Network

 
AuthorCommitMessageCommit Date
Bob RobeyBob Robey
518cc64fa57Minor cleanup in calc_neighbors_local
Bob RobeyBob Robey
79dd399285cFixing edge case for gpu calc neighbor version
Bob RobeyBob Robey
7560caceaa8Adding new files and deleting old
Bob RobeyBob Robey
c887e12b2a0Rewriting l7_push_update to l7 standards
Bob RobeyBob Robey
73f985b8bc7Moving memory allocation and free into setup and free calls for push
Bob RobeyBob Robey
ac78ff6db4aPutting in sparse communication instead of allgathers in calc_neighbors_local
Bob RobeyBob Robey
e43420692b6Fixing scripts for openmpi 1.6.3
Bob RobeyBob Robey
fff371ef770MMerge branch 'master' of github.com:losalamos/CLAMR
Bob RobeyBob Robey
1d38d4a403eFixing sizes for perfect hash in init
Bob RobeyBob Robey
2e78741af98Fixing compiler warnings
Bob RobeyBob Robey
b8f16e83abcFixing crashes at 128 processors
Bob RobeyBob Robey
d61ea353587Fixing memory leak with hash table for MPI
Bob RobeyBob Robey
2c6d85ceac4More fixes to parallel neighbor finding special cases
Bob RobeyBob Robey
3ba4961b850Fixing some bugs due to rare combinatioins for neighbor calcs
Bob RobeyBob Robey
1edfdd1c498Fixing negative j index in read hash
Bob RobeyBob Robey
b1321433bc0Putting in compact hash table for MPI/GPU
Bob RobeyBob Robey
ee02fd02e2eGetting opt level 3 to work for MPI/GPU version
Bob RobeyBob Robey
d7e14d7219fAdding update of current hashkey to compact hash table for GPU
Bob RobeyBob Robey
2e5511c8772Fixes to compact hash algorithm
Bob RobeyBob Robey
d7e6cbb8783Adding test for reusing same location again in hash
Bob RobeyBob Robey
1267a2c6973Putting back in compact hash for mpi parallel version
Bob RobeyBob Robey
9007c4ee57aFixing boundary cell neighbors
Bob RobeyBob Robey
d6a87f44e32Adding more debug changes
Bob RobeyBob Robey
328f4098500Putting back in debug changes
Bob RobeyBob Robey
764e72980c2Backing out of compact hash changes for parallel runs
Bob RobeyBob Robey
e8ed298975eFixing bug on boundary cell neighbor
Bob RobeyBob Robey
ef0bb4ba2bcAdding OPT LEVEL 3 for MPI/GPU
Bob RobeyBob Robey
19dfda0a2adRemoving setting of corners for compact hash
Bob RobeyBob Robey
88663e9a2f4MMerge branch 'master' of github.com:losalamos/CLAMR
Bob RobeyBob Robey
55ad1718ea4Removing debug statement
Bob RobeyBob Robey
596a45b7348Adding compact hash calls for MPI CPU version of code
Bob RobeyBob Robey
7e339d5f64ePutting in 1 write multiple reads for cpu hash table
Bob RobeyBob Robey
36073417b35Getting gpu compact hash code bypassed on mac laptop
Bob RobeyBob Robey
d17b3551727Adding compact hash code to gpu
Bob RobeyBob Robey
a3c14322efbRestructuring kernels to allow compact hash to be added
Bob RobeyBob Robey
4869e6cc680More compiler issues on moonlight
Bob RobeyBob Robey
6dcce807f87Fixing compiler problems on moonlight
Bob RobeyBob Robey
ae74cc081a0Fixing compile problems
Bob RobeyBob Robey
8877c66423cAdding structure for OPT_LEVEL 4 for GPU
Bob RobeyBob Robey
6112279573fFixing compile on moonlight -- ulong already defined
Bob RobeyBob Robey
daf0b04c0f0Updating opt level 3 to match code for 4
Bob RobeyBob Robey
2657e191f34Reworking calc neighbors to reduce hash reads
brobeybrobey
a133714b653Adding double hashing method
Bob RobeyBob Robey
df17aed5580Adding compact hash algorithm for the CPU -- drops memory usage and for more levels of refinement, speeds up the neighbor calculations.
Bob RobeyBob Robey
c7fadaaf2e6Minor optimizations for neighbor opt_level 3
Bob RobeyBob Robey
5f33bc5951cFixing the 1 write/3 read neighbor algorithm for the single CPU and GPU code. The distributed code still needs to have these added.
Bob RobeyBob Robey
809424becf5Fixing the do_load_balance interface so that clamr_checkall works and doesn't use an updated number of cells for the gpu_do_load_balance call
Bob RobeyBob Robey
252c1ff2a0fFixing clamr_checkall to stash a copy of the new_ncells in gpu_new_ncells so that the load balance for the gpu does not use the updated value
Bob RobeyBob Robey
df9b498bfe1Adding weight array to load balance routines
Bob RobeyBob Robey
84687526cc8Removing ncells_global from load balance argument list