Skip to content

Trim redundant benchmark sweep points #871

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
PointKernel wants to merge 1 commit into
base: dev
Choose a base branch
from
+19 −19
Open

Trim redundant benchmark sweep points #871

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 8 additions & 7 deletions benchmarks/benchmark_defaults.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,12 +37,13 @@ auto constexpr SKEW = 0.5;
auto constexpr BATCH_SIZE = 1'000'000;
auto constexpr INITIAL_SIZE = 50'000'000;

auto const N_RANGE = nvbench::range(10'000'000, 100'000'000, 20'000'000);
auto const N_RANGE_CACHE =
std::vector<nvbench::int64_t>{8'000, 80'000, 800'000, 8'000'000, 80'000'000};
auto const OCCUPANCY_RANGE = nvbench::range(0.1, 0.9, 0.1);
auto const MULTIPLICITY_RANGE = std::vector<nvbench::int64_t>{1, 2, 4, 8, 16};
auto const MATCHING_RATE_RANGE = nvbench::range(0.1, 1., 0.1);
auto const SKEW_RANGE = nvbench::range(0.1, 1., 0.1);
// Sweeps keep the end points and the points where the curves bend. Denser sweeps can be run
// on demand with an nvbench axis override, e.g. `-a "Occupancy=[0.1:0.9:0.1]"`.
auto const N_RANGE = std::vector<nvbench::int64_t>{10'000'000, 50'000'000, 90'000'000};

@sleeepyjack sleeepyjack Nov 11, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
auto const N_RANGE = std::vector<nvbench::int64_t>{10'000'000, 50'000'000, 90'000'000};
auto const N_RANGE = std::vector<nvbench::int64_t>{20'000'000, 40'000'000, 80'000'000, 100'000'000};

Optional. I think having at least 4 data points on the major axis gives better interpretability. 90M lands on a weird spot where dynamic_map allocates a third submap with only a few entries in it.

auto const N_RANGE_CACHE = std::vector<nvbench::int64_t>{8'000, 800'000, 8'000'000, 80'000'000};

@sleeepyjack sleeepyjack Nov 11, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
auto const N_RANGE_CACHE = std::vector<nvbench::int64_t>{8'000, 800'000, 8'000'000, 80'000'000};
auto const N_RANGE_CACHE = std::vector<nvbench::int64_t>{2'000'000, 4'000'000, 8'000'000, 16'000'000, 80'000'000};

Small table sizes are purely launch overhead. Added two more data points where the drop should be visible.

auto const OCCUPANCY_RANGE = std::vector<nvbench::float64_t>{0.1, 0.5, 0.8, 0.9};

@sleeepyjack sleeepyjack Nov 11, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
auto const OCCUPANCY_RANGE = std::vector<nvbench::float64_t>{0.1, 0.5, 0.8, 0.9};
auto const OCCUPANCY_RANGE = nvbench::range(0.5, 0.9, 0.1);

0.1 is irrelevant.

auto const MULTIPLICITY_RANGE = std::vector<nvbench::int64_t>{1, 4, 16};
auto const MATCHING_RATE_RANGE = std::vector<nvbench::float64_t>{0.1, 0.5, 1.};
auto const SKEW_RANGE = std::vector<nvbench::float64_t>{0.1, 0.3, 1.};

} // namespace cuco::benchmark::defaults
3 changes: 1 addition & 2 deletions benchmarks/bloom_filter/defaults.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,6 @@ using BF_KEY = nvbench::int64_t;

static constexpr auto BF_N = 1'000'000'000;

auto const BF_SIZE_MB_RANGE_CACHE =
std::vector<nvbench::int64_t>{1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048};
auto const BF_SIZE_MB_RANGE_CACHE = std::vector<nvbench::int64_t>{1, 4, 16, 64, 256, 1024};

@sleeepyjack sleeepyjack Nov 11, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
auto const BF_SIZE_MB_RANGE_CACHE = std::vector<nvbench::int64_t>{1, 4, 16, 64, 256, 1024};
auto const BF_SIZE_MB_RANGE_CACHE = std::vector<nvbench::int64_t>{16, 32, 64, 128, 256, 1024};

Sizes below 16MB are basically noise or launch overhead. I would add back 32 and 128MB as this is the interesting region where throughput drops from L2 to DRAM.


} // namespace cuco::benchmark::defaults
2 changes: 1 addition & 1 deletion benchmarks/hash_function/hash_function_bench.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -201,5 +201,5 @@ NVBENCH_BENCH_TYPES(
.set_name("string_hash_function_eval")
.set_type_axes_names({"Hash"})
.add_int64_axis("NumInputs", {cuco::benchmark::defaults::N / 4})
.add_int64_axis("MinLength", {1, 4})
.add_int64_axis("MinLength", {1})
.add_int64_axis("MaxLength", {4, 32, 64});
8 changes: 4 additions & 4 deletions benchmarks/hyperloglog/hyperloglog_bench.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -151,14 +151,14 @@ NVBENCH_BENCH_TYPES(hyperloglog_e2e,
NVBENCH_TYPE_AXES(TYPE_RANGE, nvbench::type_list<distribution::uniform>))
.set_name("hyperloglog_e2e_uniform")
.set_type_axes_names({"T", "Distribution"})
.add_int64_power_of_two_axis("NumInputs", {28, 29, 30})
.add_int64_axis("SketchSizeKB", {8, 16, 32, 64, 128, 256}) // 256KB uses gmem fallback kernel
.add_int64_power_of_two_axis("NumInputs", {28})

@sleeepyjack sleeepyjack Nov 11, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
.add_int64_power_of_two_axis("NumInputs", {28})
.add_int64_power_of_two_axis("NumInputs", {29})

More stable results.

.add_int64_axis("SketchSizeKB", {32, 64, 128, 256}) // 256KB uses gmem fallback kernel
.add_int64_axis("Multiplicity", {1});

NVBENCH_BENCH_TYPES(hyperloglog_add,
NVBENCH_TYPE_AXES(TYPE_RANGE, nvbench::type_list<distribution::uniform>))
.set_name("hyperloglog_add_uniform")
.set_type_axes_names({"T", "Distribution"})
.add_int64_power_of_two_axis("NumInputs", {28, 29, 30})
.add_int64_axis("SketchSizeKB", {8, 16, 32, 64, 128, 256})
.add_int64_power_of_two_axis("NumInputs", {28})

@sleeepyjack sleeepyjack Nov 11, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
.add_int64_power_of_two_axis("NumInputs", {28})
.add_int64_power_of_two_axis("NumInputs", {29})

More stable results

@sleeepyjack sleeepyjack Nov 11, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

dang this is 8GB in keys. I guess 4Gb is fine too

.add_int64_axis("SketchSizeKB", {32, 64, 128, 256})
.add_int64_axis("Multiplicity", {1});
10 changes: 5 additions & 5 deletions benchmarks/roaring_bitmap/build_bench.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -149,29 +149,29 @@ NVBENCH_BENCH(roaring_bitmap_from_indices_unique)

NVBENCH_BENCH(roaring_bitmap_from_indices_container_cardinality)
.set_name("roaring_bitmap_from_indices_container_cardinality")
.add_int64_axis("ContainerCardinality", {1, 8, 64, 512, 4096, 8192, 32768});
.add_int64_axis("ContainerCardinality", {1, 512, 4096, 32768});

NVBENCH_BENCH(roaring_bitmap_from_sorted_indices_container_cardinality)
.set_name("roaring_bitmap_from_sorted_indices_container_cardinality")
.add_int64_axis("ContainerCardinality", {1, 8, 64, 512, 4096, 8192, 32768});
.add_int64_axis("ContainerCardinality", {1, 512, 4096, 32768});

NVBENCH_BENCH(roaring_bitmap_from_sorted_unique_indices_container_cardinality)
.set_name("roaring_bitmap_from_sorted_unique_indices_container_cardinality")
.add_int64_axis("ContainerCardinality", {1, 8, 64, 512, 4096, 8192, 32768});
.add_int64_axis("ContainerCardinality", {1, 512, 4096, 32768});

NVBENCH_BENCH_TYPES(roaring_bitmap_from_indices,
NVBENCH_TYPE_AXES(nvbench::type_list<distribution::uniform>))
.set_name("roaring_bitmap_from_indices_uniform")
.set_type_axes_names({"Distribution"})
.add_int64_power_of_two_axis("NumInputs", {20, 24, 28})
.add_int64_axis("Multiplicity", {2, 8, 32});
.add_int64_axis("Multiplicity", {2, 32});

NVBENCH_BENCH_TYPES(roaring_bitmap_from_sorted_indices,
NVBENCH_TYPE_AXES(nvbench::type_list<distribution::uniform>))
.set_name("roaring_bitmap_from_sorted_indices")
.set_type_axes_names({"Distribution"})
.add_int64_power_of_two_axis("NumInputs", {20, 24, 28})
.add_int64_axis("Multiplicity", {2, 8, 32});
.add_int64_axis("Multiplicity", {2, 32});

NVBENCH_BENCH(roaring_bitmap_from_sorted_unique_indices)
.set_name("roaring_bitmap_from_sorted_unique_indices")
Expand Down
Loading