diff options
| author | Juan Gomez Luna <juan.gomez@safari.ethz.ch> | 2021-06-16 19:46:05 +0200 |
|---|---|---|
| committer | Juan Gomez Luna <juan.gomez@safari.ethz.ch> | 2021-06-16 19:46:05 +0200 |
| commit | 3de4b495fb176eba9a0eb517a4ce05903cb67acb (patch) | |
| tree | fc6776a94549d2d4039898f183dbbeb2ce013ba9 /SEL/baselines/gpu/README | |
| parent | ef5c3688c486b80a56d3c1cded25f2b2387f2668 (diff) | |
PrIM -- first commit
Diffstat (limited to 'SEL/baselines/gpu/README')
| -rw-r--r-- | SEL/baselines/gpu/README | 21 |
1 files changed, 21 insertions, 0 deletions
diff --git a/SEL/baselines/gpu/README b/SEL/baselines/gpu/README new file mode 100644 index 0000000..e0e9a3e --- /dev/null +++ b/SEL/baselines/gpu/README @@ -0,0 +1,21 @@ +Select (SEL) + +Compilation instructions + + make + +Execution instructions + + ./select 0 50 1258291200 + +Compilation flags + + FLOAT - For single precision arrays (Default: Double precision) + INT - For integer arrays (Note: Sample predicate is only for INT) + THREADS - Thread block size (Default: 1024) + COARSENING - Coarsening factor (Default: 16 (SP and INT); 8 (DP)) + ATOMIC - Global atomics for synchronization (Default: No atomics) + +Read more +J. Gomez-Luna et al., “In-place Data Sliding Algorithms for Many-core Architectures,” ICPP 2015. + |