Rest of TPCH

Cargo.toml

@@ -75,6 +75,7 @@ getrandom = "0.2.14"
 half = { version = "^2", features = ["std", "num-traits"] }
 hashbrown = "0.14.3"
 humansize = "2.1.3"
+indicatif = "0.17.8"
 itertools = "0.13.0"
 lazy_static = "1.4.0"
 leb128 = "0.2.5"
@@ -85,12 +86,14 @@ num_enum = "0.7.2"
 parquet = "52.0.0"
 paste = "1.0.14"
 pin-project = "1.1.5"
+prettytable-rs = "0.10.0"
 prost = "0.13.0"
 prost-build = "0.13.0"
 prost-types = "0.13.0"
 pyo3 = { version = "0.21.2", features = ["extension-module", "abi3-py311"] }
 pyo3-log = "0.11.0"
 rand = "0.8.5"
+rayon = "1.10.0"
 reqwest = { version = "0.12.0", features = ["blocking"] }
 seq-macro = "0.3.5"
 serde = "1.0.197"

bench-vortex/Cargo.toml

@@ -27,13 +27,16 @@ enum-iterator = { workspace = true }
 flexbuffers = { workspace = true }
 futures = { workspace = true, features = ["executor"] }
 humansize = { workspace = true }
+indicatif = { workspace = true }
 itertools = { workspace = true }
 lazy_static = { workspace = true }
 log = { workspace = true }
 mimalloc = { workspace = true }
 object_store = { workspace = true, features = ["aws"] }
 parquet = { workspace = true, features = [] }
+prettytable-rs = { workspace = true }
 rand = { workspace = true }
+rayon = { workspace = true }
 reqwest = { workspace = true }
 serde = { workspace = true }
 simplelog = { workspace = true }

bench-vortex/benches/tpch_benchmark.rs

@@ -1,5 +1,4 @@
 use bench_vortex::tpch::dbgen::{DBGen, DBGenOptions};
-use bench_vortex::tpch::query::Q1;
 use bench_vortex::tpch::{load_datasets, Format};
 use criterion::{criterion_group, criterion_main, Criterion};
 use tokio::runtime::Builder;
@@ -10,50 +9,80 @@ fn benchmark(c: &mut Criterion) {
     // Run TPC-H data gen.
     let data_dir = DBGen::new(DBGenOptions::default()).generate().unwrap();
 
-    let mut group = c.benchmark_group("tpch q1");
-    group.sample_size(10);
-
-    let ctx = runtime
+    let vortex_no_pushdown_ctx = runtime
         .block_on(load_datasets(
             &data_dir,
             Format::Vortex {
                 disable_pushdown: false,
             },
         ))
         .unwrap();
-    group.bench_function("vortex-pushdown", |b| {
-        b.to_async(&runtime)
-            .iter(|| async { ctx.sql(Q1).await.unwrap().collect().await.unwrap() })
-    });
-
-    let ctx = runtime
+    let vortex_ctx = runtime
         .block_on(load_datasets(
             &data_dir,
             Format::Vortex {
                 disable_pushdown: true,
             },
         ))
         .unwrap();
-    group.bench_function("vortex-nopushdown", |b| {
-        b.to_async(&runtime)
-            .iter(|| async { ctx.sql(Q1).await.unwrap().collect().await.unwrap() })
-    });
-
-    let ctx = runtime
+    let csv_ctx = runtime
         .block_on(load_datasets(&data_dir, Format::Csv))
         .unwrap();
-    group.bench_function("csv", |b| {
-        b.to_async(&runtime)
-            .iter(|| async { ctx.sql(Q1).await.unwrap().collect().await.unwrap() })
-    });
-
-    let ctx = runtime
+    let arrow_ctx = runtime
         .block_on(load_datasets(&data_dir, Format::Arrow))
         .unwrap();
-    group.bench_function("arrow", |b| {
-        b.to_async(&runtime)
-            .iter(|| async { ctx.sql(Q1).await.unwrap().collect().await.unwrap() })
-    });
+
+    for q in 1..=22 {
+        if q == 15 {
+            // DataFusion does not support query 15 since it has multiple SQL statements.
+        }
+
+        let query = bench_vortex::tpch::tpch_query(q);
+
+        let mut group = c.benchmark_group(format!("tpch_q{q}"));
+        group.sample_size(10);
+
+        group.bench_function("vortex-pushdown", |b| {
+            b.to_async(&runtime).iter(|| async {
+                vortex_ctx
+                    .sql(&query)
+                    .await
+                    .unwrap()
+                    .collect()
+                    .await
+                    .unwrap()
+            })
+        });
+
+        group.bench_function("vortex-nopushdown", |b| {
+            b.to_async(&runtime).iter(|| async {
+                vortex_no_pushdown_ctx
+                    .sql(&query)
+                    .await
+                    .unwrap()
+                    .collect()
+                    .await
+                    .unwrap()
+            })
+        });
+
+        group.bench_function("csv", |b| {
+            b.to_async(&runtime)
+                .iter(|| async { csv_ctx.sql(&query).await.unwrap().collect().await.unwrap() })
+        });
+
+        group.bench_function("arrow", |b| {
+            b.to_async(&runtime).iter(|| async {
+                arrow_ctx
+                    .sql(&query)
+                    .await
+                    .unwrap()
+                    .collect()
+                    .await
+                    .unwrap()
+            })
+        });
+    }
 }
 
 criterion_group!(benches, benchmark);