Adds support for large number of items to DeviceScan (NVIDIA#2171)

* make DeviceScan offset type a template parameter * updates tests to use device interface * moves thrust scan to unsigned offset types * adjusts benchmarks to account for used offset types * uses dynamic dispatch to unsigned type * adds tparam docs for NumItemsT * fixes warning about different signedness comparison * adds check for negative num_items in thrust::scan * fixes unused param in is_negative
pciolkosz · Aug 21, 2024 · 0be218f · 0be218f
1 parent 0a27a16
commit 0be218f
Show file tree

Hide file tree

Showing 6 changed files with 159 additions and 92 deletions.
diff --git a/cub/benchmarks/bench/scan/exclusive/base.cuh b/cub/benchmarks/bench/scan/exclusive/base.cuh
@@ -129,7 +129,7 @@ static void basic(nvbench::state& state, nvbench::type_list<T, OffsetT>)
   });
 }
 
-using some_offset_types = nvbench::type_list<nvbench::int32_t>;
+using some_offset_types = nvbench::type_list<nvbench::uint32_t, nvbench::uint64_t>;
 
 NVBENCH_BENCH_TYPES(basic, NVBENCH_TYPE_AXES(all_types, some_offset_types))
   .set_name("base")