java VS 1brc

There are Java bindings for TensorFlow, but that's quite low level. I tried to see if I can get some Keras API for Scala, but I'm no expert and haven't had enough time to invest in this, so it's stuck in alpha. Maybe I develop it slow burning over the next year. A bit envious that Kotlin has a Keras-like library.

Hope this is not too late to answer your question. In theory there are no official Java tutorials for Tensorflow 2. The Java implementation is still under development at https://github.com/tensorflow/java

To give a contrasting perspective, I think the Java ecosystem is much better suited for many data science tasks, and has a growing and well-maintained set of libraries for general purpose machine learning. I won't list them all, but TF-Java, DJL et al. have implementations of many modern architectures and there are a number of excellent libraries (CoreNLP, Lucene et al.) for working with text.

Here's the latest java git repo https://github.com/tensorflow/java

This would be the code to beat. Ideally with only 8 cores but any number of cores is also very interesting.

https://github.com/gunnarmorling/1brc/discussions/710

Given that 1-billion-line-file is approximately 13GB, instead of providing a fixed database, the official repository offers a script to generate synthetic data with random readings. Just follow the instructions to create your own database.

Local disk I/O is no longer the bottleneck on modern systems: https://benhoyt.com/writings/io-is-no-longer-the-bottleneck/

In addition, the official 1BRC explicitly evaluated results on a RAM disk to avoid I/O speed entirely: https://github.com/gunnarmorling/1brc?tab=readme-ov-file#eva... "Programs are run from a RAM disk (i.o. the IO overhead for loading the file from disk is not relevant)"

You may have heard of the "The One Billion Row Challenge" (1brc) and in case you don't, go checkout Gunnar Morlings's 1brc repo.

Here’s a thread on results with duckdb, I don’t mean to discourage you taking a shot at all though: https://github.com/gunnarmorling/1brc/discussions/39

If you are in “javaland” look at billion row challenge, you will learn a lot - https://github.com/gunnarmorling/1brc

Even manual vectorization is pain...writing ASM, really?

Rust has unstable portable SIMD and a few third-party crates, C++ has that as well, C# has stable portable SIMD and a very small BLAS-like library on top of it (hell it even exercises PackedSIMD when ran in a browser) and Java is getting stable Panama vectors some time in the future (though the question of codegen quality stands open given planned changes to unsafe API).

Go among these is uniquely disadvantaged. And if that's not enough, you may want to visit 1Brc's challenge discussions and see that Go struggles get anywhere close to 2s mark with both C# and C++ are blazing past it:

https://hotforknowledge.com/2024/01/13/1brc-in-dotnet-among-...

https://github.com/gunnarmorling/1brc/discussions/67

In terms of performance: I realize that this is a somewhat "toy" issue, and it's a sample size of 1, but for the currently ongoing "One Billion Row Challenge"[1] (an ongoing Java performance competition related to parsing and aggregating a 13 GB file), all of the current top-performers are using Unsafe. More specifically, the use of Unsafe appears to have been the change for a few entries that allowed getting below the 3-second barrier in the test.

1. https://github.com/gunnarmorling/1brc