Add adjacency list backed graph implementation (#113)

* Move Target into mod.rs * Add adjacency list structure * Add AdjacencyList target methods * From impl for AdjacencyList * Impl directed graph traits for DirectedALGraph * Support DirectedALGraph in GraphBuilder * Reformat imports in csr module * Remove unused newtype * Add from_edge_list benches for adjacency list * Add benchmark for linear edge traversal * Implement UndirectedALGraph * Support Graph500 input for AL graphs * Add new graphs to the prelude * Add more logging to adjacency list creation * Add graph format to graph_app common arguments * Impl Edges for Graph500 input * Run PageRank app on both graph formats * Generate page_rank app using macro * Generate wcc app using macro * Generate sssp app using macro * Rename macro and identifiers * Fix format
neo4j-labs · Oct 27, 2023 · 1e62bc7 · 1e62bc7
1 parent 3c9b1d7
commit 1e62bc7
Show file tree

Hide file tree

Showing 19 changed files with 975 additions and 371 deletions.
diff --git a/Cargo.toml b/Cargo.toml
@@ -54,6 +54,7 @@ rayon = "1.7.0"
 reqwest = { version = "0.11", features = ["stream"] }
 serde_json = "1.0.103"
 serde = { version = "1.0.174", features = ["derive"] }
+tap = "1.0.1"
 tempfile = "3.7.0"
 thiserror = "1.0.44"
 tokio = { version = "1.29.1", features = ["full"], default-features = true }

diff --git a/crates/app/src/app.rs b/crates/app/src/app.rs
@@ -2,15 +2,18 @@ use std::{path::PathBuf, time::Instant};
 
 use graph::prelude::*;
 
+use crate::runner::gen_runner;
 use kommandozeile::*;
 use log::info;
 
 mod loading;
-mod page_rank;
+mod runner;
 mod serialize;
-mod sssp;
 mod triangle_count;
-mod wcc;
+
+gen_runner!(directed+unweighted: page_rank, graph::page_rank::page_rank, PageRankConfig);
+gen_runner!(directed+unweighted: wcc, graph::wcc::wcc_afforest_dss, WccConfig);
+gen_runner!(directed+weighted: sssp, graph::sssp::delta_stepping, DeltaSteppingConfig, f32);
 
 fn main() -> Result<()> {
     let args = setup_clap::<Args>().run()?;
@@ -19,10 +22,10 @@ fn main() -> Result<()> {
     env_logger::init();
 
     match args.algorithm {
-        Algorithm::PageRank { config } => page_rank::page_rank(args.args, config)?,
-        Algorithm::Sssp { config } => sssp::sssp(args.args, config)?,
+        Algorithm::PageRank { config } => page_rank::run(args.args, config)?,
+        Algorithm::Sssp { config } => sssp::run(args.args, config)?,
         Algorithm::TriangleCount { relabel } => triangle_count::triangle_count(args.args, relabel)?,
-        Algorithm::Wcc { config } => wcc::wcc(args.args, config)?,
+        Algorithm::Wcc { config } => wcc::run(args.args, config)?,
         Algorithm::Loading {
             undirected,
             weighted,
@@ -56,6 +59,9 @@ struct CommonArgs {
     #[clap(short, long, value_enum, default_value_t = FileFormat::EdgeList)]
     format: FileFormat,
 
+    #[clap(short, long, value_enum, default_value_t = GraphFormat::CompressedSparseRow)]
+    graph: GraphFormat,
+
     #[clap(long)]
     use_32_bit: bool,
 
@@ -66,6 +72,12 @@ struct CommonArgs {
     warmup_runs: usize,
 }
 
+#[derive(clap::ValueEnum, Debug, Clone)]
+enum GraphFormat {
+    CompressedSparseRow,
+    AdjacencyList,
+}
+
 #[derive(clap::ValueEnum, Debug, Clone)]
 enum FileFormat {
     EdgeList,

diff --git a/crates/app/src/page_rank.rs b/crates/app/src/page_rank.rs
diff --git a/crates/app/src/runner.rs b/crates/app/src/runner.rs
@@ -0,0 +1,136 @@
+macro_rules! gen_runner {
+    (directed+unweighted: $algo_name:ident, $algo_func:expr, $algo_config:ty) => {
+        mod $algo_name {
+            use graph::prelude::*;
+            crate::gen_runner!(__entry: $algo_config);
+            crate::gen_runner!(__run_file_format_all: $algo_config, ());
+            crate::gen_runner!(__run_graph_format: $algo_config, ());
+            crate::gen_runner!(__bench: $algo_func, $algo_config, DirectedNeighbors<NI>, ());
+        }
+    };
+
+    (directed+weighted: $algo_name:ident, $algo_func:expr, $algo_config:ty, $ev_type:ty) => {
+        mod $algo_name {
+            use graph::prelude::*;
+            crate::gen_runner!(__entry: $algo_config);
+            crate::gen_runner!(__run_file_format_edge_list: $algo_config, $ev_type);
+            crate::gen_runner!(__run_graph_format: $algo_config, $ev_type);
+            crate::gen_runner!(__bench: $algo_func, $algo_config, DirectedNeighborsWithValues<NI, $ev_type>, $ev_type);
+        }
+    };
+
+    (__entry: $algo_config:ty) => {
+        pub(crate) fn run(
+            args: $crate::CommonArgs,
+            config: $algo_config,
+        ) -> ::kommandozeile::Result<()> {
+            ::log::info!(
+                "Reading graph ({} bit) from: {:?}",
+                if args.use_32_bit { "32" } else { "64" },
+                args.path
+            );
+
+            if args.use_32_bit {
+                run_::<u32>(args, config)
+            } else {
+                run_::<u64>(args, config)
+            }
+        }
+    };
+
+    (__run_file_format_all: $algo_config:ty, $ev_type:ty) => {
+        fn run_<NI: Idx>(
+            args: $crate::CommonArgs,
+            config: $algo_config,
+        ) -> ::kommandozeile::Result<()>
+        where
+            NI: Idx + ::std::hash::Hash,
+        {
+            match args.format {
+                $crate::FileFormat::EdgeList => {
+                    run__::<NI, EdgeListInput<NI, $ev_type>>(args, config, EdgeListInput::default())
+                }
+                $crate::FileFormat::Graph500 => {
+                    run__::<NI, Graph500Input<NI>>(args, config, Graph500Input::default())
+                }
+            }
+        }
+    };
+
+    (__run_file_format_edge_list: $algo_config:ty, $ev_type:ty) => {
+        fn run_<NI: Idx>(
+            args: $crate::CommonArgs,
+            config: $algo_config,
+        ) -> ::kommandozeile::Result<()>
+        where
+            NI: Idx + ::std::hash::Hash,
+        {
+            match args.format {
+                $crate::FileFormat::EdgeList => {
+                    run__::<NI, EdgeListInput<NI, $ev_type>>(args, config, EdgeListInput::default())
+                }
+                $crate::FileFormat::Graph500 => {
+                    std::panic!("Graph500 is not supported for weighted graphs")
+                }
+            }
+        }
+    };
+
+    (__run_graph_format: $algo_config:ty, $ev_type:ty) => {
+        fn run__<NI, Format>(
+            args: $crate::CommonArgs,
+            config: $algo_config,
+            file_format: Format,
+        ) -> ::kommandozeile::Result<()>
+        where
+            NI: Idx + ::std::hash::Hash,
+            Format: InputCapabilities<NI>,
+            Format::GraphInput: TryFrom<InputPath<::std::path::PathBuf>>,
+            <Format as InputCapabilities<NI>>::GraphInput: Edges<NI = NI, EV = $ev_type>,
+            Error:
+                From<<Format::GraphInput as TryFrom<InputPath<::std::path::PathBuf>>>::Error>,
+        {
+            match args.graph {
+                $crate::GraphFormat::CompressedSparseRow => {
+                    run___::<DirectedCsrGraph<NI, (), $ev_type>, NI, Format>(args, config, file_format)
+                }
+                $crate::GraphFormat::AdjacencyList => {
+                    run___::<DirectedALGraph<NI, (), $ev_type>, NI, Format>(args, config, file_format)
+                }
+            }
+        }
+    };
+
+    (__bench: $algo_func:expr, $algo_config:ty, $neighbors_trait:path, $ev_type:ty) => {
+        fn run___<G, NI, Format>(
+            args: $crate::CommonArgs,
+            config: $algo_config,
+            file_format: Format,
+        ) -> ::kommandozeile::Result<()>
+        where
+            NI: Idx + ::std::hash::Hash,
+            Format: InputCapabilities<NI>,
+            Format::GraphInput: TryFrom<InputPath<::std::path::PathBuf>>,
+            <Format as InputCapabilities<NI>>::GraphInput: Edges<NI = NI, EV = $ev_type>,
+            Error:
+                From<<Format::GraphInput as TryFrom<InputPath<::std::path::PathBuf>>>::Error>,
+            G: Graph<NI> + DirectedDegrees<NI> + $neighbors_trait + Sync,
+            G: TryFrom<(Format::GraphInput, CsrLayout)>,
+            Error: From<<G as TryFrom<(Format::GraphInput, CsrLayout)>>::Error>,
+        {
+            let graph: G = GraphBuilder::new()
+                .csr_layout(CsrLayout::Sorted)
+                .file_format(file_format)
+                .path(args.path)
+                .build()?;
+
+            $crate::time(args.runs, args.warmup_runs, || {
+                $algo_func(&graph, config);
+            });
+
+            Ok(())
+        }
+    };
+}
+
+pub(crate) use gen_runner;
diff --git a/crates/app/src/serialize.rs b/crates/app/src/serialize.rs
@@ -15,6 +15,7 @@ pub(crate) fn serialize(args: CommonArgs, undirected: bool, output: PathBuf) ->
     let CommonArgs {
         path,
         format: _,
+        graph: _,
         runs: _,
         warmup_runs: _,
         use_32_bit,

diff --git a/crates/app/src/sssp.rs b/crates/app/src/sssp.rs