puhti profile, I still can not connect the supercomputer. -Siyu (Discussed in main room)

- The case in short:
- Access to Puhti is ok via webinterface
- SSH ccess via terminal not working, possibly because the passphrase was set empty?
- There's also a local cluster, that needs to be connected to Puhti
- There are also SSH keys for the connection to the local cluster from users own computer, which might cause confusion here.
- Regarding connecting to Puhti & SSH keys: https://docs.csc.fi/computing/#connecting-to-the-supercomputers
- Important notes regarding the SSH keys:
- The private key should never be shared with anyone, not even with CSC staff! It should only be stored on the local workstation.
- Also, never leave the passphrase empty when generating an SSH key pair! Please choose a secure passphrase. It should be at least 8 characters long and contain numbers, letters and special characters.
- In case no passhphrase was set, you can try generating new set of SSH keys
- **Option 1:** Use the web interface to upload and download your files (works if they are <10GB)
- **Option 2:** Use Allas: https://docs.csc.fi/data/Allas/ You can also transfer data to Allas! Might be a nice solution here.
- **Option 3:** Connecting the local cluster to Puhti, the SSH agent: https://docs.csc.fi/computing/connecting/ssh-unix/#ssh-agent-forwarding + https://docs.csc.fi/computing/connecting/ssh-unix/#authentication-agent

- **Q496** Noppe / other CSC services for training + Rishi

- Discussed in the ambassador room
- https://research.csc.fi/use-case/environment-for-web-based-big-data-courses/
- Mostly, depends on whether you need GPUs or don't (Noppe is fine if no GPUs are not needed)

## 2025-11-05 session

**Short talk:** Roihu supercomputer - current status and updates

#### Course & Webinar Advertisements

- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025, last day to register: 9.11.
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- Course: [Python for Scientific Computing](https://csc.fi/en/training-calendar/python-for-scientific-computing/) 25.-27.11.2025
- Course: [Schrödinger Suite Workshop](https://csc.fi/en/training-calendar/schrodinger-suite-workshop-2025/) 26.-27.11.2025
- Course: [Data Analysis with R](https://csc.fi/en/training-calendar/data-analysis-with-r-5/) 1.-2.12.2025

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started (Heli)
Room 2 - Ambassadors
Room 3 - CSC Computing Environment Part2 (Maria)
Room 4 - SD related (Kimmo)
Room 5 - Roihu (Rasmus)
Room 6 - Q493 (Mats)
Room 7 - Q494 (Oskar)
Room 8 - Data management application (Robin)

---

### Questions

- **Q493** Type question/topic here :
Hi,Beginner here. I'm getting started with estimating bayesian multilevel models on Puhti. I've been experimenting with smaller tasks (running the model on 10k to 50k observations in interactive sessions and batch jobs before trying the heavier jobs. However, it seems like I'm always at 0% CPU efficiency while my smaller jobs already need 10GB memory to work. I'd like to get help on how to increase the CPU efficieny before I run the models on the bigger datasets (multiple million observations). Thank you in advance! Moritz

- Room number 6 (Mats)
- **Q494** Installing software you cant find as a module. I'm trying to analyze my data on Puhti, with a code that uses CERN's ROOT software which is not included as a module on the system. I am having trouble undestanding how to download the software and could use help or is it possible to add it as a module. -Tero Lappeteläinen

- Room number 7 (Oskar)
- Try first to install via conda and tykky. If that doesn't work installing on Puhti directly with precompiled binaries should work. Inside a container one can also install with the package manager.

## 2025-10-29 session

**Short talk:** Using MATLAB on CSC supercomputers

#### Course & Webinar Advertisements

- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- Course: [Python for Scientific Computing](https://csc.fi/en/training-calendar/python-for-scientific-computing/) 25.-27.11.2025
- Course: [Schrödinger Suite Workshop](https://csc.fi/en/training-calendar/schrodinger-suite-workshop-2025/) 26.-27.11.2025
- Course: [Data Analysis with R](https://csc.fi/en/training-calendar/data-analysis-with-r-5/) 1.-2.12.2025

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started (Oskar)
Room 2 - Ambassadors
Room 3 - Q491 / Machine learning (Mats)
Room 4 - Q492 / SD Connect (Kimmo)
Room 5 - Matlab (Jaan)
Room 6 -
Room 7 -
Room 8 -

---

### Questions

- **Q490** Python (+Monai-label) use for a first timer + Panu Erästö

- Breakout room 1

- **Q491** I am using Puhti to run some code which uses Ollama to get various LLMs to describe images. I wrote my code based on the example code provided by CSC and want some advice about how to check whether this setting ("gres=gpu:v100:4,nvme:100") makes sense for my context, or whether I should change it. Similarly, wondering whether there's anything in my "seff (job number)" output that suggests that settings should be changed. Finally, I am using lots of GPU credits (Which is fine if necessary) and I'm not knowledgeable enough to understand whether I'm being inefficient with my code, or whether this is just a "GPU-heavy" use case. - Adeline Clarke

- Breakout room 3

- **Q492** I need some help with the SD Connect command line tool sd-lock-util / Anni

- Breakout room 4

## 2025-10-22 session

**Short talk:** Climate Digital Twin and how to access the data produced
- https://destine.ecmwf.int/climate-change-adaptation-digital-twin-climate-dt/
- https://docs.lumi-supercomputer.eu/firststeps/loggingin-webui/
- https://github.com/DestinE-Climate-DT/climatedt-community-resources/blob/main/introduction/climate_dt_data_access.md

#### Course & Webinar Advertisements

- Webinar: [Enhancing Data Support: Practical Reproducibility, Part 2](https://csc.fi/en/training-calendar/enhancing-data-support-practical-reproducibility-part-2/) 29.10.2025
- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- Course: [Schrödinger Suite Workshop](https://csc.fi/en/training-calendar/schrodinger-suite-workshop-2025/) 26.-27.11.2025
- Course: [Data Analysis with R](https://csc.fi/en/training-calendar/data-analysis-with-r-5/) 1.-2.12.2025

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - Q487 & other machine learning questions (Mats)
Room 4 - Q488 & other R questions (Heli)
Room 5 - Rstudio/SDS (Kimmo)
Room 6 - Tracy and Climate Digital Twin and how to access the data produced
Room 7 -
Room 8 -

---

### Questions

- **Q487** I am hoping to use Puhti to run some python scripts which use Ollama to get various LLMs to describe a set of images. I have been following the process described under 'Inference with vLLM' here: https://docs.csc.fi/support/tutorials/ml-llm/ and the linked github, and I have the code working locally with smaller LLMs, but I'm struggling to get it working on Puhti. I'm hoping someone can help me figure out what I am missing. + Adeline

- Room 3

- **Q488** Run R scripts on Puhti with uploaded data (Moritz Oberndorfer, complete supercomputing beginner)
I want to estimate some statistical models in R using openly available birth register data which I uploaded to my Puhti folder. The models would probably take weeks to estimate on a normal machine. Thanks!

- Room 4

- **Q489** Gpaw relaxation script- Had this problem since last Q and A (Azin Alesafar)

## 2025-10-15 session
Short talk: No short talk this time

### Course & Webinar Advertisements
- Course: Portable GPU Programming 21-24.10.2025
- Webinar: Enhancing Data Support: Practical
- Reproducibility, Part 2 29.10.2025
- Course: Practical Deep Learning, 11.-12.11.2025
- Course: CSC Computing Environment Part2: Next steps 12.-13.11.2025
- Course: High Performance R 17.-18.11.2025
- Course: Using Containers in Supercomputing Environment 25.-27.11.2025
- Course: Schrödinger Suite Workshop 26.-27.11.2025
- Course: Data Analysis with R 1.-2.12.2025 (registration opens soon; updated course concept)
See CSC’s training calendar for more

### Zoom breakout rooms and topics
Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - SD Desktop / Kimmo
Room 4 - R package installation
Room 5 - Pouta/Mahti question
Room 6 -
Room 7 -
Room 8 -

### Questions

- Q483 Gleb. I am desighning a Pouta+Mahti computational backend for an analytical pypeline needed for the Digital Twin that we are (still) developing. My questions are 1) about the best/advised practices on how fast the VMs can be initiated/stopped based on my load-based request, and 2) what are the best practises to accomodate scheduled workload in Mahti with some non-scheduled computational tasks that I have in other projects.

- Answer / breakout room number Room 5
- Q484 Is it possible to access Allas data and transfer files directly to $LOCAL_SCRATCH? :) + Chito

- breakout room number Room 1
- Instructions: https://csc-training.github.io/csc-env-eff/hands-on/data-io/tutorial_allas-in-batch-jobs.html
https://docs.csc.fi/data/Allas/allas_batchjobs/
- Q485 Hei, minulla on ongelma SD-desktopiin pääsemisen kanssa, mihin break out roomiin liityn? / Hilla

- breakout room number 3
- Q486 I have some problems with using Puhti. Tislam

For using the PUHTI supercomputer for training large-scale image data.During my batch job I request a subtantial runtime 6 hours, the jobs are consistenly and automaticallt truncated after only few minutes with completing few epochs. This premature truncation is hindering my training process. could you please help me to findout the problem? Toriqul Islam

- breakout room number 6

## 2025-10-08 session

**Short talk:** Quantum computing services at CSC

#### Course & Webinar Advertisements

- Course: [Geocomputing on the supercomputer](https://csc.fi/en/training-calendar/geocomputing-on-the-supercomputer-2/) 8.-9.10.2025
- Course: [Moving your AI training jobs to LUMI: A Hands-On Workshop](https://lumi-supercomputer.eu/events/moving-your-ai-training-jobs-to-lumi-a-hands-on-workshop/), 8.-9.10.2025
- Webinar: [Enhancing Data Support: Understanding Reproducibility, Part 1](https://csc.fi/en/training-calendar/enhancing-data-support-understanding-reproducibility-part-1/) 9.10.2025
- Course: [Portable GPU Programming](https://csc.fi/en/training-calendar/portable-gpu-programming/) 21-24.10.2025
- Webinar: [Enhancing Data Support: Practical Reproducibility, Part 2](https://csc.fi/en/training-calendar/enhancing-data-support-practical-reproducibility-part-2/) 29.10.2025
- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- Course: [Schrödinger Suite Workshop](https://csc.fi/en/training-calendar/schrodinger-suite-workshop-2025/) 26.-27.11.2025

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - Quantum / Mikael
Room 4 - SD / Kimmo
Room 5 - Workflow / Rasmus
Room 6 - Q481
Room 7 - SD services / Francesca
Room 8 -

---

### Questions

- **Q480** I am running something on interactive mode sinteractive command. I had the broken pipe message and got disconnected. If i do squeue -u $USER, i see that the interactive session is still running. Is there a wayto connect to the same session using JOBID? I am unable to attend zoom session today. Answer here would be sufficient for me. Thank you. :) + Sadiksha

- If you are using mahti/puhti, you can directly ssh gpu node. squeue -u $USER will list you the NODELIST e.g., g1101, then you can open another ternimal, ssh g1101. More info: https://docs.csc.fi/computing/connecting/#:~:text=This%20also%20applies%20to%20compute%20nodes%2C%20although%20just%20the%20ones%20where%20you%20have%20a%20job%20running.%20Use%20the%20squeue%20command%20to%20see%20which%20node(s)%20your%20job%20is%20on%2C%20and%20then%20connect%20to%20a%20node%20using%20ssh.
- If you are using LUMI, you can access the running gpu node by
srun --overlap --pty --jobid=<jobid> -w nid00XXXX $SHELL.
More information can be found https://docs.lumi-supercomputer.eu/runjobs/scheduled-jobs/interactive/.

- **Q481** Could CSC consider to offer the following software as pre-installed on Puhti: FragPipe and DIA-NN? These are data analysis tools for proteomics data. I ask this on behalf of two research groups (Uni Turku & Uni Helsinki, ~40 researchers in total). DIA-NN: https://github.com/vdemichev/diann. FragPipe: https://github.com/Nesvilab/FragPipe. + Jalmari Kettunen

- Nino, breakout room number 8

- **Q482** Relaxation with GPAW- slow process plus getting "Out of Memory error". would like to make sure I am using resources efficiently. In Mahti. + Azin Alesafar

- Lustre is under heavy load. recovery_status:status: RECOVERING , hopefully that would complete soon

## 2025-10-01 session

**Short talk:** Transferring arbitrarily large files with Funet Filesender

#### Course & Webinar Advertisements

- Course: [Geocomputing on the supercomputer](https://csc.fi/en/training-calendar/geocomputing-on-the-supercomputer-2/) 8.-9.10.2025
- Course: [Moving your AI training jobs to LUMI: A Hands-On Workshop](https://lumi-supercomputer.eu/events/moving-your-ai-training-jobs-to-lumi-a-hands-on-workshop/), 8.-9.10.2025
- Webinar: [Enhancing Data Support: Understanding Reproducibility, Part 1](https://csc.fi/en/training-calendar/enhancing-data-support-understanding-reproducibility-part-1/) 9.10.2025
- Course: [Portable GPU Programming](https://csc.fi/en/training-calendar/portable-gpu-programming/) 21-24.10.2025
- Webinar: [Enhancing Data Support: Practical Reproducibility, Part 2](https://csc.fi/en/training-calendar/enhancing-data-support-practical-reproducibility-part-2/) 29.10.2025
- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- Course: [Schrödinger Suite Workshop](https://csc.fi/en/training-calendar/schrodinger-suite-workshop-2025/) 26.-27.11.2025

See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - CSC Computing Environment course support session
Room 4 - Matlab (Jaan)
Room 5 -
Room 6 -
Room 7 -
Room 8 -

---

### Questions

:::danger
If you wish you can type your question here before the session. We'll respond to them during the session. **Please join the Zoom call if you have a question, if you cannot join the call you can send the question by email to <[e-mail]>.**
Previous questions and answers can be found in the archives, see links section above.
:::

- **Q479** Puhti basic batch job configuration not working

I have the following config:

#!/bin/bash
#SBATCH --account=project_2015383
#SBATCH --partition=gpu
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=1G
#SBATCH --time=00:00:10
#SBATCH --gres=gpu:v100:1

module purge
module load tensorflow/2.18

srun python3 Skin-Cancer-Classification-Final.py

And it throws the following error:
$ srun run.sh
srun: error: AssocMaxSubmitJobLimit
srun: error: Unable to allocate resources: Job violates accounting/QOS policy (job submit limit, user's size and/or time limits)

I've read FAQ and none of the standard reasons apply (I indicate project with plenty of credits, there are no other jobs in the squeue, requested resources look pretty basic..)

- You should run `sbatch run.sh` rather than `srun run.sh`

It worked lol, thank you.

## 2025-09-24 session

**Short talk:** Software available on Puhti/Mahti and their effective use ([PDF](https://a3s.fi/saren-2001659-pub/Software_on_Puhti_and_Mahti_2025-09-24.pdf))

#### Course & Webinar Advertisements

- Course: [CSC Computing Environment Part1: Basics](https://csc.fi/en/training-calendar/csc-computing-environment-part-1-basics/) 1.-2.10.2025
- Course: [Data Analysis with R](https://csc.fi/en/training-calendar/data-analysis-with-r-4/) 6.-7.10.2025
- Course: [Geocomputing on the supercomputer](https://csc.fi/en/training-calendar/geocomputing-on-the-supercomputer-2/) 8.-9.10.2025
- Webinar: [Enhancing Data Support: Understanding Reproducibility, Part 1](https://csc.fi/en/training-calendar/enhancing-data-support-understanding-reproducibility-part-1/) 9.10.2025
- Course: [Portable GPU Programming](https://csc.fi/en/training-calendar/portable-gpu-programming/) 21-24.10.2025
- Webinar: [Enhancing Data Support: Practical Reproducibility, Part 2](https://csc.fi/en/training-calendar/enhancing-data-support-practical-reproducibility-part-2/) 29.10.2025
- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - CSC Computing Environment pre-course support session
Room 4 - SDS (Kimmo,Anni)
Room 5 - Tykky
Room 6 - Puhti/scratch/data upload
Room 7 - Disc quota issue
Room 8 - slurm script
Room 9 -
Room 10 -

---

### Questions

:::danger
If you wish you can type your question here before the session. We'll respond to them during the session. **Please join the Zoom call if you have a question, if you cannot join the call you can send the question by email to <[e-mail]>.**
Previous questions and answers can be found in the archives, see links section above.
:::

Q474 I need hands-on help with sd-lock-util and ssh-connection for SD Apply / Anni

Please join breakout room #4

Q475 I need help with tykky /Rodney

- A: write an environemt.yml file and run tykky's conda-containerize on that.

Q476 Hi, As we discuss last week that I have 25GB public data and you suggest me to use PUHTI scratch/project_20... directory to upload data but I couldn't upload using upload tab . Please help me in this regars/ Toriqul Islam(tislam)
I am very sorry, could you please tell me how join room 6
-it will be possible after this short talk
Please join breakout room #6
- A: The web interface file transfer tool has some file size restrictions. You may need to use some other tool. Check our [file transfer](https://docs.csc.fi/data/moving/) Docs page.

Q477 Hi, I am trying to increase the disc quota but does not work / Himat

Please join breakout room #7

Q478 Hi, I'm working on a multi-node training job and I'm not sure if my slurm script can facilitate the communication between the nodes (two nodes) correctly. Could you help me check my slurm script? Thanks! Chao Wang

breakout room no 8

## 2025-09-17 session

**Short talk:** no short talk this time

#### Course & Webinar Advertisements

- Course: [CSC Computing Environment Part1: Basics](https://csc.fi/en/training-calendar/csc-computing-environment-part-1-basics/) 1.-2.10.2025
- Course: [Data Analysis with R](https://csc.fi/en/training-calendar/data-analysis-with-r-4/) 6.-7.10.2025
- Course: [Geocomputing on the supercomputer](https://csc.fi/en/training-calendar/geocomputing-on-the-supercomputer-2/) 8.-9.10.2025
- Webinar: [Enhancing Data Support: Understanding Reproducibility, Part 1](https://csc.fi/en/training-calendar/enhancing-data-support-understanding-reproducibility-part-1/) 9.10.2025
- Webinar: [Enhancing Data Support: Practical Reproducibility, Part 2](https://csc.fi/en/training-calendar/enhancing-data-support-practical-reproducibility-part-2/) 29.10.2025
- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - Bioinformatics tools
Room 4 - Getting started with HPC
Room 5 - Machine learning + sbatch
Room 6 -
Room 7 - VS Code Server with Mahti/Puhti
Room 8 -

---

### Questions

:::danger
If you wish you can type your question here before the session. We'll respond to them during the session. **Please join the Zoom call if you have a question, if you cannot join the call you can send the question by email to <[e-mail]>.**
Previous questions and answers can be found in the archives, see links section above.
:::

- **Q470** I need help to find out if some bioinformatics tools are installed in CSC. If not, I need help with installing:
1. Bam-readcount https://github.com/genome/bam-readcount
2. Facets-suite https://github.com/mskcc/facets-suite
3. PhyloWGS https://github.com/MarkeyBBSRF/Phylowgs_workflow
4. PyClone https://github.com/Roth-Lab/pyclone as well.
5. Additionally, samtools and singularity :) + Sadiksha
- We have a list of installed software on our Docs pages: [Bioscience software](https://docs.csc.fi/apps/by_discipline/#biosciences).
- The above list is not complete so it is also a good idea to check with "module spider" command, e.g. `module spider samtools`
- Samtools is available. The other mentioned software is not. We can look at installing it in a breakout room.
- The "singularity" is now known as "apptainer" and it exists
- General info on [installing own software](https://csc-training.github.io/csc-env-eff/part-2/installing/)
- Breakout room number 4 and Ari-Matti

- **Q471** Question about resource allocation (sbatch jobs) for ML training + Sophie
- Breakout room number 5 with Mats

- **Q472** Resource allocation / squeue for ML training :) + Hyunkyung Choo
- Breakout room number 5 with Mats

- **Q473** How should I credit CSC in a publication? I've used Mahti extensively / Oskar
- Please follow our doc page for citing CSC: https://docs.csc.fi/support/faq/how-to-cite-csc/
- thank you :) - Oskar

## 2025-09-10 session

**Short talk:** Topical issues on data management trainings and CSC service catalog ([slides](https://a3s.fi/Marias_share/data_management_2025.pdf))

#### Course & Webinar Advertisements

- Course: [CSC Computing Environment Part1: Basics](https://csc.fi/en/training-calendar/csc-computing-environment-part-1-basics/) 1.-2.10.2025
- Course: [Data Analysis with R](https://csc.fi/en/training-calendar/data-analysis-with-r-4/) 6.-7.10.2025
- Course: [Geocomputing on the supercomputer](https://csc.fi/en/training-calendar/geocomputing-on-the-supercomputer-2/) 8.-9.10.2025
- Webinar: [Enhancing Data Support: Understanding Reproducibility, Part 1](https://csc.fi/en/training-calendar/enhancing-data-support-understanding-reproducibility-part-1/) 9.10.2025
- Webinar: [Enhancing Data Support: Practical Reproducibility, Part 2](https://csc.fi/en/training-calendar/enhancing-data-support-practical-reproducibility-part-2/) 29.10.2025
- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - Localcolabfold (Ari-Matti)
Room 4 -
Room 5 -
Room 6 -
Room 7 -
Room 8 -

---

### Questions

:::danger
If you wish you can type your question here before the session. We'll respond to them during the session. **Please join the Zoom call if you have a question, if you cannot join the call you can send the question by email to <[e-mail]>.**
Previous questions and answers can be found in the archives, see links section above.
:::

- **Q468** Puhti and **Mahti** will be decommissioned next year -- what happens to the individual servers the computers are made of? Could few of them be donated to poor research group with rack, but only desktop machines there? ;) Jani L.
- I think you mean Puhti and **Mahti**, not Rahti :)
- Sorry, fixed! :D
- As far as I'm aware this is still open. Historically these have sometimes been donated to educational institutions, so if you're interested you could always try to get in contact via [e-mail] :-)
- **Q469** Questions about optimizing localcolabfold/alphafold2, using local db with localcolabfold + Tatu
- Answer / breakout room number

## 2025-09-03 session

**Short talk:** Upcoming billing unit changes ([slides](https://a3s.fi/heli/billing_unit_update.pdf), [recording](https://video.csc.fi/media/t/0_ks2poes3), [blog post](https://research.csc.fi/2025/06/02/billing-unit-renewal-schedule-and-changes/))

#### Course & Webinar Advertisements

- Workshop: [CodeRefinery](https://csc.fi/en/training-calendar/coderefinery/) 9.9.-22.10.2025 -> Practical workshops on tools and techniques for researchers writing code that would like to make their work more reproducible. In the first 3 sessions you will learn how to keep track of your code using version control with git. Other sessions introduce Jupyter (for keeping documentation and code together), Snakemake (a workflow manager), Sphinx (for creating nice looking documentation) and more! Online workshop with possibility to join in-person classroom at CSC in Keilaniemi
- Course: [CSC Computing Environment Part1: Basics](https://csc.fi/en/training-calendar/csc-computing-environment-part-1-basics/) 1.-2.10.2025
- Course: [Data Analysis with R](https://csc.fi/en/training-calendar/data-analysis-with-r-4/) 6.-7.10.2025
- Course: [Geocomputing on the supercomputer](https://csc.fi/en/training-calendar/geocomputing-on-the-supercomputer-2/) 8.-9.10.2025
- Webinar: [Enhancing Data Support: Understanding Reproducibility, Part 1](https://csc.fi/en/training-calendar/enhancing-data-support-understanding-reproducibility-part-1/) 9.10.2025
- Webinar: [Enhancing Data Support: Practical Reproducibility, Part 2](https://csc.fi/en/training-calendar/enhancing-data-support-practical-reproducibility-part-2/) 29.10.2025
- Course: [Practical Deep Learning](https://csc.fi/en/training-calendar/practical-deep-learning-6/), 11.-12.11.2025
- Course: [CSC Computing Environment Part2: Next steps](https://csc.fi/en/training-calendar/csc-computing-environment-part-2-next-steps/) 12.-13.11.2025
- Course: [High Performance R](https://csc.fi/en/training-calendar/high-performance-r-2/) 17.-18.11.2025
- Course: [Using Containers in Supercomputing Environment](https://csc.fi/en/training-calendar/online-using-containers-in-supercomputing-environment/) 25.-27.11.2025
- See [CSC's training calendar](https://csc.fi/en/trainings/training-calendar/) for more

### Zoom breakout rooms and topics

Room 1 - Getting started
Room 2 - Ambassadors
Room 3 - Billing unit changes (Joonas)
Room 4 - R-libraries in SD Desktop
Room 5 - Matlab (Jaan)
Room 6 - Schrödinger (Nino)
Room 7 -
Room 8 -

---

### Questions

:::danger
If you wish you can type your question here before the session. We'll respond to them during the session. **Please join the Zoom call if you have a question, if you cannot join the call you can send the question by email to <[e-mail]>.**
Previous questions and answers can be found in the archives, see links section above.
:::

- **Q462** Quite basic question: Our LUMI project (and its CPU/GPU units) are reaching their end, so we got a new project and resources allocated. Will we still have access to the scratch folder of the previous project, since ~40TB of data are saved there, or should we move these to the new project folder? - Ioanna
- According to LUMI docs the data will be available for 90 days after the project ended: https://docs.lumi-supercomputer.eu/storage/
- So yes, you should transfer the files to the new project (within 90 days). There should be several warning emails about the data access ending to the old project.

- **Q463** When starting Matlab through MPS a parallel pool already exists, however with only one worker and according to Matlab the number of cores is only 1 regardless of requested number of cores - Vincent Verhoeven
- Did you submit the job as in the docs? https://docs.csc.fi/apps/matlab/#submitting-parallel-jobs
- `j = batch(c, @funcParallel, 1, {8}, 'Pool', 8, 'CurrentFolder', '.', 'AutoAddClientPath', false)` the argument `'Pool', 8` defines how many workers are started (one core per worker)
- To the best of my knowledge, yes. The NumWorkersRange is also [n + 1, n + 1] where n is 'Pool', n which fits the documentation
- which matlab version are you using? r2024b? Yes
- we need to investigate, how did you check how many worker matlab starts?
- on puhti: `cd ~/.matlab/generic_cluster_jobs/puhti`
- `cd Job_x` replace the `x` with your job number
- `cat submitScript.sh`
- processes in the slurm script are identified by `--ntasks`
- cores are `--cpus-per-task

    #!/bin/sh
    sbatch --job-name=MATLAB_R2024b_Job26  --output='/users/tollande/.matlab/generic_cluster_jobs/puhti//Job26/Job26.log' --export=NONE --ntasks=9 --cpus-per-task=1 --ntasks-per-core=1 -A project_2001659 --mem-per-cpu=2g -t 00:15:00 -p test --licenses=mdcs:9 '/users/tollande/.matlab/generic_cluster_jobs/puhti//Job26/environmentWrapper.sh'

- https://docs.csc.fi/apps/matlab/#submitting-serial-jobs and add to

    c = parcluster;
    % Replace 'project_id' to your project identifier, otherwise the script will fail.
    c.AdditionalProperties.ComputingProject = 'project_id';
    c.AdditionalProperties.Partition = 'small';
    c.AdditionalProperties.WallTime = '00:15:00';
    c.AdditionalProperties.CPUsPerNode = '';
    c.AdditionalProperties.MemPerCPU = '4g';
    c.AdditionalProperties.GpuCard = '';
    c.AdditionalProperties.GPUsPerNode = '';
    c.AdditionalProperties.EmailAddress = '';
    c.NumThreads = 2;

- **Q464** I have few general questions how to run efficiently alphafold or localcolabfold, so general talk in room on this area would be appreciated + Tatu
- Unfortunately the specialists on these tools are away at the moment. Next week they should be back.

- **Q465** Schrödinger writes: "Starting in 2025-4 will use Job Server by default. Customers who currently use Job Control will need to deploy and configure Job Server." The CSC slurm depends on account from schrodinger_hosts. Job Server does not use schrodinger_hosts. You have a plan for that change? Jukka Lehtonen
- Room 6

- **Q466** Installing packages in python jupyter lab using pip and I had path issues
- The recommeded way to use pip install in Jupyter Lab is documented here: https://docs.csc.fi/computing/webinterface/jupyter/#installing-packages
- Often just `!pip install` works, not always. Doing this should work every time:

import sys 
!{sys.executable} -m pip install --user <package>

- **Q467** Not an actual question but **please update the short talk archive for this autumn** :) Currently the archive is ending at summer break 2025.
- Answer / breakout room number