poulpy-cpu-avx 0.4.4

A crate providing concrete AVX accelerated CPU implementations of poulpy-hal through its open extension points
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75

//! Scalar-vector product (SVP) operations for [`NTT120Avx`](super::NTT120Avx).
//!
//! Implements the `SvpPPol*` and `SvpApply*` OEP traits. SVP multiplies a
//! single scalar polynomial (prepared into NTT domain as `SvpPPol`, q120c format)
//! against a `VecZnxDft` (q120b format), producing a `VecZnxDft` result.
//!
//! - **Prepare**: NTT a `ScalarZnx` into a prepared `SvpPPol` (q120c).
//! - **Apply DFT-to-DFT**: pointwise multiply a prepared scalar against each limb
//!   of a `VecZnxDft`. Available in overwrite, accumulate, and in-place variants.

use poulpy_hal::{
    layouts::{Module, ScalarZnxToRef, SvpPPolToMut, SvpPPolToRef, VecZnxDftToMut, VecZnxDftToRef},
    oep::{SvpApplyDftToDftAddImpl, SvpApplyDftToDftImpl, SvpApplyDftToDftInplaceImpl, SvpPrepareImpl},
    reference::ntt120::svp::{
        ntt120_svp_apply_dft_to_dft, ntt120_svp_apply_dft_to_dft_add, ntt120_svp_apply_dft_to_dft_inplace, ntt120_svp_prepare,
    },
};

use super::NTT120Avx;

unsafe impl SvpPrepareImpl<Self> for NTT120Avx {
    fn svp_prepare_impl<R, A>(module: &Module<Self>, res: &mut R, res_col: usize, a: &A, a_col: usize)
    where
        R: SvpPPolToMut<Self>,
        A: ScalarZnxToRef,
    {
        ntt120_svp_prepare::<R, A, Self>(module, res, res_col, a, a_col);
    }
}

unsafe impl SvpApplyDftToDftImpl<Self> for NTT120Avx {
    fn svp_apply_dft_to_dft_impl<R, A, C>(
        module: &Module<Self>,
        res: &mut R,
        res_col: usize,
        a: &A,
        a_col: usize,
        b: &C,
        b_col: usize,
    ) where
        R: VecZnxDftToMut<Self>,
        A: SvpPPolToRef<Self>,
        C: VecZnxDftToRef<Self>,
    {
        ntt120_svp_apply_dft_to_dft::<R, A, C, Self>(module, res, res_col, a, a_col, b, b_col);
    }
}

unsafe impl SvpApplyDftToDftAddImpl<Self> for NTT120Avx {
    fn svp_apply_dft_to_dft_add_impl<R, A, C>(
        module: &Module<Self>,
        res: &mut R,
        res_col: usize,
        a: &A,
        a_col: usize,
        b: &C,
        b_col: usize,
    ) where
        R: VecZnxDftToMut<Self>,
        A: SvpPPolToRef<Self>,
        C: VecZnxDftToRef<Self>,
    {
        ntt120_svp_apply_dft_to_dft_add::<R, A, C, Self>(module, res, res_col, a, a_col, b, b_col);
    }
}

unsafe impl SvpApplyDftToDftInplaceImpl for NTT120Avx {
    fn svp_apply_dft_to_dft_inplace_impl<R, A>(module: &Module<Self>, res: &mut R, res_col: usize, a: &A, a_col: usize)
    where
        R: VecZnxDftToMut<Self>,
        A: SvpPPolToRef<Self>,
    {
        ntt120_svp_apply_dft_to_dft_inplace::<R, A, Self>(module, res, res_col, a, a_col);
    }
}