1️⃣🐝🏎️ The One Billion Row Challenge

Status Jan 12: As there has been such a large number of entries to this challenge so far (100+), and this is becoming hard to manage, please only create new submissions if you expect them to run in 10 seconds or less on the evaluation machine.

Status Jan 1: This challenge is open for submissions!

The One Billion Row Challenge (1BRC) is a fun exploration of how far modern Java can be pushed for aggregating one billion rows from a text file. Grab all your (virtual) threads, reach out to SIMD, optimize your GC, or pull any other trick, and create the fastest implementation for solving this task!

The text file contains temperature values for a range of weather stations. Each row is one measurement in the format <string: station name>;<double: measurement>, with the measurement value having exactly one fractional digit. The following shows ten rows as an example:

Hamburg;12.0
Bulawayo;8.9
Palembang;38.8
St. John's;15.2
Cracow;12.6
Bridgetown;26.9
Istanbul;6.2
Roseau;34.4
Conakry;31.2
Istanbul;23.0

The task is to write a Java program which reads the file, calculates the min, mean, and max temperature value per weather station, and emits the results on stdout like this (i.e. sorted alphabetically by station name, and the result values per station in the format <min>/<mean>/<max>, rounded to one fractional digit):

{Abha=-23.0/18.0/59.2, Abidjan=-16.2/26.0/67.3, Abéché=-10.0/29.4/69.0, Accra=-10.1/26.4/66.4, Addis Ababa=-23.7/16.0/67.0, Adelaide=-27.8/17.3/58.5, ...}

Submit your implementation by Jan 31 2024 and become part of the leaderboard!

Results

These are the results from running all entries into the challenge on eight cores of a Hetzner AX161 dedicated server (32 core AMD EPYC™ 7502P (Zen2), 128 GB RAM).

#	Result (m:s.ms)	Implementation	JDK	Submitter	Notes
1	00:02.195	link	21.0.1-graal	Thomas Wuerthinger, Quan Anh Mai, Alfonso² Peterssen	GraalVM native binary, uses Unsafe
2	00:02.248	link	21.0.1-graal	Artsiom Korzun	GraalVM native binary, uses Unsafe
3*	00:02.313	link	21.0.1-graal	Roy van Rijn	GraalVM native binary, uses Unsafe
3*	00:02.336	link	21.0.1-graal	Van Phu DO	GraalVM native binary, uses Unsafe
	00:02.575	link	21.0.1-open	Quan Anh Mai	uses Unsafe
	00:02.909	link	21.0.1-graal	Jaromir Hamala	uses Unsafe
	00:03.258	link	21.0.1-open	Quan Anh Mai
	00:03.376	link	21.0.1-graal	Marko Topolnik	uses Unsafe
	00:03.714	link	21.0.1-graal	Jason Nochlin
	00:03.718	link	21.0.1-graal	zerninv	uses Unsafe
	00:03.902	link	21.0.1-open	Juan Parera
	00:03.959	link	21.0.1-open	gonix
	00:03.966	link	21.0.1-open	Jin Cong Ho	uses Unsafe
	00:03.990	link	21.0.1-graal	Elliot Barlas	uses Unsafe
	00:04.066	link	21.0.1-open	JesseVanRooy	uses Unsafe
	00:04.154	link	21.0.1-open	John Ziamos	uses Unsafe
	00:04.551	link	21.0.1-graal	Roman Musin	uses Unsafe
	00:04.741	link	21.0.1-open	Cliff Click	uses Unsafe
	00:04.823	link	21.0.1-graal	Jamal Mulla	uses Unsafe
	00:04.920	link	21.0.1-graal	Subrahmanyam
	00:04.959	link	21.0.1-graal	Yavuz Tas	uses Unsafe
	00:05.142	link	21.0.1-open	Arjen Wisse
	00:05.235	link	21.0.1-open	unbounded
	00:05.336	link	21.0.1-tem	Peter Levart
	00:05.478	link	21.0.1-open	Olivier Bourgain	uses Unsafe
	00:05.887	link	21.0.1-graal	Charlie Evans	uses Unsafe
	00:05.960	link	21.0.1-graal	Vaidhy Mayilrangam	uses Unsafe
	00:05.979	link	21.0.1-graal	Sam Pullara
	00:06.166	link	21.0.1-open	Jamie Stansfield
	00:06.257	link	21.0.1-graal	Stefan Sprenger	uses Unsafe
	00:06.415	link	21.0.1-open	Arman Sharif	uses Unsafe
	00:06.654	link	21.0.1-graal	Jaroslav Bachorik
	00:06.576	link	21.0.1-open	Andrew Sun	uses Unsafe
	00:06.715	link	21.0.1-open	Algirdas Raščius
	00:06.872	link	21.0.1-open	Dr Ian Preston
	00:07.240	link	java	giovannicuccu
	00:07.680	link	21.0.1-graal	Xylitol	uses Unsafe
	00:07.730	link	21.0.1-open	Johannes Schüth
	00:07.925	link	21.0.1-graal	Ricardo Pieper
	00:07.913	link	21.0.1-open	parkertimmins
	00:08.167	link	21.0.1-tem	Dimitar Dimitrov
	00:08.214	link	21.0.1-open	deemkeen
	00:08.398	link	21.0.1-open	Parth Mudgal	uses Unsafe
	00:08.489	link	21.0.1-graal	Bang NGUYEN
	00:08.517	link	21.0.1-graal	ags	uses Unsafe
	00:08.557	link	21.0.1-graal	Adrià Cabeza
	00:08.622	link	21.0.1-graal	Keshavram Kuduwa
	00:08.689	link	21.0.1-open	Roman Stoffel
	00:08.752	link	21.0.1-graal	Anita SV
	00:08.892	link	21.0.1-open	Roman Romanchuk
	00:09.020	link	21.0.1-open	yemreinci
	00:09.071	link	21.0.1-open	Gabriel Reid
	00:09.352	link	21.0.1-graal	Filip Hrisafov
	00:09.867	link	21.0.1-graal	Ricardo Pieper
	00:09.945	link	21.0.1-open	Anthony Goubard
	00:10.092	link	21.0.1-graal	Pratham
	00:10.127	link	21.0.1-open	Parth Mudgal	uses Unsafe
	00:11.577	link	21.0.1-open	Eve
	00:10.473	link	21.0.1-open	Anton Rybochkin
	00:11.119	link	21.0.1-open	lawrey
	00:11.167	link	21.0.1-open	Nick Palmer
	00:11.405	link	21.0.1-graal	Rafael Merino García
	00:11.433	link	21.0.1-graal	Jatin Gala
	00:11.805	link	21.0.1-graal	Cool_Mineman
	00:11.878	link	21.0.1-open	karthikeyan97	uses Unsafe
	00:11.934	link	21.0.1-open	arjenvaneerde
	00:12.051	link	21.0.1-open	Dmitry Bufistov
	00:12.102	link	java	Yann Moisan
	00:12.220	link	21.0.1-open	Richard Startin
	00:12.495	link	21.0.1-graal	Samuel Yvon	GraalVM native binary
	00:12.568	link	21.0.1-graal	Vlad
	00:12.800	link	java	Yonatan Graber
	00:13.013	link	21.0.1-graal	Thanh Duong
	00:13.071	link	21.0.1-open	Dr Ian Preston
	00:13.817	link	21.0.1-open	Carlo
	00:14.502	link	21.0.1-graal	eriklumme
	00:14.772	link	21.0.1-open	Kevin McMurtrie
	00:14.867	link	21.0.1-open	Michael Berry
	00:15.662	link	21.0.1-open	Serghei Motpan
	00:17.179	link	21.0.1-open	Jairo Graterón
	00:17.490	link	21.0.1-open	Gergely Kiss
	00:17.255	link	21.0.1-open	tkosachev
	00:17.520	link	21.0.1-open	Farid
	00:17.717	link	21.0.1-open	Oleh Marchenko
	00:17.815	link	21.0.1-open	Hallvard Trætteberg
	00:17.932	link	21.0.1-open	Bartłomiej Pietrzyk
	00:18.251	link	21.0.1-graal	Markus Ebner
	00:18.448	link	21.0.1-open	Moysés Borges Furtado
	00:18.771	link	21.0.1-graal	David Kopec
	00:18.902	link	21.0.1-graal	Maxime
	00:19.357	link	21.0.1-graalce	Roman Schweitzer
	00:20.691	link	21.0.1-graal	Kidlike	GraalVM native binary
	00:21.989	link	21.0.1-open	couragelee
	00:22.457	link	21.0.1-open	Ramzi Ben Yahya
	00:22.471	link	21.0.1-open	Shivam Agarwal
	00:24.986	link	21.0.1-open	kumarsaurav123
	00:26.500	link	21.0.1-open	Bruno Félix
	00:28.381	link	21.0.1-open	Hampus
	00:29.741	link	21.0.1-open	Matteo Vaccari
	00:32.018	link	21.0.1-open	Aurelian Tutuianu
	00:34.388	link	21.0.1-tem	Tobi
	00:35.875	link	21.0.1-open	MahmoudFawzyKhalil
	00:36.180	link	21.0.1-open	Horia Chiorean
	00:38.340	link	21.0.1-open	AbstractKamen
	00:41.982	link	21.0.1-open	Chris Riccomini
	00:42.893	link	21.0.1-open	javamak
	00:46.597	link	21.0.1-open	Maeda-san
	00:58.811	link	21.0.1-open	Ujjwal Bharti
	01:05.094	link	21.0.1-open	Mudit Saxena
	01:06.790	link	21.0.1-open	Karl Heinz Marbaise
	01:06.944	link	21.0.1-open	santanu
	01:07.014	link	21.0.1-open	pedestrianlove
	01:08.811	link	21.0.1-open	Aleš Justin
	01:08.908	link	21.0.1-open	itaske
	01:09.595	link	21.0.1-tem	Antonio Goncalves
	01:09.882	link	21.0.1-open	Prabhu R
	01:14.815	link	21.0.1-open	twohardthings
	01:25.801	link	21.0.1-open	ivanklaric
	01:33.594	link	21.0.1-open	Gaurav Mathur
	01:56.607	link	21.0.1-open	Abhilash
	03:43.521	link	21.0.1-open	김예환 Ye-Hwan Kim (Sam)
	03:59.760	link	21.0.1-open	Samson
	---
	04:49.679	link (Baseline)	21.0.1-open	Gunnar Morling

* These two entries have such a similar runtime (below the error margin I can reliably measure), that they share position #1 in the leaderboar.

Note that I am not super-scientific in the way I'm running the contenders (see Evaluating Results for the details). This is not a high-fidelity micro-benchmark and there can be variations of ~ +-5% between runs. So don't be too hung up on the exact ordering of your entry compared to others in close proximity. The primary purpose of this challenge is to learn something new, have fun along the way, and inspire others to do the same. The leaderboard is only means to an end for achieving this goal. If you observe drastically different results though, please open an issue.

See Entering the Challenge for instructions how to enter the challenge with your own implementation. The Show & Tell features a wide range of 1BRC entries built using other languages, databases, and tools.

Bonus Results

This section lists results from running the fastest N entries with different configurations. As entries have been optimized towards the specific conditions of the original challenge description and set-up (such as size of the key set), challenge entries may perform very differently across different configurations. These bonus results are provided here for informational purposes only. For the 1BRC challenge, only the results in the previous section are of importance.

32 Cores / 64 Threads

For officially evaluating entries into the challenge, each contender is run on eight cores of the evaluation machine (AMD EPYC™ 7502P). Here are the results from running the top 15 entries (as of commit 2c26b511) on all 32 cores / 64 threads (i.e. SMT is enabled) of the machine:

#	Result (m:s.ms)	Implementation	JDK	Submitter	Notes
1	00:00.799	link	21.0.1-graal	Thomas Wuerthinger	GraalVM native binary
2	00:00.933	link	21.0.1-graal	Roy van Rijn	GraalVM native binary
3	00:01.236	link	21.0.1-graal	Artsiom Korzun
	00:01.380	link	21.0.1-open	merykittyunsafe
	00:01.383	link	21.0.1-open	Cliff Click
	00:01.429	link	21.0.1-open	John Ziamos
	00:01.464	link	21.0.1-open	Olivier Bourgain
	00:01.603	link	21.0.1-open	Van Phu DO
	00:01.748	link	21.0.1-graal	Yavuz Tas
	00:01.778	link	21.0.1-open	Quan Anh Mai
	00:01.942	link	21.0.1-graal	Marko Topolnik
	00:01.972	link	21.0.1-graal	Elliot Barlas
	00:02.111	link	21.0.1-graal	Jamal Mulla
	00:02.644	link	21.0.1-graal	Vaidhy Mayilrangam
	00:03.697	link	21.0.1-graal	Jason Nochlin

10K Key Set

The 1BRC challenge data set contains 413 distinct weather stations, whereas the rules allow for 10,000 different station names to occur. Here are the results from running the top 15 entries (as of commit 2c26b511) against 1,000,000,000 measurement values across 10K stations (created via ./create_measurements3.sh 1000000000), using eight cores on the evaluation machine:

#	Result (m:s.ms)	Implementation	JDK	Submitter	Notes
1	00:04.589	link	21.0.1-graal	Artsiom Korzun
2	00:05.296	link	21.0.1-graal	Roy van Rijn	GraalVM native binary
3	00:05.308	link	21.0.1-graal	Thomas Wuerthinger	GraalVM native binary
	00:05.881	link	21.0.1-graal	Marko Topolnik
	00:07.120	link	21.0.1-graal	Jamal Mulla
	00:07.915	link	21.0.1-open	Cliff Click
	00:08.979	link	21.0.1-graal	Yavuz Tas
	00:10.052	link	21.0.1-open	merykittyunsafe
	00:10.134	link	21.0.1-graal	Vaidhy Mayilrangam
	00:10.599	link	21.0.1-graal	Elliot Barlas
	00:12.750	link	21.0.1-open	Quan Anh Mai
	---
	DNF	link	21.0.1-graal	Jason Nochlin	Didn't complete in 60 sec
	DNF	link	21.0.1-open	Van Phu DO	Didn't complete in 60 sec
	DNF	link	21.0.1-open	John Ziamos	Didn't complete in 60 sec
	DNF	link	21.0.1-open	Olivier Bourgain	Failed with java.lang.OutOfMemoryError: Java heap space

Prerequisites

Java 21 must be installed on your system.

Running the Challenge

This repository contains two programs:

• dev.morling.onebrc.CreateMeasurements (invoked via create_measurements.sh): Creates the file measurements.txt in the root directory of this project with a configurable number of random measurement values
• dev.morling.onebrc.CalculateAverage (invoked via calculate_average_baseline.sh): Calculates the average values for the file measurements.txt

Execute the following steps to run the challenge:

1. Build the project using Apache Maven:

   ./mvnw clean verify
   

2. Create the measurements file with 1B rows (just once):

   ./create_measurements.sh 1000000000
   

   This will take a few minutes. Attention: the generated file has a size of approx. 12 GB, so make sure to have enough diskspace.

   If you're running the challenge with a non-Java language, there's a non-authoritative Python script to generate the measurements file at src/main/python/create_measurements.py. The authoritative method for generating the measurements is the Java program dev.morling.onebrc.CreateMeasurements.

3. Calculate the average measurement values:

   ./calculate_average_baseline.sh
   

   The provided naive example implementation uses the Java streams API for processing the file and completes the task in ~2 min on environment used for result evaluation. It serves as the base line for comparing your own implementation.

4. Optimize the heck out of it:

   Adjust the CalculateAverage program to speed it up, in any way you see fit (just sticking to a few rules described below). Options include parallelizing the computation, using the (incubating) Vector API, memory-mapping different sections of the file concurrently, using AppCDS, GraalVM, CRaC, etc. for speeding up the application start-up, choosing and tuning the garbage collector, and much more.

Flamegraph/Profiling

A tip is that if you have jbang installed, you can get a flamegraph of your program by running async-profiler via ap-loader:

jbang --javaagent=ap-loader@jvm-profiling-tools/ap-loader=start,event=cpu,file=profile.html -m dev.morling.onebrc.CalculateAverage_yourname target/average-1.0.0-SNAPSHOT.jar

or directly on the .java file:

jbang --javaagent=ap-loader@jvm-profiling-tools/ap-loader=start,event=cpu,file=profile.html src/main/java/dev/morling/onebrc/CalculateAverage_yourname

When you run this, it will generate a flamegraph in profile.html. You can then open this in a browser and see where your program is spending its time.

Rules and limits

• Any of these Java distributions may be used:
  • Any builds provided by SDKMan
  • Early access builds available on openjdk.net may be used (including EA builds for OpenJDK projects like Valhalla)
  • Builds on builds.shipilev.net If you want to use a build not available via these channels, reach out to discuss whether it can be considered.
• No external library dependencies may be used
• Implementations must be provided as a single source file
• The computation must happen at application runtime, i.e. you cannot process the measurements file at build time (for instance, when using GraalVM) and just bake the result into the binary
• Input value ranges are as follows:
  • Station name: non null UTF-8 string of min length 1 character and max length 100 bytes, containing neither ; nor \n characters. (i.e. this could be 100 one-byte characters, or 50 two-byte characters, etc.)
  • Temperature value: non null double between -99.9 (inclusive) and 99.9 (inclusive), always with one fractional digit
• There is a maximum of 10,000 unique station names
• Line endings in the file are \n characters on all platforms
• Implementations must not rely on specifics of a given data set, e.g. any valid station name as per the constraints above and any data distribution (number of measurements per station) must be supported
• The rounding of output values must be done using the semantics of IEEE 754 rounding-direction "roundTowardPositive"

Entering the Challenge

To submit your own implementation to 1BRC, follow these steps:

• Create a fork of the onebrc GitHub repository.
• Run ./create_fork.sh <your_GH_user> to copy the baseline implementation to your personal files, or do this manually:
  • Create a copy of CalculateAverage_baseline.java, named CalculateAverage_<your_GH_user>.java, e.g. CalculateAverage_doloreswilson.java.
  • Create a copy of calculate_average_baseline.sh, named calculate_average_<your_GH_user>.sh, e.g. calculate_average_doloreswilson.sh.
  • Adjust that script so that it references your implementation class name. If needed, provide any JVM arguments via the JAVA_OPTS variable in that script. Make sure that script does not write anything to standard output other than calculation results.
  • (Optional) OpenJDK 21 is used by default. If a custom JDK build is required, create a copy of prepare_baseline.sh, named prepare_<your_GH_user>.sh, e.g. prepare_doloreswilson.sh. Include the SDKMAN command sdk use java [version] in the your prepare script.
  • (Optional) If you'd like to use native binaries (GraalVM), add all the required build logic to your prepare_<your_GH_user>.sh script.
• Make that implementation fast. Really fast.
• Run the test suite by executing /test.sh <your_GH_user>; if any differences are reported, fix them before submitting your implementation.
• Create a pull request against the upstream repository, clearly stating
  • The name of your implementation class.
  • The execution time of the program on your system and specs of the same (CPU, number of cores, RAM). This is for informative purposes only, the official runtime will be determined as described below.
• I will run the program and determine its performance as described in the next section, and enter the result to the scoreboard.

Note: I reserve the right to not evaluate specific submissions if I feel doubtful about the implementation (I.e. I won't run your Bitcoin miner ;).

If you'd like to discuss any potential ideas for implementing 1BRC with the community, you can use the GitHub Discussions of this repository. Please keep it friendly and civil.

The challenge runs until Jan 31 2024. Any submissions (i.e. pull requests) created after Jan 31 2024 23:59 UTC will not be considered.

Evaluating Results

Results are determined by running the program on a Hetzner AX161 dedicated server (32 core AMD EPYC™ 7502P (Zen2), 128 GB RAM).

Programs are run from a RAM disk (i.o. the IO overhead for loading the file from disk is not relevant), using 8 cores of the machine. Each contender must pass the 1BRC test suite (/test.sh). The hyperfine program is used for measuring execution times of the launch scripts of all entries, i.e. end-to-end times are measured. Each contender is run five times in a row. The slowest and the fastest runs are discarded. The mean value of the remaining three runs is the result for that contender and will be added to the results table above. The exact same measurements.txt file is used for evaluating all contenders. See the script evaluate.sh for the exact implementation of the evaluation steps.

Prize

If you enter this challenge, you may learn something new, get to inspire others, and take pride in seeing your name listed in the scoreboard above. Rumor has it that the winner may receive a unique 1️⃣🐝🏎️ t-shirt, too!

FAQ

Q: Can I use Kotlin or other JVM languages other than Java?
A: No, this challenge is focussed on Java only. Feel free to inofficially share implementations significantly outperforming any listed results, though.

Q: Can I use non-JVM languages and/or tools?
A: No, this challenge is focussed on Java only. Feel free to inofficially share interesting implementations and results though. For instance it would be interesting to see how DuckDB fares with this task.

Q: I've got an implementation—but it's not in Java. Can I share it somewhere?
A: Whilst non-Java solutions cannot be formally submitted to the challenge, you are welcome to share them over in the Show and tell GitHub discussion area.

Q: Can I use JNI?
A: Submissions must be completely implemented in Java, i.e. you cannot write JNI glue code in C/C++. You could use AOT compilation of Java code via GraalVM though, either by AOT-compiling the entire application, or by creating a native library (see here.

Q: What is the encoding of the measurements.txt file?
A: The file is encoded with UTF-8.

Q: Can I make assumptions on the names of the weather stations showing up in the data set?
A: No, while only a fixed set of station names is used by the data set generator, any solution should work with arbitrary UTF-8 station names (for the sake of simplicity, names are guaranteed to contain no ; or \n characters).

Q: Can I copy code from other submissions?
A: Yes, you can. The primary focus of the challenge is about learning something new, rather than "winning". When you do so, please give credit to the relevant source submissions. Please don't re-submit other entries with no or only trivial improvements.

Q: Which operating system is used for evaluation?
A: Fedora 39.

Q: My solution runs in 2 sec on my machine. Am I the fastest 1BRC-er in the world?
A: Probably not :) 1BRC results are reported in wallclock time, thus results of different implementations are only comparable when obtained on the same machine. If for instance an implementation is faster on a 32 core workstation than on the 8 core evaluation instance, this doesn't allow for any conclusions. When sharing 1BRC results, you should also always share the result of running the baseline implementation on the same hardware.

Q: Why 1️⃣🐝🏎️ ?
A: It's the abbreviation of the project name: One Billion Row Challenge.

1BRC on the Web

A list of external resources such as blog posts and videos, discussing 1BRC and specific implementations:

• Cliff Click discussing his 1BRC solution on the Coffee Compiler Club (video)
• 1️⃣🐝🏎️🦆 (1BRC in SQL with DuckDB), by Robin Moffatt (blog post)
• 1 billion rows challenge in PostgreSQL and ClickHouse, by Francesco Tisiot (blog post)
• The One Billion Row Challenge with Snowflake, by Sean Falconer (blog post)
• One billion row challenge using base R, by David Schoch (blog post)
• 1 Billion Row Challenge with Apache Pinot, by Hubert Dulay (blog post)
• One Billion Row Challenge In C, by Danny Van Kooten (blog post)
• One Billion Row Challenge in Racket, by Bogdan Popa (blog post)
• The One Billion Row Challenge - .NET Edition, by Frank A. Krueger (podcast)
• One Billion Row Challenge, by Ragnar Groot Koerkamp (blog post)

Sponsorship

A big thank you to my employer Decodable for funding the evaluation environment and supporting this challenge!

License

This code base is available under the Apache License, version 2.

Code of Conduct

Be excellent to each other! More than winning, the purpose of this challenge is to have fun and learn something new.