CCDB(https://ccdb.alliancecan.ca/, aka Compute Canada Database)
First, you should debug the code locally to ensure it runs smoothly. Also, make sure that all required custom functions are included in the directory you plan to upload.
Secondly, download Globus for transferring code and data.
Third, establish a connection with cedar using the built-in Command Prompt in Windows.
1
2 | ssh -Y username@cedar.computecanada.ca
password
|
Create an executable file. First, navigate to the directory where the starting code for execution is located using the "cd" command.
Below is the template for the executable file. Prepare it in advance, and after creation, simply copy and paste it in. To save, use Ctrl+O, then Enter, and then Ctrl+X.
1
2
3
4
5
6
7
8
9
10
11 | #!/bin/bash
#SBATCH --time=20-00:00
#SBATCH --nodes=1
#SBATCH --cpus-per-task=12
#SBATCH --account=def-XXXXX
#SBATCH --mem-per-cpu=250G
#SBATCH --output=job_output.log
module load matlab/2023a
srun matlab -nodisplay -singleCompThread -r "MAIN_SPI_RASTER"
|
or
1
2
3
4
5
6
7
8
9
10
11 | #!/bin/bash
#SBATCH --time=20-00:00
#SBATCH --gres=gpu:4
#SBATCH --cpus-per-task=12
#SBATCH --account=def-XXXX
#SBATCH --mem-per-cpu=16384M
#SBATCH --output=job_output.log
module load matlab/2023a
srun matlab -nodisplay -singleCompThread -r "MAIN_SPI_RASTER"
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19 | $ module spider matlab
Versions:
matlab/2014a
matlab/2016b
matlab/2017a
matlab/2018a
matlab/2018b
matlab/2019a
matlab/2019b
matlab/2020a
matlab/2020b
matlab/2020b.4
matlab/2020b.6
matlab/2021a.1
matlab/2021a.5
matlab/2021b.3
matlab/2022a
matlab/2022b.2
matlab/2023a.3
|
Python
1
2
3
4
5
6
7
8
9
10 | #!/bin/bash
#SBATCH --time=1-00:00
#SBATCH --gres=gpu:1
#SBATCH --cpus-per-task=1
#SBATCH --account=def-XXXX
#SBATCH --mem-per-cpu=6384M
#SBATCH --output=job_output.log
module load python/3.11.2
python cc.py
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18 | $ module spider python
Versions:
python/2.7.14
python/2.7.18
python/3.5.4
python/3.6.3
python/3.6.10
python/3.7.0
python/3.7.4
python/3.7.7
python/3.7.9
python/3.8.0
python/3.8.2
python/3.8.10
python/3.9.6
python/3.10.2
python/3.11.2
python/3.11.5
|
After successfully submitting the job, you can check its status at any time using the command "sq". Once the job has completed, the system won't provide any notifications, so it's important to keep an eye on its status.
Use scancel with the job ID to cancel a job:
1
2 | /*Check the running status.*/
$ diskusage_report
|
Notes
| Name |
Storage |
Max Duration |
Globus endpoint |
| Cedar |
1T |
28 days |
computecanada#cedar-globus |
| Graham |
1T |
7 days |
computecanada#graham-globus |
| Beluga |
1T |
7 days |
computecanada#beluga-dtn |
| Narval |
1T |
7 days |
Compute Canada - Narval |
When running on CC, never use absolute paths; always use relative paths.
The CC platform has a maximum file limit of 500K, so it's important to use resources efficiently and reduce the number of files during runtime.
1
2 | % An absolute path
/project/usernumber/username/Mat/SPEI/input/prec/prec_192404.tif
|
CC is a UNIX system and cannot directly run .EXE files. To run an .EXE file on CC, you first need to check if Wine is installed on CC. You can do this by typing 'wine' in the Command Prompt. If it returns the following, it means Wine is already installed.
wine
Usage: wine PROGRAM [ARGUMENTS...] Run the specified program
wine --help Display this help and exit
wine --version Output version information and exit
Check out the size of folder on Compute Canada
% Check the size of the folder
$ du -sh foldername_1 foldername_2
Copy a folder to a specific path
% Copy the target folder to a specific folder
$ cp -r source_folder_path destination_folder_path
References
No comments:
Post a Comment