Performance
These benchmarks compare Fatou's formatting speed with Runic and JuliaFormatter, and its language-server speed and memory use with LanguageServer.jl and JETLS. Fatou runs as a compiled Rust program; the comparison tools run in Julia.
Formatter
We compare Fatou with Runic and JuliaFormatter on source files from JuliaSyntax.jl and DataFrames.jl. Each tool uses its default formatting style.
The single-file and project benchmarks keep each tool loaded between runs, so Julia startup and compilation do not count. The cold-start benchmark below includes those costs.
- Corpora: JuliaSyntax.jl
v0.4.10(09576ca), DataFrames.jlv1.8.2(946c72a) - Versions: Fatou
0.20.0, Runic1.9.0, JuliaFormatter2.12.6, Julia1.13.0 - Host: AMD Ryzen 9 7900 12-Core Processor (Linux x86_64)
- Machine:
terra - Warm-loop iterations: 50 single, 20 project; 3 warmup
- Cold-start iterations: 5 fresh-process runs (single file)
Single files
The three files cover parser internals (parse_stream.jl), a large data table (kinds.jl), and docstring- and macro-heavy library code (abstractdataframe.jl).
The charts show time relative to Fatou: 1 is the Fatou baseline, and 10 means ten times as long. Lower is faster. Each step on the logarithmic axis is a tenfold increase. Tooltips and expandable tables show the measured times and throughput.
[{"scenario":"parse_stream.jl","tool":"Fatou","relative_time":1.0,"throughput_mbps":10.264,"files_ok":1,"total_bytes":41937,"median_ms":4.085757,"relative":"baseline"},{"scenario":"parse_stream.jl","tool":"Runic","relative_time":3.8770810403066065,"throughput_mbps":2.647,"files_ok":1,"total_bytes":41937,"median_ms":15.840811,"relative":"3.88x"},{"scenario":"parse_stream.jl","tool":"JuliaFormatter","relative_time":1.9196322737744804,"throughput_mbps":5.347,"files_ok":1,"total_bytes":41937,"median_ms":7.843151,"relative":"1.92x"},{"scenario":"kinds.jl","tool":"Fatou","relative_time":1.0,"throughput_mbps":15.785,"files_ok":1,"total_bytes":24442,"median_ms":1.548404,"relative":"baseline"},{"scenario":"kinds.jl","tool":"Runic","relative_time":8.127778021756596,"throughput_mbps":1.942,"files_ok":1,"total_bytes":24442,"median_ms":12.585084,"relative":"8.13x"},{"scenario":"kinds.jl","tool":"JuliaFormatter","relative_time":2.643322414563641,"throughput_mbps":5.972,"files_ok":1,"total_bytes":24442,"median_ms":4.092931,"relative":"2.64x"},{"scenario":"abstractdataframe.jl","tool":"Fatou","relative_time":1.0,"throughput_mbps":18.042,"files_ok":1,"total_bytes":100388,"median_ms":5.564171,"relative":"baseline"},{"scenario":"abstractdataframe.jl","tool":"Runic","relative_time":5.655966899651359,"throughput_mbps":3.19,"files_ok":1,"total_bytes":100388,"median_ms":31.470767,"relative":"5.66x"},{"scenario":"abstractdataframe.jl","tool":"JuliaFormatter","relative_time":2.455231336348218,"throughput_mbps":7.348,"files_ok":1,"total_bytes":100388,"median_ms":13.661327,"relative":"2.46x"}]Data table
parse_stream.jl (JuliaSyntax/src/parse_stream.jl)
| Tool | Files | Bytes | Median (ms) | Throughput (MB/s) | Relative |
|---|---|---|---|---|---|
| Fatou | 1 | 41,937 | 4.1 | 10.26 | baseline |
| Runic | 1 | 41,937 | 15.8 | 2.65 | 3.88x |
| JuliaFormatter | 1 | 41,937 | 7.8 | 5.35 | 1.92x |
kinds.jl (JuliaSyntax/src/kinds.jl)
| Tool | Files | Bytes | Median (ms) | Throughput (MB/s) | Relative |
|---|---|---|---|---|---|
| Fatou | 1 | 24,442 | 1.5 | 15.79 | baseline |
| Runic | 1 | 24,442 | 12.6 | 1.94 | 8.13x |
| JuliaFormatter | 1 | 24,442 | 4.1 | 5.97 | 2.64x |
abstractdataframe.jl (DataFrames/src/abstractdataframe/abstractdataframe.jl)
| Tool | Files | Bytes | Median (ms) | Throughput (MB/s) | Relative |
|---|---|---|---|---|---|
| Fatou | 1 | 100,388 | 5.6 | 18.04 | baseline |
| Runic | 1 | 100,388 | 31.5 | 3.19 | 5.66x |
| JuliaFormatter | 1 | 100,388 | 13.7 | 7.35 | 2.46x |
Whole projects
This comparison checks formatting across each project's src/ tree, including file discovery, reading files, and parallel processing. Runic is omitted because it has no in-process directory API. The tables list any files a tool skipped.
[{"scenario":"JuliaSyntax","tool":"Fatou","relative_time":1.0,"throughput_mbps":31.366,"files_ok":15,"total_bytes":332783,"median_ms":10.609532,"relative":"baseline"},{"scenario":"JuliaSyntax","tool":"JuliaFormatter","relative_time":2.5709462962174014,"throughput_mbps":12.2,"files_ok":15,"total_bytes":332783,"median_ms":27.276537,"relative":"2.57x"},{"scenario":"DataFrames","tool":"Fatou","relative_time":1.0,"throughput_mbps":87.524,"files_ok":36,"total_bytes":870581,"median_ms":9.946773,"relative":"baseline"},{"scenario":"DataFrames","tool":"JuliaFormatter","relative_time":3.4198382731766372,"throughput_mbps":25.593,"files_ok":36,"total_bytes":870581,"median_ms":34.016355,"relative":"3.42x"}]Data table
JuliaSyntax (JuliaSyntax/src)
| Tool | Files | Bytes | Median (ms) | Throughput (MB/s) | Relative |
|---|---|---|---|---|---|
| Fatou | 15 | 332,783 | 10.6 | 31.37 | baseline |
| JuliaFormatter | 15 | 332,783 | 27.3 | 12.20 | 2.57x |
DataFrames (DataFrames/src)
| Tool | Files | Bytes | Median (ms) | Throughput (MB/s) | Relative |
|---|---|---|---|---|---|
| Fatou | 36 | 870,581 | 9.9 | 87.52 | baseline |
| JuliaFormatter | 36 | 870,581 | 34.0 | 25.59 | 3.42x |
Cold start
Each run starts a fresh process, formats parse_stream.jl once, and exits. This is the cost of a command-line invocation, including Julia startup, package loading, and compilation for Runic and JuliaFormatter.
[{"tool":"Fatou","relative_time":1.0,"median_ms":5.907823,"throughput_mbps":7.099,"relative":"baseline"},{"tool":"Runic","relative_time":32.0115479085951,"median_ms":189.118559,"throughput_mbps":0.222,"relative":"32.01x"},{"tool":"JuliaFormatter","relative_time":48.59833156815971,"median_ms":287.110341,"throughput_mbps":0.146,"relative":"48.60x"}]Data table
| Tool | Median time | Throughput (MB/s) | vs Fatou |
|---|---|---|---|
| Fatou | 5.9 ms | 7.10 | baseline |
| Runic | 189.1 ms | 0.22 | 32.01x |
| JuliaFormatter | 287.1 ms | 0.15 | 48.60x |
Language servers
We compare Fatou with LanguageServer.jl and JETLS in the same DataFrames.jl workspace. The benchmark opens the same files, collects diagnostics, and sends editor requests to each server.
The servers provide different analyses: Fatou analyzes source statically, LanguageServer.jl indexes the Julia environment, and JETLS performs type inference through JET. These results show the time and memory each editor session uses.
- Workspace: DataFrames.jl
v1.8.2(946c72a), instantiated - Session: 5 files opened (343 KiB of source), diagnostics, symbols, and hovers
- Navigation target:
_namesinsrc/abstractdataframe/selection.jlat line 383 - Versions: Fatou
0.18.0, LanguageServer.jl5.0.0, JETLS7e01ca58(2026-08-08), Julia1.12.6 - Host: AMD Ryzen 9 7900 12-Core Processor (Linux x86_64, 61 GB RAM)
- Machine:
terra
Speed
The startup measurements cover three stages:
- Initialize measures the server's response to the editor's initialization request.
- Workspace ready measures the time from process start until background indexing settles.
- Open files ready measures the time from opening the files until diagnostics arrive and background work settles.
After startup, we time requests for document symbols, hover information, definitions, references, and renaming. The chart shows median response times. Tooltips and expandable tables include the 95th percentile (p95) and the number of results returned, which can differ between servers. The p95 is the time within which 95% of requests finish. Lower is faster.
Readiness
[{"server":"Fatou","metric":"Initialize","value":0.001},{"server":"Fatou","metric":"Workspace ready","value":0.152},{"server":"Fatou","metric":"Open files ready","value":0.054},{"server":"LanguageServer.jl","metric":"Initialize","value":4.14},{"server":"LanguageServer.jl","metric":"Workspace ready","value":27.698},{"server":"LanguageServer.jl","metric":"Open files ready","value":3.297},{"server":"JETLS","metric":"Initialize","value":5.884},{"server":"JETLS","metric":"Workspace ready","value":9.881},{"server":"JETLS","metric":"Open files ready","value":13.537}]Readiness data
| Server | Initialize | Workspace ready | Open files ready |
|---|---|---|---|
| Fatou | 1 ms | 152 ms | 54 ms |
| LanguageServer.jl | 4.14 s | 27.70 s | 3.30 s |
| JETLS | 5.88 s | 9.88 s | 13.54 s |
Warm requests
[{"server":"Fatou","metric":"Document symbols","value":1.286,"p95":1.684,"returned_work":"54–123 symbols (median 70), 24 KiB"},{"server":"Fatou","metric":"Hover","value":0.055,"p95":0.13,"returned_work":"1 result, 162 B"},{"server":"Fatou","metric":"Go to definition","value":0.279,"p95":0.4,"returned_work":"5 locations in 4 files, 879 B"},{"server":"Fatou","metric":"Find references","value":0.542,"p95":0.684,"returned_work":"135 locations in 17 files, 23 KiB"},{"server":"Fatou","metric":"Rename","value":0.609,"p95":0.689,"returned_work":"135 edits in 17 files, 17 KiB"},{"server":"LanguageServer.jl","metric":"Document symbols","value":5.416,"p95":7.962,"returned_work":"372–626 symbols (median 519), 117 KiB"},{"server":"LanguageServer.jl","metric":"Hover","value":0.086,"p95":0.353,"returned_work":"1 result, 709 B"},{"server":"LanguageServer.jl","metric":"Go to definition","value":1.303,"p95":1.379,"returned_work":"5 locations in 4 files, 884 B"},{"server":"LanguageServer.jl","metric":"Find references","value":54.304,"p95":84.146,"returned_work":"135 locations in 17 files, 23 KiB"},{"server":"LanguageServer.jl","metric":"Rename","value":54.571,"p95":59.513,"returned_work":"135 edits in 17 files, 18 KiB"},{"server":"JETLS","metric":"Document symbols","value":2.085,"p95":5.838,"returned_work":"348–578 symbols (median 477), 131 KiB"},{"server":"JETLS","metric":"Hover","value":1.277,"p95":4.232,"returned_work":"1 result, 1 KiB"},{"server":"JETLS","metric":"Go to definition","value":0.62,"p95":0.871,"returned_work":"2 locations in 1 file, 761 B"},{"server":"JETLS","metric":"Find references","value":1.067,"p95":4.385,"returned_work":"135 locations in 17 files, 23 KiB"},{"server":"JETLS","metric":"Rename","value":22.355,"p95":41.46,"returned_work":"135 edits in 17 files, 18 KiB"}]Request timings and returned work
| Server | Request | Median | p95 | Returned work |
|---|---|---|---|---|
| Fatou | Document symbols | 1.29 ms | 1.68 ms | 54–123 symbols (median 70), 24 KiB |
| Fatou | Hover | 0.06 ms | 0.13 ms | 1 result, 162 B |
| Fatou | Go to definition | 0.28 ms | 0.40 ms | 5 locations in 4 files, 879 B |
| Fatou | Find references | 0.54 ms | 0.68 ms | 135 locations in 17 files, 23 KiB |
| Fatou | Rename | 0.61 ms | 0.69 ms | 135 edits in 17 files, 17 KiB |
| LanguageServer.jl | Document symbols | 5.42 ms | 7.96 ms | 372–626 symbols (median 519), 117 KiB |
| LanguageServer.jl | Hover | 0.09 ms | 0.35 ms | 1 result, 709 B |
| LanguageServer.jl | Go to definition | 1.30 ms | 1.38 ms | 5 locations in 4 files, 884 B |
| LanguageServer.jl | Find references | 54.30 ms | 84.15 ms | 135 locations in 17 files, 23 KiB |
| LanguageServer.jl | Rename | 54.57 ms | 59.51 ms | 135 edits in 17 files, 18 KiB |
| JETLS | Document symbols | 2.08 ms | 5.84 ms | 348–578 symbols (median 477), 131 KiB |
| JETLS | Hover | 1.28 ms | 4.23 ms | 1 result, 1 KiB |
| JETLS | Go to definition | 0.62 ms | 0.87 ms | 2 locations in 1 file, 761 B |
| JETLS | Find references | 1.07 ms | 4.38 ms | 135 locations in 17 files, 23 KiB |
| JETLS | Rename | 22.36 ms | 41.46 ms | 135 edits in 17 files, 18 KiB |
Warm requests show median / p95. Each target ran 20 measured rounds after 2 warmup rounds; symbols and hover span 3 files, while definition, references, and rename use _names in src/abstractdataframe/selection.jl.
Memory
The chart shows resident memory (RSS) for each server and its child processes. Baseline is after initialization, Settled is after the open files have been analyzed, and Peak is the highest sample during the session. Each server is measured in one session.
[{"server":"Fatou","metric":"Baseline","value":79.2},{"server":"Fatou","metric":"Settled","value":84.3},{"server":"Fatou","metric":"Peak","value":93.1},{"server":"LanguageServer.jl","metric":"Baseline","value":965.0},{"server":"LanguageServer.jl","metric":"Settled","value":1104.0},{"server":"LanguageServer.jl","metric":"Peak","value":1567.4},{"server":"JETLS","metric":"Baseline","value":792.5},{"server":"JETLS","metric":"Settled","value":1997.9},{"server":"JETLS","metric":"Peak","value":2110.9}]Memory data
| Server | Baseline | Settled | Peak | vs Fatou | Settled after | Doing |
|---|---|---|---|---|---|---|
| Fatou | 79 MB | 84 MB | 93 MB | baseline | 9 s | static analysis, no Julia runtime |
| LanguageServer.jl | 965 MB | 1104 MB | 1567 MB | 13.1x | 41 s | Julia runtime plus a SymbolServer pass over the environment |
| JETLS | 792 MB | 1998 MB | 2111 MB | 23.7x | 33 s | Julia runtime plus type inference through JET |
Command-line memory use
For a CI job or pre-commit hook, Fatou runs once and exits. This table shows peak memory use for format, lint, and parse on individual files and whole source trees. Whole-tree runs process files in parallel, so several syntax trees can be in memory at once.
| Command | Over | Input | Peak RSS | Wall |
|---|---|---|---|---|
fatou format --check | src tree | 850 KiB | 39.2 MB | 20 ms |
fatou lint | src tree | 850 KiB | 26.6 MB | 30 ms |
fatou format --check | one file | 98 KiB | 10.3 MB | 10 ms |
fatou lint | one file | 98 KiB | 9.3 MB | 10 ms |
fatou parse | one file | 98 KiB | 7.7 MB | 10 ms |
fatou --version | process floor | - | 4.6 MB | 0 ms |
true | harness floor | - | 2.2 MB | 0 ms |
The src tree is 36 files, 850 KiB. Lowest of 5 runs each, measured with GNU time.
Reproduce the results
Run these commands from the repository root in the devenv shell:
task bench-corpus # Fetch the pinned Julia projects.
task bench # Measure formatter speed.
task bench-lsenv # Set up LanguageServer.jl and JETLS.
task bench-lsp # Measure language-server speed and memory.The commands write bench/results.json and bench/memory.json. These committed files supply the charts, machine details, and tool versions shown above. Building the documentation reads them without running the benchmarks.

