Skip to content

Update text of SVE2 support for xsimd fundable project. #316

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
SylvainCorlay merged 1 commit into from
Apr 2, 2026
Merged

Update text of SVE2 support for xsimd fundable project. #316

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion src/components/fundable/descriptions/SVE2SupportInXsimd.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,10 +1,12 @@
#### Overview
xsimd is a C++ header-only library that abstracts SIMD (vectorization) intrinsics behind a single, generic API. The same code - `xsimd::batch<float>` - compiles to optimal machine code on x86 SSE/AVX, ARM NEON, RISC-V, and WebAssembly, with no runtime overhead. When an intrinsic is missing on a given target, xsimd falls back gracefully rather than failing or leaving the developer to write platform-specific branches. This is why projects like Mozilla Firefox, Apache Arrow, and Pythran have adopted it as their vectorization layer.
xsimd is a C++ header-only library that abstracts SIMD (vectorization) intrinsics behind a single, generic API. The same code - `xsimd::batch<float>` - compiles to optimal machine code on x86 SSE/AVX, ARM NEON, RISC-V, and WebAssembly, with no runtime overhead. When an intrinsic is missing on a given target, xsimd falls back gracefully rather than failing or leaving the developer to write platform-specific branches. This is why projects like Mozilla Firefox, Apache Arrow, Meta Velox, KDE Krita, and Pythran have adopted it as their vectorization layer.

Writing SIMD intrinsics by hand is notoriously painful. Each architecture exposes different types, different naming conventions, and different gotchas — alignment rules, predicate registers, lane-width surprises, compiler bugs. Targeting two architectures means two separate, opaque, hard-to-test code paths. xsimd eliminates that cost entirely.

ARM is now a first-class server architecture. AWS Graviton4 — broadly deployed today across EC2 instance families — ships with SVE2, ARM's latest vector ISA. Google and Microsoft have both announced their own ARM-based server chips (Axion and Cobalt 200 respectively), signalling an industry-wide shift. SVE2 brings a materially richer instruction set than NEON, and xsimd currently cannot exploit it — falling back to SVE and NEON on hardware that is capable of significantly more.

We propose to implement a complete xsimd::sve2 backend — all arithmetic, mathematical, and reduction operations . The work will be fully tested, documented, and merged upstream into the public repository under the existing BSD 3-Clause license.

Funding this development will directly enhance the cutting-edge data processing workflow using Arrow and Velox, paving the way for the future of ARM computing!

##### Are you interested in this project? Either entirely or partially, contact us for more information on how to help us fund it
Loading