Releases · ut-parla/Parla.py

Release for SC22 Artifact submission.

What's Changed since 0.0.0

Adding simple kokkos example by @dialecticDolt in #1
Working virtual execution environments by @insertinterestingnamehere in #4
Start running threads after the whole threadpool got initialized by @bozhiyou in #5
Merge auto data move by @zhangsiyu1103 in #6
Clean up changes for remove clone/clone_here from inner/blocked_cholesky by @bozhiyou in #7
Add initialize_thread to component instances. by @arthurp in #20
Blocked cyclic based multi-GPU blocked cholesky + clone_here bug fix by @nicelhc13 in #21
Added number of iterations for tests on the multi-GPU Cholesky by @nicelhc13 in #27
Fix data movement calls for partitioners by @insertinterestingnamehere in #33
Create a new stream for each task. by @arthurp in #32
Automatic data movement based on partitioners w/ moving semantics by @bozhiyou in #22
Automate data movement in QR benchmark by @bozhiyou in #35
Fix our asarray workaround. by @insertinterestingnamehere in #41
Initial working version of persistent data reservation. by @insertinterestingnamehere in #40
Fix Persistent Memory Reservation by @insertinterestingnamehere in #48
Change CPU architecture to have one device. by @arthurp in #45
Fix parla cpu/multiload/numpy dependency cycle. by @insertinterestingnamehere in #58
Add tutorial section for data partitioning/movement by @bozhiyou in #86
Tutorial 2 update by @ag1548 in #92
Tutorial 0 update by @ag1548 in #91
Tutorial 1 update by @ag1548 in #90
Tutorial by @sestephens73 in #94
feat: single device PArray with automatic datamovement by @yinengy in #96
fix: hide cupy in system without GPU by @yinengy in #97
fix: update setup.py for submodule and also requirements.txt by @yinengy in #101
feat: support coherence protocol on PArray for multi devices by @yinengy in #100
fix: add locks to prevent race in updating coherence protocol by @yinengy in #102
Merge dev-scheduler by @nicelhc13 in #103
fix: remove the use of gpu module by @yinengy in #104
Dev fix by @nicelhc13 in #105
Refactoring task classes by @nicelhc13 in #107
Runahead scheduling by @nicelhc13 in #108
Mapper memory tracking and policy updates by @sestephens73 in #110
Fix data race among multiple readers by @nicelhc13 in #111
Fix parray tracking, removing over-deallocation bug by @sestephens73 in #113
Attempt to fix intialization crash by making initialization seqential by @dialecticDolt in #114
Mapper dependency load balance by @sestephens73 in #115
Handle task colocations through two queues by @nicelhc13 in #117
Memory tracking fixes by @sestephens73 in #120
Fix counter + free worker thread bugs by @nicelhc13 in #121
New Parray: support fine-grained data movement by @yinengy in #119
Syncing dev with main by @sestephens73 in #122
Miscellaneous refactoring and robustness enhancement by @bozhiyou in #112
Miscellaneous refactoring, continued by @sestephens73 in #124
Callback dev wlr deadlock by @dialecticDolt in #129
Add Parla-corresponding Dask Cholesky on GPU and CPU by @nicelhc13 in #127
Add cholesky implementations including the Dask version, and fix a data movement task related bug. by @nicelhc13 in #130