Merge pull request #146 from Alexhuszagh/itoa

Increment the itoa benchmark versions.

README.md

@@ -4,7 +4,9 @@ High-performance numeric conversion routines for use in a `no_std` environment.
 
 **Similar Projects**
 
-If you want a minimal, stable, and compile-time friendly version of lexical's float-parsing algorithm, see [minimal-lexical](https://github.com/Alexhuszagh/minimal-lexical). If you want a minimal, performant float parser, recent versions of the Rust standard library should be [sufficient](https://github.com/rust-lang/rust/pull/86761).
+If you want a minimal, stable, and compile-time friendly version of lexical's float-parsing algorithm, see [minimal-lexical](https://github.com/Alexhuszagh/minimal-lexical).
+
+If you want a minimal, performant float parser, recent versions of the Rust standard library should be [sufficient](https://github.com/rust-lang/rust/pull/86761). For high-performance integer formatters, look at [itoa](https://docs.rs/itoa/latest/itoa/). The [metrics](#metrics) section contains a detailed comparison of various crates and their performance in comparison to lexical.
 
 **Table of Contents**
 
@@ -263,7 +265,7 @@ Lexical is extensively used in production, the same float parsing algorithm has
 
 ## Metrics
 
-Various benchmarks, binary sizes, and compile times are shown here:
+Various benchmarks, binary sizes, and compile times are shown here. All the benchmarks can be found on [lexical-benchmarks](https://github.com/Alexhuszagh/lexical-benchmarks?tab=readme-ov-file#latest-results). All benchmarks used a black box to avoid optimizing out the result and leading to misleading metrics.
 
 **Build Timings**
 
@@ -278,29 +280,76 @@ The binary sizes of stripped binaries compiled at optimization level "2". For a
 ![Parse Stripped - Optimization Level "2"](https://raw.githubusercontent.com/Alexhuszagh/rust-lexical/main/assets/size_parse_stripped_opt2_posix.svg)
 ![Write Stripped - Optimization Level "2"](https://raw.githubusercontent.com/Alexhuszagh/rust-lexical/main/assets/size_write_stripped_opt2_posix.svg)
 
-**Benchmarks -- Parse Integer**
+### Benchmarks — Parse Integer
+
+**Random**
+
+A benchmark on randomly-generated integers uniformly distributed over the entire range.
+
+![Uniform Random Data](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/json_random%20-%20parse%20int%20-%20core,lexical.png)
+
+**Simple**
+
+A benchmark on randomly-generated integers from 1-1000.
+
+![Simple Random Data](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/json_simple%20-%20parse%20int%20-%20core,lexical.png)
+
+### Benchmarks — Parse Float
+
+**Real-World Datasets**
+
+A benchmark on parsing floats from various real-world data sets, including Canada, Mesh, and astronomical data (earth).
+
+![Canada](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/canada%20-%20parse%20float%20-%20core,lexical.png)
+
+![Earth](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/earth%20-%20parse%20float%20-%20core,lexical.png)
+
+![Mesh](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/mesh%20-%20parse%20float%20-%20core,lexical.png)
+
+**Random**
+
+A benchmark on randomly-generated integers uniformly distributed over the entire range.
+
+![Random Big Integer](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/random_big_ints%20-%20parse%20float%20-%20core,lexical.png)
+
+**Simple**
+
+A benchmark on randomly-generated integers from 1-1000.
+
+![Random Simple](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/random_simple_int64%20-%20parse%20float%20-%20core,lexical.png)
+
+### Benchmarks — Write Integer
+
+**Random**
+
+A benchmark on randomly-generated integers uniformly distributed over the entire range.
+
+![Random Uniform](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/json_chain_random%20-%20write%20int%20-%20fmt,itoa,lexical.png)
+
+**Simple**
+
+![Random Simple](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/json_simple%20-%20write%20int%20-%20fmt,itoa,lexical.png)
 
-A benchmark on randomly-generated integers uniformly distributed over the entire range. For a more fine-tuned breakdown, see [benchmarks](https://github.com/Alexhuszagh/rust-lexical/blob/main/lexical-parse-integer/docs/Benchmarks.md).
+**Large**
 
-![Uniform Random Data](https://raw.githubusercontent.com/Alexhuszagh/rust-lexical/main/lexical-parse-integer/assets/random_uniform.svg)
+![Random Large](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/random_large%20-%20write%20int%20-%20fmt,itoa,lexical.png)
 
-**Benchmarks -- Parse Float**
+### Benchmarks — Write Float
 
-A benchmark on parsing floats from various real-world data sets. For a more fine-tuned breakdown, see [benchmarks](https://github.com/Alexhuszagh/rust-lexical/blob/main/lexical-parse-float/docs/Benchmarks.md).
+**Big Integer**
 
-![Real Data](https://raw.githubusercontent.com/Alexhuszagh/rust-lexical/main/lexical-parse-float/assets/real.svg)
+A benchmarks for values with a large integers.
 
-**Benchmarks -- Write Integer**
+![Big Integers](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/random_big_ints%20-%20write%20float%20-%20dtoa,fmt,lexical,ryu.png)
 
-A benchmark on writing random integers uniformly distributed over the entire range. For a more fine-tuned breakdown, see [benchmarks](https://github.com/Alexhuszagh/rust-lexical/blob/main/lexical-write-integer/docs/Benchmarks.md).
+**Simple 64-Bit Inteers**
 
-![Uniform Random Data](https://raw.githubusercontent.com/Alexhuszagh/rust-lexical/main/lexical-write-integer/assets/random_uniform.svg)
+![Simple Int64](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/random_simple_int64%20-%20write%20float%20-%20dtoa,fmt,lexical,ryu.png)
 
-**Benchmarks -- Write Float**
 
-A benchmark on writing floats generated via a random-number generator and parsed from a JSON document. For a more fine-tuned breakdown, see [benchmarks](https://github.com/Alexhuszagh/rust-lexical/blob/main/lexical-write-float/docs/Benchmarks.md).
+**Random**
 
-![Random Data](https://raw.githubusercontent.com/Alexhuszagh/rust-lexical/main/lexical-write-float/assets/json.svg)
+![Random](https://github.com/Alexhuszagh/lexical-benchmarks/raw/main/results/da4728e/plot/json%20-%20write%20float%20-%20dtoa,fmt,lexical,ryu.png)
 
 ## Safety
 

lexical-benchmark/input.rs

@@ -449,16 +449,11 @@ macro_rules! ryu_generator {
 
 macro_rules! itoa_generator {
     ($group:ident, $name:expr, $iter:expr) => {{
-        use lexical_util::constants::BUFFER_SIZE;
-        let mut buffer = vec![b'0'; BUFFER_SIZE];
+        let mut buffer = itoa::Buffer::new();
         $group.bench_function($name, |bench| {
             bench.iter(|| {
                 $iter.for_each(|&x| {
-                    itoa::write(&mut buffer, x).unwrap();
-                    black_box(&buffer);
-                    unsafe {
-                        buffer.set_len(0);
-                    }
+                    black_box(buffer.format(x));
                 })
             })
         });

lexical-benchmark/parse-float/canada.rs

@@ -13,7 +13,8 @@ fn canada(criterion: &mut Criterion) {
     group.measurement_time(Duration::from_secs(5));
 
     let data = input::read_lines("canada.txt");
-    parse_float_generator!(group, "canada", data.iter(), f64);
+    parse_float_generator!(group, "f32", data.iter(), f32);
+    parse_float_generator!(group, "f64", data.iter(), f64);
 }
 
 criterion_group!(canada_benches, canada);

lexical-benchmark/parse-float/earth.rs

@@ -13,7 +13,8 @@ fn earth(criterion: &mut Criterion) {
     group.measurement_time(Duration::from_secs(5));
 
     let data = input::read_csv("earth.csv");
-    parse_float_generator!(group, "earth", data.iter(), f64);
+    parse_float_generator!(group, "f32", data.iter(), f32);
+    parse_float_generator!(group, "f64", data.iter(), f64);
 }
 
 criterion_group!(earth_benches, earth);

lexical-benchmark/parse-float/mesh.rs

@@ -13,7 +13,8 @@ fn mesh(criterion: &mut Criterion) {
     group.measurement_time(Duration::from_secs(5));
 
     let data = input::read_lines("mesh.txt");
-    parse_float_generator!(group, "mesh", data.iter(), f64);
+    parse_float_generator!(group, "f32", data.iter(), f32);
+    parse_float_generator!(group, "f64", data.iter(), f64);
 }
 
 criterion_group!(mesh_benches, mesh);

lexical-benchmark/write-integer/Cargo.toml

@@ -17,7 +17,7 @@ features = []
 
 [dev-dependencies]
 criterion = { version = "0.5.0", features = ["html_reports"] }
-itoa = { version = "0.4", features = ["i128"] }
+itoa = { version = "1.0.11" }
 fastrand = "2.1.0"
 lazy_static = "1"
 serde = { version = "1.0", features = ["derive"] }