highway VS nsimd

The bf16 dot instruction replaces 6 instructions: https://github.com/google/highway/blob/master/hwy/ops/x86_12...

[0] for those interested in Highway.

It's also mentioned in [1], which starts off

> Today we're sharing open source code that can sort arrays of numbers about ten times as fast as the C++ std::sort, and outperforms state of the art architecture-specific algorithms, while being portable across all modern CPU architectures. Below we discuss how we achieved this.

[0] https://github.com/google/highway

[1] https://opensource.googleblog.com/2022/06/Vectorized%20and%2..., which has an associated paper at https://arxiv.org/pdf/2205.05982.pdf.

Thanks so much!

Everyone working on this self-selected into contributing, so I think of it less as my team than ... a team?

Specifically want to call out: Jan Wassenberg (author of https://github.com/google/highway) and I started gemma.cpp as a small project just a few months ago + Phil Culliton, Dan Zheng, and Paul Chang + of course the GDM Gemma team.

C++ users can enjoy Highway [1].

[1] https://github.com/google/highway/

At that point it is better to have some kind of DSL that should not be in the main language, because it would target a much lower level than a typical program. The best effort I've seen in this scene was Google's Highway [1] (not to be confused with HighwayHash) and I even once attempted to recreate it in Rust, but it is still distanced from my ideal.

[1] https://github.com/google/highway

Interesting, thanks for sharing :)

At the time we open-sourced Highway, the standardization process had already started and there were some discussions.

I'm curious why stdlib is the only path you see to default? Compare the activity level of https://github.com/VcDevel/std-simd vs https://github.com/google/highway. As to open-source usage, after years of std::experimental, I see <200 search hits [1], vs >400 for Highway [2], even after excluding several library users.

But that aside, I'm not convinced standardization is the best path for a SIMD library. We and external users extend Highway on a weekly basis as new use cases arise. What if we deferred those changes to 3-monthly meetings, or had to wait for one meeting per WD, CD, (FCD), DIS, (FDIS) stage before it's standardized? Standardization seems more useful for rarely-changing things.

1: https://sourcegraph.com/search?q=context:global+std::experim...

2: https://sourcegraph.com/search?q=context:global+HWY_NAMESPAC...

Thanks for the link. We were previously using GFNI for bit reversal and 8-bit shifts, and I just extended that to our 8-bit BroadcastSignBit (https://github.com/google/highway/pull/1784).

You could study Google's Highway library [1].

[1] https://github.com/google/highway

Runtime feature detection need not be rare nor hard, it's a few dozen lines of boilerplate. You can even write your code just once: see https://github.com/google/highway#examples.

I've decided to switch to something better, and after a few hours of searching, I found this repository: NSIMD https://github.com/agenium-scale/nsimd FastDifferentialCoding https://github.com/lemire/FastDifferentialCoding VS https://github.com/VcDevel/Vc XSIMD https://github.com/xtensor-stack/xsimd

> - Waiting for Cross-Platform standardized SIMD vector datatypes

which language has standardized SIMD vector datatypes ? most languages don't even have any ability to express SIMD while in C++ I can just use Vc (https://github.com/VcDevel/Vc), nsimd (https://github.com/agenium-scale/nsimd) or one of the other ton of alternatives, and have stuff that JustWorksTM on more architectures than most languages even support

- Using nonstandard extensions, libraries or home-baked solutions to run computations in parallel on many cores or on different processors than the CPU

what are the other native languages with a standardized memory model for atomics ? and, what's the problem with using libraries ? it's not like you're going to use C# or Java's built-in threadpools if you are doing any serious work, no ? Do they even have something as easy to use as https://github.com/taskflow/taskflow ?

- Debugging cross-platform code using couts, cerrs and printfs

because people never use console.log in JS or System.println in C# maybe ?

- Forced to use boost for even quite elementary operations on std::strings.

can you point to non-trivial java projects that do not use Apache Commons ? Also, the boost string algorithms are header-only so you will end up with exactly the same binaries that if it was in some std::string_algorithms namespace:

https://gcc.godbolt.org/z/43vKadbde