zflame - Flamegraph Profiling

zflame is a cutting-edge flamegraph profiling tool designed for the Zig programming language, aimed at simplifying performance analysis and optimization. By leveraging Zig's low-level capabilities, zflame provides detailed, interactive flamegraphs that help developers identify and address performance bottlenecks in their applications.

Features

• 🔥 Generate flamegraphs from various profiler formats (perf, DTrace, sample, etc.)
• 📊 Differential flamegraphs for performance regression analysis
• 🎨 Customizable color schemes and rendering options
• 📈 Stack trace collapsing with multiple algorithm implementations
• 🚀 Streaming parser design for handling large datasets
• 🔧 Both CLI tool and library APIs available

Installation

Requirements

• Zig 0.15.1 or later
• No external dependencies required

Building from Source

git clone https://github.com/hendriknielaender/zflame
cd zflame
zig build -Doptimize=ReleaseFast

The binary will be available at zig-out/bin/zflame.

Usage

CLI Tool

Generate a flamegraph from perf output:

# Record performance data
perf record -F 99 -g ./your_program

# Generate perf script output
perf script > perf.out

# Create flamegraph
zflame perf perf.out > flamegraph.svg

Supported input formats:

• perf - Linux perf events
• dtrace - DTrace stack traces
• sample - Instruments.app sample format
• vtune - Intel VTune Profiler
• xctrace - Xcode Instruments

Differential Flamegraphs

Compare performance between two runs:

zflame diff-folded before.folded after.folded | zflame flamegraph > diff.svg

Library Usage

const std = @import("std");
const zflame = @import("zflame");

pub fn main() !void {
    var gpa = std.heap.GeneralPurposeAllocator(.{}){};
    defer _ = gpa.deinit();
    const allocator = gpa.allocator();

    // Parse perf output
    const perf_data = try std.fs.cwd().readFileAlloc(allocator, "perf.out", 1024 * 1024);
    defer allocator.free(perf_data);

    // Collapse stack traces
    var folder = try zflame.perf.Folder.init(.{});
    defer folder.deinit();
    
    const collapsed = try folder.collapse(allocator, perf_data);
    defer allocator.free(collapsed);

    // Generate flamegraph
    const options = zflame.flamegraph.Options{
        .title = "CPU Profile",
        .count_name = "samples",
        .color_scheme = .hot,
    };
    
    const svg = try zflame.flamegraph.generate(allocator, collapsed, options);
    defer allocator.free(svg);
    
    try std.fs.cwd().writeFile("flamegraph.svg", svg);
}

Performance

Benchmarks available in benchmarks/ directory.

Architecture

The project follows a modular design:

src/
├── collapse/        # Stack trace collapsing algorithms
│   ├── perf.zig    # Linux perf format
│   ├── dtrace.zig  # DTrace stacks
│   └── ...         # Other formats
├── flamegraph/      # SVG generation
│   ├── color.zig   # Color schemes
│   └── parser.zig  # Folded format parser
├── differential.zig # Differential analysis
└── main.zig        # CLI entry point

Acknowledgments

This project is a Zig port of inferno by Jon Gjengset. The original Rust implementation provided the algorithmic foundation and design inspiration for zflame.

Additional thanks to:

• Brendan Gregg for inventing flamegraphs and the original implementation

License

MIT License - see LICENSE for details.

Related Projects

• inferno - The original Rust implementation
• FlameGraph - Original Perl implementation