Add H3 #4
Merged
Add H3 #4
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode characters
added 2 commits October 28, 2021 10:57
jirikraus requested changes Oct 28, 2021
| const int bottom = 0; | ||
| #endif | ||
| | ||
| // TODO: Use MPI_Sendrecv to exchange the data with the top and bottom neighbors |
There was a problem hiding this comment.
I think a bit more help with the argument for the MPI_Sendrecv calls might be needed.
There was a problem hiding this comment.
I agree. Maybe sketch the call in the // TODO a bit. One might also give parts of the options.
Co-authored-by: Jiri Kraus <jkraus@nvidia.com>
Co-authored-by: Jiri Kraus <jkraus@nvidia.com>
Co-authored-by: Jiri Kraus <jkraus@nvidia.com>
Co-authored-by: Jiri Kraus <jkraus@nvidia.com>
Co-authored-by: Jiri Kraus <jkraus@nvidia.com>
Co-authored-by: Jiri Kraus <jkraus@nvidia.com>
AndiH reviewed Oct 29, 2021
| | ||
| ## Hands-ON 5: Multi GPU parallelization with CUDA-aware MPI | ||
| | ||
| ## Task 1: Parallelize the jacobi solver for multiple GPUs using CUDA-aware MPI |
There was a problem hiding this comment.
Suggested change
| ## Task 1: Parallelize the jacobi solver for multiple GPUs using CUDA-aware MPI | |
| ## Task 1: Parallelize Jacobi Solver for Multiple GPUs using CUDA-aware MPI |
| const int bottom = 0; | ||
| #endif | ||
| | ||
| // TODO: Use MPI_Sendrecv to exchange the data with the top and bottom neighbors |
There was a problem hiding this comment.
I agree. Maybe sketch the call in the // TODO a bit. One might also give parts of the options.
Comment on lines +18 to +32
| - Initialize the MPI application | ||
| - Include the MPI header file | ||
| - Determine the local rank and the number of MPI processes | ||
| - Query the number of GPUs visible to the calling process. | ||
| - Use a local communicator to assign one GPU to each MPI process | ||
| - Finalize MPI at the end of the application | ||
| - Compute the 1-D domain decomposition | ||
| - Compute the local chunk size to distribute (ny-2) lines among the process | ||
| - in case `(ny-2)%size != 0` the last process should calculate the remaining rows | ||
| - determine the global (`iy_start_global, iy_end_global`) and local (`iy_start, iy_end`) start and end points in the 2-dimensional grid. | ||
| - Use MPI to exchange the boundaries | ||
| - Compute the top and the bottom neighbor | ||
| - we are using reflecting/periodic boundaries on top and bottom, so rank0's Top neighbor is (size-1) and rank(size-1) bottom neighbor is rank 0 | ||
| - Use MPI_Sendrecv to exchange data between the neighbors | ||
| - use the self-defined MPI_REAL_TYPE. This allows an easy switch between single- and double precision |
There was a problem hiding this comment.
I think that are too many complicated things which can go wrong…
| make run | ||
| ``` | ||
| | ||
| ## Task 1: Optimize load balancing |
There was a problem hiding this comment.
Suggested change
| ## Task 1: Optimize load balancing | |
| ## Advanced Task: Optimize Load Balancing |
This is how it was meant, right?
Co-authored-by: Jiri Kraus <jkraus@nvidia.com>
Co-authored-by: Andreas Herten <a.herten+github@gmail.com>
Co-authored-by: Andreas Herten <a.herten+github@gmail.com>
Co-authored-by: Andreas Herten <a.herten+github@gmail.com>
Co-authored-by: Andreas Herten <a.herten+github@gmail.com>
Co-authored-by: Andreas Herten <a.herten+github@gmail.com>
| We can close this right? |
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit. This suggestion is invalid because no changes were made to the code. Suggestions cannot be applied while the pull request is closed. Suggestions cannot be applied while viewing a subset of changes. Only one suggestion per line can be applied in a batch. Add this suggestion to a batch that can be applied as a single commit. Applying suggestions on deleted lines is not supported. You must change the existing code in this line in order to create a valid suggestion. Outdated suggestions cannot be applied. This suggestion has been applied or marked resolved. Suggestions cannot be applied from pending reviews. Suggestions cannot be applied on multi-line comments. Suggestions cannot be applied while the pull request is queued to merge. Suggestion cannot be applied right now. Please check back later.
No description provided.