an allocator test I whipped up real quick

test-SmpAllocator.zig

const std = @import("std");
const Allocator = std.mem.Allocator;

var debug_allocator: std.heap.DebugAllocator(.{
    .safety = false,
    .stack_trace_frames = 0,
}) = .init;

pub fn main() !void {
    //const gpa = debug_allocator.allocator();
    const gpa = std.heap.smp_allocator;

    var thread_pool: std.Thread.Pool = undefined;
    try thread_pool.init(.{
        .allocator = gpa,
        .n_jobs = std.Thread.getCpuCount() catch @panic("cpu count"),
    });
    defer thread_pool.deinit();

    var wg: std.Thread.WaitGroup = .{};
    defer wg.wait();

    for (0..1000) |i| {
        thread_pool.spawnWg(&wg, run, .{ gpa, i });
    }
}

fn run(gpa: Allocator, n: usize) void {
    runFallible(gpa, n) catch @panic("OOM");
}

fn runFallible(gpa: Allocator, n: usize) !void {
    var numbers: std.ArrayListUnmanaged(usize) = .empty;
    defer numbers.deinit(gpa);

    for (0..n) |i| {
        try numbers.append(gpa, i);
    }

    var strings: std.AutoArrayHashMapUnmanaged(usize, []const u8) = .empty;
    defer {
        for (strings.values()) |v| gpa.free(v);
        strings.deinit(gpa);
    }

    for (numbers.items) |i| {
        try strings.put(gpa, n, try std.fmt.allocPrint(gpa, "{d}", .{i}));
    }
}

Benchmark 1 (737 runs): ./test-CAllocator
  measurement          mean ± σ            min … max           outliers         delta
  wall_time          6.75ms ±  533us    5.29ms … 9.14ms          9 ( 1%)        0%
  peak_rss           15.3MB ±  836KB    12.7MB … 17.4MB          4 ( 1%)        0%
  cpu_cycles         77.9M  ± 5.44M     59.8M  … 96.7M           9 ( 1%)        0%
  instructions        274M  ± 35.1K      274M  …  274M          56 ( 8%)        0%
  cache_references    770K  ± 59.6K      653K  … 1.21M          64 ( 9%)        0%
  cache_misses       39.3K  ± 5.14K     28.1K  … 69.0K          27 ( 4%)        0%
  branch_misses      57.9K  ± 2.86K     50.8K  … 68.5K          19 ( 3%)        0%
Benchmark 2 (1033 runs): ./test-SmpAllocator
  measurement          mean ± σ            min … max           outliers         delta
  wall_time          4.82ms ±  313us    4.16ms … 6.81ms         45 ( 4%)        ⚡- 28.7% ±  0.6%
  peak_rss           6.04MB ±  751KB    3.93MB … 8.39MB          1 ( 0%)        ⚡- 60.5% ±  0.5%
  cpu_cycles         57.5M  ± 3.50M     45.6M  … 67.5M           2 ( 0%)        ⚡- 26.2% ±  0.5%
  instructions        182M  ± 5.10K      182M  …  182M          30 ( 3%)        ⚡- 33.7% ±  0.0%
  cache_references    548K  ± 29.9K      469K  …  688K          32 ( 3%)        ⚡- 28.9% ±  0.5%
  cache_misses       28.0K  ± 3.50K     18.4K  … 45.3K           8 ( 1%)        ⚡- 28.9% ±  1.0%
  branch_misses      25.2K  ±  817      22.8K  … 29.3K          18 ( 2%)        ⚡- 56.5% ±  0.3%