switch to ahash for perf (#543)

Cargo.toml

@@ -3,7 +3,6 @@ resolver = "2"
 members = [ "vegafusion-common", "vegafusion-core", "vegafusion-runtime", "vegafusion-python", "vegafusion-wasm", "vegafusion-server", "examples/rust-examples",]
 
 [workspace.dependencies]
-deterministic-hash = "1.0.1"
 async-trait = "0.1.73"
 futures = "0.3.21"
 url = "2.3.1"
@@ -39,6 +38,9 @@ version = "0.51.0"
 version = "0.4.35"
 default-features = false
 
+[workspace.dependencies.ahash]
+version = "0.8.11"
+
 [workspace.dependencies.chrono-tz]
 version = "0.9.0"
 features = [ "case-insensitive", "filter-by-regex",]

vegafusion-common/Cargo.toml

@@ -14,9 +14,6 @@ proto = ["datafusion-proto", "datafusion-proto-common", "prost"]
 [dependencies]
 thiserror = "^1.0.29"
 
-[dependencies.deterministic-hash]
-workspace = true
-
 [dependencies.chrono]
 workspace = true
 optional = true
@@ -25,6 +22,9 @@ optional = true
 workspace = true
 optional = true
 
+[dependencies.ahash]
+workspace = true
+
 [dependencies.serde_json]
 workspace = true
 default-features = false

vegafusion-common/src/data/table.rs

@@ -1,12 +1,14 @@
 use datafusion_common::ScalarValue;
 
+use ahash::RandomState;
 use arrow::{
     array::{ArrayData, ArrayRef, StructArray, UInt32Array},
     compute::concat_batches,
     datatypes::{DataType, Field, Schema, SchemaRef},
     ipc::{reader::StreamReader, writer::StreamWriter},
     record_batch::RecordBatch,
 };
+use std::hash::BuildHasher;
 
 use crate::{
     data::{ORDER_COL, ORDER_COL_DTYPE},
@@ -19,7 +21,6 @@ use arrow::array::{
 };
 #[cfg(feature = "prettyprint")]
 use arrow::util::pretty::pretty_format_batches;
-use std::hash::DefaultHasher;
 use std::{
     hash::{Hash, Hasher},
     io::Cursor,
@@ -434,7 +435,7 @@ impl VegaFusionTable {
     }
 
     pub fn get_hash(&self) -> u64 {
-        let mut hasher = deterministic_hash::DeterministicHasher::new(DefaultHasher::new());
+        let mut hasher = RandomState::with_seed(123).build_hasher();
         self.hash(&mut hasher);
         hasher.finish()
     }

vegafusion-core/Cargo.toml

@@ -25,7 +25,7 @@ log = "0.4.22"
 [dependencies.lazy_static]
 workspace = true
 
-[dependencies.deterministic-hash]
+[dependencies.ahash]
 workspace = true
 
 [dependencies.prost]

vegafusion-core/src/data/dataset.rs

@@ -1,6 +1,4 @@
 use crate::error::Result;
-use std::collections::hash_map::DefaultHasher;
-use std::hash::{Hash, Hasher};
 use vegafusion_common::data::table::VegaFusionTable;
 use vegafusion_common::datafusion_expr::LogicalPlan;
 
@@ -14,11 +12,9 @@ impl VegaFusionDataset {
     pub fn fingerprint(&self) -> String {
         match self {
             VegaFusionDataset::Table { hash, .. } => hash.to_string(),
-            VegaFusionDataset::Plan { plan } => {
-                let mut hasher = deterministic_hash::DeterministicHasher::new(DefaultHasher::new());
-                plan.hash(&mut hasher);
-                hasher.finish().to_string()
-            }
+            VegaFusionDataset::Plan { plan } => ahash::RandomState::with_seed(123)
+                .hash_one(plan)
+                .to_string(),
         }
     }
 
@@ -29,9 +25,7 @@ impl VegaFusionDataset {
 
     pub fn from_table_ipc_bytes(ipc_bytes: &[u8]) -> Result<Self> {
         // Hash ipc bytes
-        let mut hasher = deterministic_hash::DeterministicHasher::new(DefaultHasher::new());
-        ipc_bytes.hash(&mut hasher);
-        let hash = hasher.finish();
+        let hash = ahash::RandomState::with_seed(123).hash_one(&ipc_bytes);
         let table = VegaFusionTable::from_ipc_bytes(ipc_bytes)?;
         Ok(Self::Table { table, hash })
     }

vegafusion-core/src/task_graph/graph.rs

@@ -14,9 +14,8 @@ use crate::task_graph::task_value::TaskValue;
 use crate::proto::gen::tasks::task::TaskKind;
 use crate::proto::gen::tasks::task_value::Data;
 use crate::proto::gen::tasks::TaskValue as ProtoTaskValue;
-use std::collections::hash_map::DefaultHasher;
 use std::convert::TryFrom;
-use std::hash::{Hash, Hasher};
+use std::hash::{BuildHasher, Hash, Hasher};
 
 struct PetgraphEdge {
     output_var: Option<Variable>,
@@ -208,8 +207,7 @@ impl TaskGraph {
         let mut id_fingerprints: Vec<u64> = Vec::with_capacity(self.nodes.len());
         for (i, node) in self.nodes.iter().enumerate() {
             let task = node.task();
-            let mut hasher = deterministic_hash::DeterministicHasher::new(DefaultHasher::new());
-
+            let mut hasher = ahash::RandomState::with_seed(123).build_hasher();
             if let TaskKind::Value(value) = task.task_kind() {
                 // Only hash the distinction between Scalar and Table, not the value itself.
                 // The state fingerprint takes the value into account.
@@ -249,7 +247,7 @@ impl TaskGraph {
         let mut state_fingerprints: Vec<u64> = Vec::with_capacity(self.nodes.len());
         for (i, node) in self.nodes.iter().enumerate() {
             let task = node.task();
-            let mut hasher = deterministic_hash::DeterministicHasher::new(DefaultHasher::new());
+            let mut hasher = ahash::RandomState::with_seed(123).build_hasher();
 
             if matches!(task.task_kind(), TaskKind::Value(_)) {
                 // Hash the task with inline TaskValue

vegafusion-python/Cargo.toml

@@ -34,9 +34,6 @@ workspace = true
 workspace = true
 features = ["tls"]
 
-[dependencies.deterministic-hash]
-version = "1.0.1"
-
 [dependencies.serde]
 version = "1.0.137"
 features = ["derive"]

vegafusion-runtime/Cargo.toml

@@ -43,7 +43,6 @@ async-lock = "2.8.0"
 tempfile = "3.3.0"
 futures-util = "0.3.21"
 bytes = "1.1.0"
-deterministic-hash = "1.0.1"
 log = "0.4.17"
 env_logger = "0.10.0"
 ordered-float = "3.6.0"