should be it

This commit is contained in:
2025-10-24 19:21:19 -05:00
parent a4b23fc57c
commit f09560c7b1
14047 changed files with 3161551 additions and 1 deletions

View File

@@ -0,0 +1,24 @@
# name: benchmark/micro/compression/roaring/roaring_array_read.benchmark
# description: Scanning 1GB of ints compressed with Array Containers
# group: [roaring]
name Roaring Scan Array Container
group roaring
storage persistent v1.2.0
load
DROP TABLE IF EXISTS tbl;
PRAGMA force_compression='Roaring';
CREATE TABLE tbl AS SELECT case when i%25=0 then 1337 else null end as a FROM range(0, 250_000_000) tbl(i);
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('tbl') where segment_type in ('VALIDITY')
----
Roaring
run
select count(*) from tbl WHERE a IS NOT NULL;
result I
10000000

View File

@@ -0,0 +1,26 @@
# name: benchmark/micro/compression/roaring/roaring_array_store.benchmark
# description: Writing 250m tuples, with validity masks compressed with Roaring Bitmap Compression (Array Containers)
# group: [roaring]
name Roaring Write Array Container
group roaring
storage persistent v1.2.0
load
CREATE TABLE data_source AS SELECT case when i%25=0 then 1337 else null end as a FROM range(0, 250_000_000) tbl(i);
PRAGMA force_compression='Roaring';
SET checkpoint_threshold = '10.0 GB';
CREATE TABLE test_compression as FROM data_source;
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('test_compression') where segment_type in ('VALIDITY')
----
Roaring
run
CREATE TABLE tbl AS FROM data_source;
checkpoint;
cleanup
DROP TABLE IF EXISTS tbl;

View File

@@ -0,0 +1,24 @@
# name: benchmark/micro/compression/roaring/roaring_bitset_read.benchmark
# description: Scanning 1GB of ints compressed with Run Containers (inverted)
# group: [roaring]
name Roaring Scan Run Container Inverted
group roaring
storage persistent v1.2.0
load
DROP TABLE IF EXISTS tbl;
PRAGMA force_compression='Roaring';
CREATE TABLE tbl AS SELECT case when i%3=0 then 1337 else null end as a FROM range(0, 250_000_000) tbl(i);
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('tbl') where segment_type in ('VALIDITY')
----
Roaring
run
select count(*) from tbl WHERE a IS NOT NULL;
result I
83333334

View File

@@ -0,0 +1,26 @@
# name: benchmark/micro/compression/roaring/roaring_bitset_store.benchmark
# description: Writing 250m tuples, with validity masks compressed with Roaring Bitmap Compression (Run Containers (inverted))
# group: [roaring]
name Roaring Write Run Container Inverted
group roaring
storage persistent v1.2.0
load
CREATE TABLE data_source AS SELECT case when i%3=0 then 1337 else null end as a FROM range(0, 250_000_000) tbl(i);
PRAGMA force_compression='roaring';
SET checkpoint_threshold = '10.0 GB';
CREATE TABLE test_compression as FROM data_source;
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('test_compression') where segment_type in ('VALIDITY')
----
Roaring
run
CREATE TABLE tbl AS FROM data_source;
checkpoint;
cleanup
DROP TABLE IF EXISTS tbl;

View File

@@ -0,0 +1,24 @@
# name: benchmark/micro/compression/roaring/roaring_inverted_array_read.benchmark
# description: Scanning 1GB of ints compressed with Array Containers (inverted)
# group: [roaring]
name Roaring Scan Array Container Inverted
group roaring
storage persistent v1.2.0
load
DROP TABLE IF EXISTS tbl;
PRAGMA force_compression='Roaring';
CREATE TABLE tbl AS SELECT case when i%25=0 then null else 1337 end as a FROM range(0, 250_000_000) tbl(i);
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('tbl') where segment_type in ('VALIDITY')
----
Roaring
run
select count(*) from tbl WHERE a IS NOT NULL;
result I
240000000

View File

@@ -0,0 +1,26 @@
# name: benchmark/micro/compression/roaring/roaring_inverted_array_store.benchmark
# description: Writing 250m tuples, with validity masks compressed with Roaring Bitmap Compression (Array Containers (inverted))
# group: [roaring]
name Roaring Scan Array Container Inverted
group roaring
storage persistent v1.2.0
load
CREATE TABLE data_source AS SELECT case when i%25=0 then null else 1337 end as a FROM range(0, 250_000_000) tbl(i);
PRAGMA force_compression='roaring';
SET checkpoint_threshold = '10.0 GB';
CREATE TABLE test_compression as FROM data_source;
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('test_compression') where segment_type in ('VALIDITY')
----
Roaring
run
CREATE TABLE tbl AS FROM data_source;
checkpoint;
cleanup
DROP TABLE IF EXISTS tbl;

View File

@@ -0,0 +1,24 @@
# name: benchmark/micro/compression/roaring/roaring_inverted_run_read.benchmark
# description: Scanning 1GB of ints compressed with Run Containers (inverted)
# group: [roaring]
name Roaring Scan Run Container Inverted
group roaring
storage persistent v1.2.0
load
DROP TABLE IF EXISTS tbl;
PRAGMA force_compression='Roaring';
CREATE TABLE tbl AS SELECT case when i = 0 or (i % 512 != 0 and (i % 512) < 350 or (i % 512) > 450) then 1337 else null end as a FROM range(0, 250_000_000) tbl(i);
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('tbl') where segment_type in ('VALIDITY')
----
Roaring
run
select count(*) from tbl WHERE a IS NOT NULL;
result I
200195338

View File

@@ -0,0 +1,26 @@
# name: benchmark/micro/compression/roaring/roaring_inverted_run_store.benchmark
# description: Writing 250m tuples, with validity masks compressed with Roaring Bitmap Compression (Run Containers (inverted))
# group: [roaring]
name Roaring Write Run Container Inverted
group roaring
storage persistent v1.2.0
load
CREATE TABLE data_source AS SELECT case when i = 0 or (i % 512 != 0 and (i % 512) < 350 or (i % 512) > 450) then 1337 else null end as a FROM range(0, 250_000_000) tbl(i);
PRAGMA force_compression='Roaring';
SET checkpoint_threshold = '10.0 GB';
CREATE TABLE test_compression as FROM data_source;
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('test_compression') where segment_type in ('VALIDITY')
----
Roaring
run
CREATE TABLE tbl AS FROM data_source;
checkpoint;
cleanup
DROP TABLE IF EXISTS tbl;

View File

@@ -0,0 +1,24 @@
# name: benchmark/micro/compression/roaring/roaring_run_read.benchmark
# description: Scanning 1GB of ints compressed with Run Containers (inverted)
# group: [roaring]
name Roaring Scan Run Container Inverted
group roaring
storage persistent v1.2.0
load
DROP TABLE IF EXISTS tbl;
PRAGMA force_compression='Roaring';
CREATE TABLE tbl AS SELECT case when i = 0 or (i % 512 != 0 and (i % 512) < 350 or (i % 512) > 450) then null else 1337 end as a FROM range(0, 250_000_000) tbl(i);
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('tbl') where segment_type in ('VALIDITY')
----
Roaring
run
select count(*) from tbl WHERE a IS NOT NULL;
result I
49804662

View File

@@ -0,0 +1,27 @@
# name: benchmark/micro/compression/roaring/roaring_run_store.benchmark
# description: Writing 250m tuples, with validity masks compressed with Roaring Bitmap Compression (Run Containers)
# group: [roaring]
name Roaring Write Run Container
group roaring
storage persistent v1.2.0
# Roughly 8 runs per Vector
load
CREATE TABLE data_source AS SELECT case when i = 0 or (i % 512 != 0 and (i % 512) < 350 or (i % 512) > 450) then null else 1337 end as a FROM range(0, 250_000_000) tbl(i);
PRAGMA force_compression='Roaring';
SET checkpoint_threshold = '10.0 GB';
CREATE TABLE test_compression as FROM data_source;
checkpoint;
assert I
select DISTINCT compression from pragma_storage_info('test_compression') where segment_type in ('VALIDITY')
----
Roaring
run
CREATE TABLE tbl AS FROM data_source;
checkpoint;
cleanup
DROP TABLE IF EXISTS tbl;