-
In faiss some features like IVFPQFastScan rely on at least AVX2 level simd, and we could turn on the flag at the compilation, but we can not decide it dynamicly, that is if we turn on the opt flag on …
-
Right now the [`ImageBuffer`](https://docs.rs/image/0.25.2/image/struct.ImageBuffer.html) type has a `rows()` function to iterate over the rows of the image, but ~~[`GenericImageView`](https://docs.rs…
-
This is to throw some ideas after the immenent release.
1. add -ffast-math compiler option
2. try SIMD-STL.
- It appears that we can enable SIMD (especially AVX/SSE) for many STL (std::*) fu…
-
[EXL (EVA extended library)](https://eprint.iacr.org/2021/1505) supports general EXL vector sizes that are translated into CKKS ciphertext slots by the following:
* User specifies their arbitrary inp…
asraa updated
2 months ago
-
I made [a recipe](https://github.com/procxx/conan-boost-ex-simd) for fork of the removed Boost.SIMD library (I archived it too [here](https://github.com/procxx/boost.simd)) (not planned for conan-cent…
-
```rust
#![feature(portable_simd)]
use std::simd::{u8x16, LaneCount, Simd, SimdElement, SimdInt, SimdPartialOrd, SupportedLaneCount};
fn splat(x: T) -> Simd
where
T: SimdElement,
…
-
This likely caused poor assembly for `Simd::splat` in some circumstances:
https://rust-lang.zulipchat.com/#narrow/stream/257879-project-portable-simd/topic/Very.20bad.20.60u16x8.3A.3Asplat.60.20codeg…
-
Currently NanoRT does not utilize SIMD/AVX.
Also no quantized BVH support.
It'd be better to start to consider optimization and quantization.
Fortunately, recent CPU architecture(AlderLake, Z…
-
Since we use many of our own types, we might be able to use vector and matrix types from System.Numerics instead of OpenTK types which may result in significant speedups with calculations.
We shoul…
-
We should document the various env vars (`{}ARROW_SIMD_LEVEL{`}, `{}ARROW_RUNTIME_SIMD_LEVEL{`}, `{}ARROW_USER_SIMD_LEVEL{`}, others?).
We should also document what the defaults are (and what that …