Problem:

Given a list of posts, compute the top 5 related posts for each post based on the number of shared tags.

Steps

• Read the posts JSON file.
• Iterate over the posts and populate a map containing: tag -> List<int>, with the int representing the post index of each post with that tag.
• Iterate over the posts and for each post:
  • Create a map: PostIndex -> int to track the number of shared tags
  • For each tag, Iterate over the posts that have that tag
  • For each post, increment the shared tag count in the map.
• Sort the related posts by the number of shared tags.
• Write the top 5 related posts for each post to a new JSON file.

Run Benchmark

./run.sh go | rust | python | all

# windows (powershell)
./run.ps1 go | rust | python | all
# OR
pwsh ./run.ps1 go | rust | python | all

# Docker (check the dockerfiles/base.Dockerfile for available variables)
./gen_dockerfile.sh -b go | rust | python | all
# THEN

./docker_run.sh go | rust | python | all
# OR use the image directly
docker run -e TEST_NAME=go -it --rm go_databench

Rules

No:

• FFI (including assembly inlining)
• Unsafe code blocks
• Custom benchmarking
• Disabling runtime checks (bounds etc)
• Specific hardware targeting
• SIMD for single threaded solutions
• Hardcoding number of posts
• Lazy evaluation (Unless results are computed at runtime and timed)
• Computation Caching

Must:

• Support up to 100,000 posts
• Support UTF8 strings
• Parse json at runtime
• Support up to 100 tags
• Represent tags as strings
• Be production ready
• Use less than 8GB of memory

Updated Results from github workflow (raw data)

VM Specs ( AWS c7a.xlarge-4vCPU-8GB-RAM-Ubuntu 22.04 )

Language	Time (5k posts)	20k posts	60k posts	Total
Julia HO1	6.80 ms	23.00 ms	99.33 ms	129.13 ms
D HO1	11.21 ms	42.84 ms	122.06 ms	176.11 ms
D (v2)	13.97 ms	$\textsf{\color{lightgreen}146.70 ms}$	$\textsf{\color{lightgreen}1.14 s}$	1.30 s
Rust	$\textsf{\color{lightgreen}11.54 ms}$	149.96 ms	1.30 s	1.46 s
c3	13.00 ms	164.67 ms	1.33 s	1.51 s
C++	16.10 ms	202.67 ms	1.72 s	1.94 s
Zig	17.00 ms	233.67 ms	1.99 s	2.24 s
Odin	18.74 ms	248.22 ms	2.12 s	2.39 s
Go	19.96 ms	284.42 ms	2.49 s	2.79 s
Neat	22.52 ms	301.30 ms	2.54 s	2.87 s
Java (JIT)	24.60 ms	299.00 ms	2.62 s	2.94 s
C# (JIT)	22.53 ms	314.86 ms	2.76 s	3.10 s
C# (AOT)	21.48 ms	318.20 ms	2.79 s	3.12 s
Haskell	26.65 ms	347.84 ms	2.81 s	3.19 s
Nim	22.34 ms	337.00 ms	2.95 s	3.31 s
F# (JIT)	25.02 ms	354.38 ms	3.02 s	3.40 s
Julia	23.66 ms	350.96 ms	3.10 s	3.47 s
Vlang	26.39 ms	372.67 ms	3.24 s	3.64 s
D	29.92 ms	413.98 ms	3.60 s	4.05 s
Swift	36.61 ms	482.22 ms	4.19 s	4.71 s
F# (AOT)	37.92 ms	570.90 ms	5.07 s	5.68 s
Java (GraalVM)	33.30 ms	504.00 ms	5.47 s	6.01 s
Crystal	45.64 ms	690.35 ms	6.03 s	6.77 s
Numba	66.15 ms	827.15 ms	6.94 s	7.83 s
Pypy	72.62 ms	858.90 ms	7.38 s	8.31 s
LuaJIT	73.52 ms	930.36 ms	7.84 s	8.84 s
JS (Bun)	80.80 ms	975.00 ms	8.49 s	9.55 s
Dart AOT	69.80 ms	1.07 s	9.43 s	10.57 s
Dart VM	61.50 ms	989.67 ms	9.94 s	10.99 s
JS (Deno)	96.40 ms	1.17 s	10.61 s	11.88 s
JS (Node)	99.00 ms	1.12 s	11.03 s	12.25 s
Clojure	111.90 ms	1.31 s	10.97 s	12.39 s
Common Lisp (SBCL)	154.00 ms	1.34 s	11.20 s	12.69 s
Ocaml	99.40 ms	1.46 s	13.05 s	14.61 s
Typed Racket	136.36 ms	1.96 s	16.31 s	18.41 s
Racket	135.03 ms	2.04 s	16.69 s	18.87 s
Scala Native	287.70 ms	3.51 s	30.07 s	33.87 s
LuaJIT (JIT OFF)	630.13 ms	8.72 s	83.86 s	93.21 s
Erlang	758.95 ms	12.50 s	107.46 s	120.72 s
Lua	976.54 ms	15.09 s	136.84 s	152.90 s
Python	1.51 s	24.84 s	215.18 s	241.53 s
Lobster (JIT)	1.63 s	25.43 s	227.06 s	254.13 s
Ruby	1.78 s	28.77 s	254.93 s	285.48 s

Multicore Results

Language	Time (5k posts)	20k posts	60k posts	Total
D Concurrent (v2)	7.21 ms	$\textsf{\color{lightgreen}54.85 ms}$	$\textsf{\color{lightgreen}326.77 ms}$	388.83 ms
C# Concurrent (JIT)	6.56 ms	55.32 ms	450.54 ms	512.42 ms
C++ Concurrent	$\textsf{\color{lightgreen}5.00 ms}$	58.00 ms	477.00 ms	540.00 ms
C# Concurrent (AOT)	5.27 ms	60.57 ms	487.37 ms	553.22 ms
D Concurrent	8.84 ms	76.34 ms	560.38 ms	645.56 ms
Rust Concurrent	5.43 ms	69.22 ms	602.36 ms	677.00 ms
Nim Concurrent	6.17 ms	90.26 ms	657.74 ms	754.17 ms
Go Concurrent	6.89 ms	92.48 ms	771.56 ms	870.92 ms
F# Concurrent (AOT)	8.40 ms	114.00 ms	1.00 s	1.13 s
F# Concurrent	8.80 ms	122.33 ms	1.08 s	1.21 s
Swift Concurrent	13.24 ms	148.23 ms	1.20 s	1.37 s
Julia Concurrent	11.02 ms	159.49 ms	1.40 s	1.57 s
Java Concurrent (JIT)	73.40 ms	221.67 ms	1.41 s	1.70 s
Numba Concurrent	25.82 ms	223.30 ms	1.62 s	1.87 s
Zig Concurrent	17.50 ms	222.24 ms	1.86 s	2.10 s
Java (GraalVM) Concurrent	20.00 ms	294.67 ms	1.81 s	2.12 s

Old Results with details (on my machine)

Language	Processing Time	Total (+ I/O)	Details
Rust	-	4.5s	Initial
Rust v2	-	2.60s	Replace std HashMap with fxHashMap by phazer99
Rust v3	-	1.28s	Preallocate and reuse map and unstable sort by vdrmn and Darksonn
Rust v4	-	0.13s	Use Post index as key instead of Pointer and Binary Heap by RB5009
Rust v5	38ms	52ms	Rm hashing from loop and use vec[count] instead of map[index]count by RB5009
Rust v6	23ms	36ms	Optimized Binary Heap Ops by scottlamb
Rust Rayon	9ms	22ms	Parallelize by masmullin2000
Rust Rayon	8ms	22ms	Remove comparison out of hot loop
⠀	⠀	⠀	⠀
Go	-	1.5s	Initial
Go v2	-	80ms	Add rust optimizations
Go v3	56ms	70ms	Use goccy/go-json
Go v3	34ms	55ms	Use generic binaryheap by DrBlury
Go v4	26ms	50ms	Replace binary heap with custom priority queue
Go v5	20ms	43ms	Remove comparison out of hot loop
Go Con	10ms	33ms	Go concurrency by tirprox and DrBlury
Go Con v2	5ms	29ms	Use arena, use waitgroup, rm binheap by DrBlury
⠀	⠀	⠀	⠀
Python	-	7.81s	Initial
Python v2	1.35s	1.53s	Add rust optimizations by dave-andersen
Numpy	0.57s	0.85s	Numpy implementation by Copper280z
⠀	⠀	⠀	⠀
Crystal	50ms	96ms	Inital w/ previous optimizations
Crystal v2	33ms	72ms	Replace binary heap with custom priority queue
⠀	⠀	⠀	⠀
Odin	110ms	397ms	Ported from golang code
Odin v2	104ms	404ms	Remove comparison out of hot loop
⠀	⠀	⠀	⠀
Dart VM	125ms	530ms	Ported from golang code
Dart bin	274ms	360ms	Compiled executable
⠀	⠀	⠀	⠀
Vlang	339ms	560ms	Ported from golang code
⠀	⠀	⠀	⠀
Zig	80ms	110ms	Provided by akhildevelops

Footnotes

1. Uses specialized datastructures meant for demonstration purposes: more ↩ ↩²