Skip to content

Commit f7beeb0

Browse files
committed
Added Thermite December content
1 parent a644a0c commit f7beeb0

File tree

1 file changed

+22
-0
lines changed

1 file changed

+22
-0
lines changed

content/posts/newsletter-017/index.md

+22
Original file line numberDiff line numberDiff line change
@@ -218,6 +218,28 @@ _Discussions:
218218
[rayon]: https://github.com/rayon-rs/rayon
219219
[crates.io]: https://crates.io/
220220

221+
### [Termite SIMD]: Melt Your CPU
222+
223+
Thermite is a WIP SIMD library focused on providing portable SIMD (Single
224+
Instruction, Multiple Data) acceleration of SoA (Structure of Arrays)
225+
algorithms, using consistent-length SIMD vectors for lockstep iteration and
226+
computation. Extensive research and work has gone into minimizing wasted CPU
227+
cycles and making the most out of what your CPU can do.
228+
229+
The goal of Thermite is to provide highly optimized feature-rich backends for
230+
SSE2, SSE4.2, AVX and AVX2, with planned support for AVX512, ARM/Aarch64 NEON,
231+
and WASM SIMD extensions. In addition to that, Thermite will include a highly
232+
optimized vectorized math library with many special math functions and
233+
algorithms, specialized for both single and double precision.
234+
235+
Thermite has been worked on for a little over a month now. The AVX2 backend and
236+
vectorized math library almost fully implemented. Pre-AVX2/WASM/ARM backends are
237+
a work in progress. The latest documentation is available
238+
[here][thermite-documentation].
239+
240+
[thermite-documentation]: https://raygon-renderer.github.io/thermite/
241+
[Termite SIMD]: https://github.com/raygon-renderer/thermite
242+
221243
## Popular Workgroup Issues in Github
222244

223245
<!-- Up to 10 links to interesting issues -->

0 commit comments

Comments
 (0)