CNF:CNF Data Generation: Difference between revisions
Jump to navigation
Jump to search
Spyridon97 (talk | contribs) |
Spyridon97 (talk | contribs) |
||
| Line 39: | Line 39: | ||
# Speedup: | # Speedup: | ||
## [[File:partons-speedup.png|500px|File:partons-speedup.png]] | ## [[File:partons-speedup.png|500px|File:partons-speedup.png]] | ||
== Fall 2020 == | |||
# Fix max connections problem with the same username on database | |||
# Update vtk to 9.0.1 inside the container | |||
# Explore if it's possible/worthy to do an exclusive allocation of cores rather than nodes. | |||
Revision as of 12:24, 31 August 2020
Repository
Documentation
Summer 2020
- Add README.md file ✔
Code
Summer 2020
- Set up a MySQL server on Wahab for partons ✔
- Create a container with all the required dependencies ✔
- Split Mitch's Code into:
- A program that splits kinematic values for each core using:
- Uniform Data ✔
- Mesh Points ✔
- Splitting way
- First implementation: split the kinematics dataset like. This way we create areas of kinematics. ✔
- N0_0... N0_m-1,
- N1_0... N1_m-1,
- N2_0... N2_m-1,
- ...,
- Nn-1_0... Nn-1_m-1.
- Second implementation: split the kinematics dataset by creating chunks of size 1 until each kinematics dataset for each node is filled. This way all nodes will have points of similar kinematics. ✔
- N0_0, N1_0, N2_0 .. Nn-1_0,
- N0_1, N1_1, N2_1 .. Nn-1_1,
- N0_2, N1_2, N2_2 .. Nn-1_2,
- ...,
- N0_m-1, N1_m-1, N2_m-1 .. Nn-1_m-1
- First implementation: split the kinematics dataset like. This way we create areas of kinematics. ✔
- A program that computes the Partons Data per core using the kinematics values and the models:
- Compton Form Factors ✔
- Cross-Sections (TODO)
- A program that merges the results of each core into one file ✔
- A program that updates the weights of a mesh using partonsData ✔
- A program that splits kinematic values for each core using:
- Enable shared memory parallelization using CPU affinity, which means pinning a process to a specific core. ✔
- Create a script that runs partons in parallel using shared memory ✔
- Create a script that runs partons in parallel using distributed + shared memory ✔
- Speedup:
Fall 2020
- Fix max connections problem with the same username on database
- Update vtk to 9.0.1 inside the container
- Explore if it's possible/worthy to do an exclusive allocation of cores rather than nodes.