vatai mpk issues - Githubissues

vatai / mpk

Matrix powers kernel

1 stars 0 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Implement having the results in the same memory location as the input

#42 vatai opened 4 years ago
0
Investigate (and fix) why big matrices get stuck

#41 vatai opened 4 years ago
0
Make proper results to compare communications for various matrices.

#40 utsavsinghal5 closed 4 years ago
3
Prove or verify that mirroring will give us same partitions as we get from METIS

#39 utsavsinghal5 opened 4 years ago
1
Make a separate method for phase processing loop in `comm_comp_patterns`

#38 utsavsinghal5 closed 4 years ago
1
Reading pattern and symmetric matrices

#37 vatai closed 4 years ago
1
Output files have always the same name

#36 vatai closed 4 years ago
1
pdmpk testing (LAPJV testing)

#35 vatai closed 4 years ago
5
Get same partition at the top and bottom (Symmetry, Up-down mirror idea)

#34 utsavsinghal5 opened 4 years ago
1
Implement LAPJV (input generaton) in PDMPK

#33 vatai closed 4 years ago
7
[pdmpk] Optimise `init_idcs`

#32 vatai closed 4 years ago
1
[pdmpk,dmpk] Implement a faster search for the optimal permutation of new partition labels

#31 vatai closed 4 years ago
0
[pdmpk] Unify `init_dict` and `comm_dict`

#30 vatai closed 4 years ago
0
Implement partial vertices

#29 vatai closed 4 years ago
2
Communication hiding

#28 vatai opened 5 years ago
1
Apply optimisations pernode/hybrid optimisation

#27 vatai opened 5 years ago
1
For graph bcsstk28 to program does not produce the correct results

#26 vatai closed 5 years ago
9
Metis does not support loops

#25 vatai closed 5 years ago
2
Remove code duplication in makefiles using include

#24 vatai closed 5 years ago
0
Crash with relatively small size

#23 vatai closed 5 years ago
2
Reduce the size of comm_table

#22 vatai closed 4 years ago
2
Remove the redundant communication of the vertices which can be computed in the partition.

#21 utsavsinghal5 opened 5 years ago
1
Measure performance penalty for unbalanced communication.

#20 vatai opened 5 years ago
0
Optimize communication (more verbose `store_part[]`)

#19 vatai opened 5 years ago
0
Figure out number of phases backward which can do communication with current phase.

#18 utsavsinghal5 closed 5 years ago
3
Redesign `comm_data_t`

#17 vatai closed 5 years ago
1
Avoid copying to/from the send/receive buffers

#16 vatai closed 5 years ago
6
Run it on the university cluster

#15 vatai closed 5 years ago
1
Compare the speed MPI with the OpenMP version

#14 vatai opened 5 years ago
2
Implement measurements

#13 vatai closed 5 years ago
1
Implement `nphase` equals `0` case for mpi

#12 vatai closed 5 years ago
0
Record commit checksum in log files

#11 vatai closed 4 years ago
1
Implement skirt for MPI

#10 vatai closed 5 years ago
0
Implement alltoall MPI calls and execution

#9 vatai closed 5 years ago
0
Debug `MPI_alltoallv()` parameter arrays

#8 vatai closed 5 years ago
1
Debug the `comm_table[]` generation

#7 vatai closed 5 years ago
1
Diagonal elements

#6 vatai closed 5 years ago
3
Custom matrix values

#5 vatai closed 5 years ago
1
Automate testing for the new MPI code

#4 vatai closed 5 years ago
0
Create optimal partition label selecting algorithm

#3 vatai closed 5 years ago
3
Figure out PETSc

#2 vatai opened 5 years ago
0
Doccument skirt code

#1 vatai closed 5 years ago
0