highway VS DirectXMath

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.

For those unfamiliar, like I was, DXM is DirectXMath.

Alongside MiniEngine, you’ll want to look into the DirectX Toolkit. This is a set of utilities by Microsoft that simplify graphics and game development. It contains libraries like DirectXMesh for parsing and optimizing meshes for DX12, or DirectXMath which handles 3D math operations like the OpenGL library glm. It also has utilities for gamepad input or sprite fonts. You can see a list of the headers here to get an idea of the features. You’ll definitely want to include this in your project if you don’t want to think about a lot of these solved problems (and don’t have to worry about cross-platform support).

Bad news. For SIMD there are not cross-platform intrinsics. Intel intrinsics map directly to SSE/AVX instructions and ARM intrinsics map directly to NEON instructions.

For cross-platform, your best bet is probably https://github.com/VcDevel/std-simd

There's https://eigen.tuxfamily.org/index.php?title=Main_Page But, it's tremendously complicated for anything other than large-scale linear algebra.

And, there's https://github.com/microsoft/DirectXMath But, it has obvious biases :P

I am not in gamedev, but work with 3D graphics, we use DirectX 11, so DirectXMath was a natural choice, it is header only, it supports SIMD instructions (SSE, AVX, NEON etc.), it can even be used on Linux (has no dependence on Windows). It of course just one choice: https://github.com/Microsoft/DirectXMath.

I never really used GLM, but Eigen was substantially slower than DirectXMath https://github.com/microsoft/DirectXMath for these things. Despite the name, 99% of that library is OS agnostic, only a few small pieces (like projection matrix formula) are specific to Direct3D. When enabled with corresponding macros, inline functions from that library normally compile into pretty efficient manually vectorized SSE, AVX or NEON code.

The only major issue, DirectXMath doesn’t support FP64 precision.

If you’re the author, consider comparisons with the industry standards, glm and DirectXMath, which both ensure easy interoperability with the two graphics APIs.

Good article, but note that if the hardware supports the division instruction, will be much faster than the described workarounds.

Personally, I recently did what’s written in 2 cases: FP32 division on ARMv7, and FP64 division on GPUs who don’t support that instruction.

For ARM CPUs, not only they have FRECPE, they also have FRECPS for the iteration step. An example there: https://github.com/microsoft/DirectXMath/blob/jan2021/Inc/Di...

For GPUs, Microsoft classified FP64 division as “extended double shader instruction” and the support is optional. However, GPUs are guaranteed to support FP32 division. The result of FP32 division provides an awesome starting point for Newton-Raphson refinement in FP64 precision.

For graphics DX math is a very good library.

I wonder how does it compare with Microsoft’s implementation, there: https://github.com/microsoft/DirectXMath/blob/jan2021/Inc/Di...

Based on the code your version is probably much faster. It would be interesting to compare precision still, MS uses 17-degree polynomial there.