Various improvement to memory management and API

[module]: added enum for backend
[VecZnx, VecZnxDft, VecZnxBig, VmpPMat]: added ptr to data
[VecZnxBorrow]: removed
[VecZnxAPI]: removed
This commit is contained in:
Jean-Philippe Bossuat
2025-03-17 12:07:40 +01:00
parent 97a1559bf2
commit 46c577409e
28 changed files with 896 additions and 1064 deletions

View File

@@ -1,6 +1,6 @@
use base2k::{
Encoding, Infos, Module, Sampling, Scalar, SvpPPol, SvpPPolOps, VecZnx, VecZnxApi, VecZnxBig,
VecZnxBigOps, VecZnxDft, VecZnxDftOps, VecZnxOps, FFT64,
alloc_aligned, Encoding, Infos, Module, Sampling, Scalar, SvpPPol, SvpPPolOps, VecZnx,
VecZnxBig, VecZnxBigOps, VecZnxDft, VecZnxDftOps, VecZnxOps, MODULETYPE,
};
use itertools::izip;
use sampling::source::Source;
@@ -11,9 +11,9 @@ fn main() {
let cols: usize = 3;
let msg_cols: usize = 2;
let log_scale: usize = msg_cols * log_base2k - 5;
let module: Module = Module::new::<FFT64>(n);
let module: Module = Module::new(n, MODULETYPE::FFT64);
let mut carry: Vec<u8> = vec![0; module.vec_znx_big_normalize_tmp_bytes()];
let mut carry: Vec<u8> = alloc_aligned(module.vec_znx_big_normalize_tmp_bytes());
let seed: [u8; 32] = [0; 32];
let mut source: Source = Source::new(seed);

View File

@@ -1,13 +1,13 @@
use base2k::{
Encoding, Free, Infos, Module, VecZnx, VecZnxApi, VecZnxBig, VecZnxBigOps, VecZnxDft,
VecZnxDftOps, VecZnxOps, VecZnxVec, VmpPMat, VmpPMatOps, FFT64,
alloc_aligned, Encoding, Infos, Module, VecZnx, VecZnxBig, VecZnxBigOps, VecZnxDft,
VecZnxDftOps, VecZnxOps, VecZnxVec, VmpPMat, VmpPMatOps, MODULETYPE,
};
fn main() {
let log_n: i32 = 5;
let n: usize = 1 << log_n;
let module: Module = Module::new::<FFT64>(n);
let module: Module = Module::new(n, MODULETYPE::FFT64);
let log_base2k: usize = 15;
let cols: usize = 5;
let log_k: usize = log_base2k * cols - 5;
@@ -19,7 +19,7 @@ fn main() {
let tmp_bytes: usize = module.vmp_prepare_tmp_bytes(rows, cols)
| module.vmp_apply_dft_tmp_bytes(cols, cols, rows, cols);
let mut buf: Vec<u8> = vec![0; tmp_bytes];
let mut buf: Vec<u8> = alloc_aligned(tmp_bytes);
let mut a_values: Vec<i64> = vec![i64::default(); n];
a_values[1] = (1 << log_base2k) + 1;
@@ -37,7 +37,7 @@ fn main() {
});
(0..rows).for_each(|i| {
vecznx[i].data[i * n + 1] = 1 as i64;
vecznx[i].raw_mut()[i * n + 1] = 1 as i64;
});
let slices: Vec<&[i64]> = vecznx.dblptr();
@@ -60,8 +60,6 @@ fn main() {
res.print(res.cols(), n);
module.free();
c_dft.free();
vmp_pmat.free();
//println!("{:?}", values_res)
println!("{:?}", values_res)
}

View File

@@ -1,5 +1,5 @@
use crate::ffi::znx::znx_zero_i64_ref;
use crate::{VecZnx, VecZnxBorrow, VecZnxCommon};
use crate::{Infos, VecZnx};
use itertools::izip;
use rug::{Assign, Float};
use std::cmp::min;
@@ -89,42 +89,7 @@ impl Encoding for VecZnx {
}
}
impl Encoding for VecZnxBorrow {
fn encode_vec_i64(&mut self, log_base2k: usize, log_k: usize, data: &[i64], log_max: usize) {
encode_vec_i64(self, log_base2k, log_k, data, log_max)
}
fn decode_vec_i64(&self, log_base2k: usize, log_k: usize, data: &mut [i64]) {
decode_vec_i64(self, log_base2k, log_k, data)
}
fn decode_vec_float(&self, log_base2k: usize, data: &mut [Float]) {
decode_vec_float(self, log_base2k, data)
}
fn encode_coeff_i64(
&mut self,
log_base2k: usize,
log_k: usize,
i: usize,
value: i64,
log_max: usize,
) {
encode_coeff_i64(self, log_base2k, log_k, i, value, log_max)
}
fn decode_coeff_i64(&self, log_base2k: usize, log_k: usize, i: usize) -> i64 {
decode_coeff_i64(self, log_base2k, log_k, i)
}
}
fn encode_vec_i64<T: VecZnxCommon>(
a: &mut T,
log_base2k: usize,
log_k: usize,
data: &[i64],
log_max: usize,
) {
fn encode_vec_i64(a: &mut VecZnx, log_base2k: usize, log_k: usize, data: &[i64], log_max: usize) {
let cols: usize = (log_k + log_base2k - 1) / log_base2k;
debug_assert!(
@@ -170,7 +135,7 @@ fn encode_vec_i64<T: VecZnxCommon>(
}
}
fn decode_vec_i64<T: VecZnxCommon>(a: &T, log_base2k: usize, log_k: usize, data: &mut [i64]) {
fn decode_vec_i64(a: &VecZnx, log_base2k: usize, log_k: usize, data: &mut [i64]) {
let cols: usize = (log_k + log_base2k - 1) / log_base2k;
debug_assert!(
data.len() >= a.n(),
@@ -194,7 +159,7 @@ fn decode_vec_i64<T: VecZnxCommon>(a: &T, log_base2k: usize, log_k: usize, data:
})
}
fn decode_vec_float<T: VecZnxCommon>(a: &T, log_base2k: usize, data: &mut [Float]) {
fn decode_vec_float(a: &VecZnx, log_base2k: usize, data: &mut [Float]) {
let cols: usize = a.cols();
debug_assert!(
data.len() >= a.n(),
@@ -224,8 +189,8 @@ fn decode_vec_float<T: VecZnxCommon>(a: &T, log_base2k: usize, data: &mut [Float
});
}
fn encode_coeff_i64<T: VecZnxCommon>(
a: &mut T,
fn encode_coeff_i64(
a: &mut VecZnx,
log_base2k: usize,
log_k: usize,
i: usize,
@@ -247,7 +212,7 @@ fn encode_coeff_i64<T: VecZnxCommon>(
// values on the last limb.
// Else we decompose values base2k.
if log_max + log_k_rem < 63 || log_k_rem == log_base2k {
a.at_mut(cols-1)[i] = value;
a.at_mut(cols - 1)[i] = value;
} else {
let mask: i64 = (1 << log_base2k) - 1;
let steps: usize = min(cols, (log_max + log_base2k - 1) / log_base2k);
@@ -268,7 +233,7 @@ fn encode_coeff_i64<T: VecZnxCommon>(
}
}
fn decode_coeff_i64<T: VecZnxCommon>(a: &T, log_base2k: usize, log_k: usize, i: usize) -> i64 {
fn decode_coeff_i64(a: &VecZnx, log_base2k: usize, log_k: usize, i: usize) -> i64 {
let cols: usize = (log_k + log_base2k - 1) / log_base2k;
debug_assert!(i < a.n());
let data: &[i64] = a.raw();

View File

@@ -1,43 +0,0 @@
use crate::ffi::svp;
use crate::ffi::vec_znx_big;
use crate::ffi::vec_znx_dft;
use crate::ffi::vmp;
use crate::{SvpPPol, VecZnxBig, VecZnxDft, VmpPMat};
/// This trait should be implemented by structs that point to
/// memory allocated through C.
pub trait Free {
// Frees the memory and self destructs.
fn free(self);
}
impl Free for VmpPMat {
/// Frees the C allocated memory of the [VmpPMat] and self destructs the struct.
fn free(self) {
unsafe { vmp::delete_vmp_pmat(self.data) };
drop(self);
}
}
impl Free for VecZnxDft {
fn free(self) {
unsafe { vec_znx_dft::delete_vec_znx_dft(self.0) };
drop(self);
}
}
impl Free for VecZnxBig {
fn free(self) {
unsafe {
vec_znx_big::delete_vec_znx_big(self.0);
}
drop(self);
}
}
impl Free for SvpPPol {
fn free(self) {
unsafe { svp::delete_svp_ppol(self.0) };
let _ = drop(self);
}
}

View File

@@ -1,5 +1,3 @@
use crate::{VecZnx, VecZnxBorrow, VmpPMat};
pub trait Infos {
/// Returns the ring degree of the receiver.
fn n(&self) -> usize;
@@ -14,71 +12,3 @@ pub trait Infos {
/// Returns the number of rows of the receiver.
fn rows(&self) -> usize;
}
impl Infos for VecZnx {
/// Returns the base 2 logarithm of the [VecZnx] degree.
fn log_n(&self) -> usize {
(usize::BITS - (self.n - 1).leading_zeros()) as _
}
/// Returns the [VecZnx] degree.
fn n(&self) -> usize {
self.n
}
/// Returns the number of cols of the [VecZnx].
fn cols(&self) -> usize {
self.data.len() / self.n
}
/// Returns the number of rows of the [VecZnx].
fn rows(&self) -> usize {
1
}
}
impl Infos for VecZnxBorrow {
/// Returns the base 2 logarithm of the [VecZnx] degree.
fn log_n(&self) -> usize {
(usize::BITS - (self.n - 1).leading_zeros()) as _
}
/// Returns the [VecZnx] degree.
fn n(&self) -> usize {
self.n
}
/// Returns the number of cols of the [VecZnx].
fn cols(&self) -> usize {
self.cols
}
/// Returns the number of rows of the [VecZnx].
fn rows(&self) -> usize {
1
}
}
impl Infos for VmpPMat {
/// Returns the ring dimension of the [VmpPMat].
fn n(&self) -> usize {
self.n
}
fn log_n(&self) -> usize {
(usize::BITS - (self.n() - 1).leading_zeros()) as _
}
/// Returns the number of rows (i.e. of [VecZnxDft]) of the [VmpPMat]
fn rows(&self) -> usize {
self.rows
}
/// Returns the number of cols of the [VmpPMat].
/// The number of cols refers to the number of cols
/// of each [VecZnxDft].
/// This method is equivalent to [Self::cols].
fn cols(&self) -> usize {
self.cols
}
}

View File

@@ -8,7 +8,6 @@ pub mod encoding;
)]
// Other modules and exports
pub mod ffi;
pub mod free;
pub mod infos;
pub mod module;
pub mod sampling;
@@ -20,7 +19,6 @@ pub mod vec_znx_dft;
pub mod vmp;
pub use encoding::*;
pub use free::*;
pub use infos::*;
pub use module::*;
pub use sampling::*;
@@ -124,11 +122,3 @@ pub fn alloc_aligned_custom<T>(size: usize, align: usize) -> Vec<T> {
pub fn alloc_aligned<T>(size: usize) -> Vec<T> {
alloc_aligned_custom::<T>(size, DEFAULTALIGN)
}
fn alias_mut_slice_to_vec<T>(slice: &[T]) -> Vec<T> {
unsafe {
let ptr: *mut T = slice.as_ptr() as *mut T;
let len: usize = slice.len();
Vec::from_raw_parts(ptr, len, len)
}
}

View File

@@ -1,26 +1,46 @@
use crate::ffi::module::{delete_module_info, module_info_t, new_module_info, MODULE};
use crate::{Free, GALOISGENERATOR};
use crate::GALOISGENERATOR;
pub type MODULETYPE = u8;
pub const FFT64: u8 = 0;
pub const NTT120: u8 = 1;
#[derive(Copy, Clone)]
#[repr(u8)]
pub enum MODULETYPE {
FFT64,
NTT120,
}
pub struct Module(pub *mut MODULE, pub usize);
pub struct Module {
pub ptr: *mut MODULE,
pub n: usize,
pub backend: MODULETYPE,
}
impl Module {
// Instantiates a new module.
pub fn new<const MODULETYPE: MODULETYPE>(n: usize) -> Self {
pub fn new(n: usize, module_type: MODULETYPE) -> Self {
unsafe {
let m: *mut module_info_t = new_module_info(n as u64, MODULETYPE as u32);
let module_type_u32: u32;
match module_type {
MODULETYPE::FFT64 => module_type_u32 = 0,
MODULETYPE::NTT120 => module_type_u32 = 1,
}
let m: *mut module_info_t = new_module_info(n as u64, module_type_u32);
if m.is_null() {
panic!("Failed to create module.");
}
Self(m, n)
Self {
ptr: m,
n: n,
backend: module_type,
}
}
}
pub fn backend(&self) -> MODULETYPE {
self.backend
}
pub fn n(&self) -> usize {
self.1
self.n
}
pub fn log_n(&self) -> usize {
@@ -53,11 +73,9 @@ impl Module {
(gal_el as i64) * gen.signum()
}
}
impl Free for Module {
fn free(self) {
unsafe { delete_module_info(self.0) }
pub fn free(self) {
unsafe { delete_module_info(self.ptr) }
drop(self);
}
}

View File

@@ -1,16 +1,16 @@
use crate::{Infos, Module, VecZnxApi};
use crate::{Infos, Module, VecZnx};
use rand_distr::{Distribution, Normal};
use sampling::source::Source;
pub trait Sampling<T: VecZnxApi + Infos> {
pub trait Sampling {
/// Fills the first `cols` cols with uniform values in \[-2^{log_base2k-1}, 2^{log_base2k-1}\]
fn fill_uniform(&self, log_base2k: usize, a: &mut T, cols: usize, source: &mut Source);
fn fill_uniform(&self, log_base2k: usize, a: &mut VecZnx, cols: usize, source: &mut Source);
/// Adds vector sampled according to the provided distribution, scaled by 2^{-log_k} and bounded to \[-bound, bound\].
fn add_dist_f64<D: Distribution<f64>>(
&self,
log_base2k: usize,
a: &mut T,
a: &mut VecZnx,
log_k: usize,
source: &mut Source,
dist: D,
@@ -21,7 +21,7 @@ pub trait Sampling<T: VecZnxApi + Infos> {
fn add_normal(
&self,
log_base2k: usize,
a: &mut T,
a: &mut VecZnx,
log_k: usize,
source: &mut Source,
sigma: f64,
@@ -29,8 +29,8 @@ pub trait Sampling<T: VecZnxApi + Infos> {
);
}
impl<T: VecZnxApi + Infos> Sampling<T> for Module {
fn fill_uniform(&self, log_base2k: usize, a: &mut T, cols: usize, source: &mut Source) {
impl Sampling for Module {
fn fill_uniform(&self, log_base2k: usize, a: &mut VecZnx, cols: usize, source: &mut Source) {
let base2k: u64 = 1 << log_base2k;
let mask: u64 = base2k - 1;
let base2k_half: i64 = (base2k >> 1) as i64;
@@ -43,7 +43,7 @@ impl<T: VecZnxApi + Infos> Sampling<T> for Module {
fn add_dist_f64<D: Distribution<f64>>(
&self,
log_base2k: usize,
a: &mut T,
a: &mut VecZnx,
log_k: usize,
source: &mut Source,
dist: D,
@@ -79,7 +79,7 @@ impl<T: VecZnxApi + Infos> Sampling<T> for Module {
fn add_normal(
&self,
log_base2k: usize,
a: &mut T,
a: &mut VecZnx,
log_k: usize,
source: &mut Source,
sigma: f64,

View File

@@ -1,13 +1,18 @@
use crate::ffi::svp;
use crate::{alias_mut_slice_to_vec, assert_alignement, Module, VecZnxApi, VecZnxDft};
use crate::ffi::vec_znx_dft::vec_znx_dft_t;
use crate::{assert_alignement, Module, VecZnx, VecZnxDft};
use crate::{alloc_aligned, cast, Infos};
use crate::{alloc_aligned, cast_mut, Infos};
use rand::seq::SliceRandom;
use rand_core::RngCore;
use rand_distr::{Distribution, WeightedIndex};
use sampling::source::Source;
pub struct Scalar(pub Vec<i64>);
pub struct Scalar {
pub n: usize,
pub data: Vec<i64>,
pub ptr: *mut i64,
}
impl Module {
pub fn new_scalar(&self) -> Scalar {
@@ -17,52 +22,70 @@ impl Module {
impl Scalar {
pub fn new(n: usize) -> Self {
Self(alloc_aligned::<i64>(n))
let mut data: Vec<i64> = alloc_aligned::<i64>(n);
let ptr: *mut i64 = data.as_mut_ptr();
Self {
n: n,
data: data,
ptr: ptr,
}
}
pub fn n(&self) -> usize {
self.0.len()
self.n
}
pub fn buffer_size(n: usize) -> usize {
n
}
pub fn from_buffer(&mut self, n: usize, buf: &mut [u8]) {
pub fn from_buffer(&mut self, n: usize, bytes: &mut [u8]) -> Self {
let size: usize = Self::buffer_size(n);
debug_assert!(
buf.len() >= size,
"invalid buffer: buf.len()={} < self.buffer_size(n={})={}",
buf.len(),
bytes.len() == size,
"invalid buffer: bytes.len()={} < self.buffer_size(n={})={}",
bytes.len(),
n,
size
);
#[cfg(debug_assertions)]
{
assert_alignement(buf.as_ptr())
assert_alignement(bytes.as_ptr())
}
unsafe {
let bytes_i64: &mut [i64] = cast_mut::<u8, i64>(bytes);
let ptr: *mut i64 = bytes_i64.as_mut_ptr();
Self {
n: n,
data: Vec::from_raw_parts(bytes_i64.as_mut_ptr(), bytes.len(), bytes.len()),
ptr: ptr,
}
}
self.0 = alias_mut_slice_to_vec(cast::<u8, i64>(&buf[..size]))
}
pub fn as_ptr(&self) -> *const i64 {
self.0.as_ptr()
self.ptr
}
pub fn raw(&self) -> &[i64] {
unsafe { std::slice::from_raw_parts_mut(self.ptr, self.n) }
}
pub fn fill_ternary_prob(&mut self, prob: f64, source: &mut Source) {
let choices: [i64; 3] = [-1, 0, 1];
let weights: [f64; 3] = [prob / 2.0, 1.0 - prob, prob / 2.0];
let dist: WeightedIndex<f64> = WeightedIndex::new(&weights).unwrap();
self.0
self.data
.iter_mut()
.for_each(|x: &mut i64| *x = choices[dist.sample(source)]);
}
pub fn fill_ternary_hw(&mut self, hw: usize, source: &mut Source) {
assert!(hw <= self.n());
self.0[..hw]
self.data[..hw]
.iter_mut()
.for_each(|x: &mut i64| *x = (((source.next_u32() & 1) as i64) << 1) - 1);
self.0.shuffle(source);
self.data.shuffle(source);
}
}
@@ -105,35 +128,23 @@ pub trait SvpPPolOps {
/// Applies the [SvpPPol] x [VecZnxDft] product, where each limb of
/// the [VecZnxDft] is multiplied with [SvpPPol].
fn svp_apply_dft<T: VecZnxApi + Infos>(
&self,
c: &mut VecZnxDft,
a: &SvpPPol,
b: &T,
b_cols: usize,
);
fn svp_apply_dft(&self, c: &mut VecZnxDft, a: &SvpPPol, b: &VecZnx, b_cols: usize);
}
impl SvpPPolOps for Module {
fn new_svp_ppol(&self) -> SvpPPol {
unsafe { SvpPPol(svp::new_svp_ppol(self.0), self.n()) }
unsafe { SvpPPol(svp::new_svp_ppol(self.ptr), self.n()) }
}
fn bytes_of_svp_ppol(&self) -> usize {
unsafe { svp::bytes_of_svp_ppol(self.0) as usize }
unsafe { svp::bytes_of_svp_ppol(self.ptr) as usize }
}
fn svp_prepare(&self, svp_ppol: &mut SvpPPol, a: &Scalar) {
unsafe { svp::svp_prepare(self.0, svp_ppol.0, a.as_ptr()) }
unsafe { svp::svp_prepare(self.ptr, svp_ppol.0, a.as_ptr()) }
}
fn svp_apply_dft<T: VecZnxApi + Infos>(
&self,
c: &mut VecZnxDft,
a: &SvpPPol,
b: &T,
b_cols: usize,
) {
fn svp_apply_dft(&self, c: &mut VecZnxDft, a: &SvpPPol, b: &VecZnx, b_cols: usize) {
debug_assert!(
c.cols() >= b_cols,
"invalid c_vector: c_vector.cols()={} < b.cols()={}",
@@ -142,8 +153,8 @@ impl SvpPPolOps for Module {
);
unsafe {
svp::svp_apply_dft(
self.0,
c.0,
self.ptr,
c.ptr as *mut vec_znx_dft_t,
b_cols as u64,
a.0,
b.as_ptr(),

View File

@@ -1,18 +1,35 @@
use crate::cast_mut;
use crate::ffi::vec_znx;
use crate::ffi::znx;
use crate::ffi::znx::znx_zero_i64_ref;
use crate::{alias_mut_slice_to_vec, alloc_aligned, assert_alignement};
use crate::{alloc_aligned, assert_alignement};
use crate::{Infos, Module};
use itertools::izip;
use std::cmp::min;
/// [VecZnx] represents a vector of small norm polynomials of Zn\[X\] with [i64] coefficients.
/// A [VecZnx] is composed of multiple Zn\[X\] polynomials stored in a single contiguous array
/// in the memory.
#[derive(Clone)]
pub struct VecZnx {
/// Polynomial degree.
n: usize,
/// Number of columns.
cols: usize,
/// Polynomial coefficients, as a contiguous array. Each col is equally spaced by n.
data: Vec<i64>,
/// Pointer to data (data can be enpty if [VecZnx] borrows space instead of owning it).
ptr: *mut i64,
}
pub trait VecZnxVec {
fn dblptr(&self) -> Vec<&[i64]>;
fn dblptr_mut(&mut self) -> Vec<&mut [i64]>;
}
impl<T: VecZnxCommon> VecZnxVec for Vec<T> {
impl VecZnxVec for Vec<VecZnx> {
fn dblptr(&self) -> Vec<&[i64]> {
self.iter().map(|v| v.raw()).collect()
}
@@ -22,328 +39,141 @@ impl<T: VecZnxCommon> VecZnxVec for Vec<T> {
}
}
pub trait VecZnxApi: AsRef<Self> + AsMut<Self> {
type Owned: VecZnxCommon;
fn from_bytes(n: usize, cols: usize, bytes: &mut [u8]) -> Self::Owned;
/// Returns the minimum size of the [u8] array required to assign a
/// new backend array.
fn bytes_of(n: usize, cols: usize) -> usize;
/// Copy the data of a onto self.
fn copy_from<A: VecZnxCommon, B: VecZnxCommon>(&mut self, a: &A)
where
Self: AsMut<B>;
/// Returns the backing array.
fn raw(&self) -> &[i64];
/// Returns the mutable backing array.
fn raw_mut(&mut self) -> &mut [i64];
/// Returns a non-mutable pointer to the backing array.
fn as_ptr(&self) -> *const i64;
/// Returns a mutable pointer to the backing array.
fn as_mut_ptr(&mut self) -> *mut i64;
/// Returns a non-mutable reference to the i-th cols.
fn at(&self, i: usize) -> &[i64];
/// Returns a mutable reference to the i-th cols .
fn at_mut(&mut self, i: usize) -> &mut [i64];
/// Returns a non-mutable pointer to the i-th cols.
fn at_ptr(&self, i: usize) -> *const i64;
/// Returns a mutable pointer to the i-th cols.
fn at_mut_ptr(&mut self, i: usize) -> *mut i64;
/// Zeroes the backing array.
fn zero(&mut self);
/// Normalization: propagates carry and ensures each coefficients
/// falls into the range [-2^{K-1}, 2^{K-1}].
fn normalize(&mut self, log_base2k: usize, carry: &mut [u8]);
/// Right shifts the coefficients by k bits.
///
/// # Arguments
///
/// * `log_base2k`: the base two logarithm of the coefficients decomposition.
/// * `k`: the shift amount.
/// * `carry`: scratch space of size at least equal to self.n() * self.cols() << 3.
///
/// # Panics
///
/// The method will panic if carry.len() < self.n() * self.cols() << 3.
fn rsh(&mut self, log_base2k: usize, k: usize, carry: &mut [u8]);
/// If self.n() > a.n(): Extracts X^{i*self.n()/a.n()} -> X^{i}.
/// If self.n() < a.n(): Extracts X^{i} -> X^{i*a.n()/self.n()}.
///
/// # Arguments
///
/// * `a`: the receiver polynomial in which the extracted coefficients are stored.
fn switch_degree<A: VecZnxCommon, B: VecZnxCommon>(&self, a: &mut A)
where
Self: AsRef<B>;
fn print(&self, cols: usize, n: usize);
}
pub fn bytes_of_vec_znx(n: usize, cols: usize) -> usize {
n * cols * 8
}
pub struct VecZnxBorrow {
pub n: usize,
pub cols: usize,
pub data: *mut i64,
}
impl AsMut<VecZnxBorrow> for VecZnxBorrow {
fn as_mut(&mut self) -> &mut VecZnxBorrow {
self
}
}
impl AsRef<VecZnxBorrow> for VecZnxBorrow {
fn as_ref(&self) -> &VecZnxBorrow {
self
}
}
impl VecZnxCommon for VecZnxBorrow {}
impl VecZnxApi for VecZnxBorrow {
type Owned = VecZnxBorrow;
/// Returns a new struct implementing [VecZnxBorrow] with the provided data as backing array.
///
/// The struct will *NOT* take ownership of buf[..[VecZnx::bytes_of]]
///
/// User must ensure that data is properly alligned and that
/// the size of data is at least equal to [VecZnx::bytes_of].
fn from_bytes(n: usize, cols: usize, bytes: &mut [u8]) -> Self::Owned {
let size = Self::bytes_of(n, cols);
debug_assert!(
bytes.len() >= size,
"invalid buffer: buf.len()={} < self.buffer_size(n={}, cols={})={}",
bytes.len(),
n,
cols,
size
);
#[cfg(debug_assertions)]
{
assert_alignement(bytes.as_ptr())
}
VecZnxBorrow {
n: n,
cols: cols,
data: cast_mut(&mut bytes[..size]).as_mut_ptr(),
}
}
fn bytes_of(n: usize, cols: usize) -> usize {
bytes_of_vec_znx(n, cols)
}
fn copy_from<A: VecZnxCommon, B: VecZnxCommon>(&mut self, a: &A)
where
Self: AsMut<B>,
{
copy_vec_znx_from::<A, B>(self.as_mut(), a);
}
fn as_ptr(&self) -> *const i64 {
self.data
}
fn as_mut_ptr(&mut self) -> *mut i64 {
self.data
}
fn raw(&self) -> &[i64] {
unsafe { std::slice::from_raw_parts(self.data, self.n * self.cols) }
}
fn raw_mut(&mut self) -> &mut [i64] {
unsafe { std::slice::from_raw_parts_mut(self.data, self.n * self.cols) }
}
fn at(&self, i: usize) -> &[i64] {
let n: usize = self.n();
&self.raw()[n * i..n * (i + 1)]
}
fn at_mut(&mut self, i: usize) -> &mut [i64] {
let n: usize = self.n();
&mut self.raw_mut()[n * i..n * (i + 1)]
}
fn at_ptr(&self, i: usize) -> *const i64 {
self.data.wrapping_add(self.n * i)
}
fn at_mut_ptr(&mut self, i: usize) -> *mut i64 {
self.data.wrapping_add(self.n * i)
}
fn zero(&mut self) {
unsafe {
znx_zero_i64_ref((self.n * self.cols) as u64, self.data);
}
}
fn normalize(&mut self, log_base2k: usize, carry: &mut [u8]) {
normalize(log_base2k, self, carry)
}
fn rsh(&mut self, log_base2k: usize, k: usize, carry: &mut [u8]) {
rsh(log_base2k, self, k, carry)
}
fn switch_degree<A: VecZnxCommon, B: VecZnxCommon>(&self, a: &mut A)
where
Self: AsRef<B>,
{
switch_degree(a, self.as_ref());
}
fn print(&self, cols: usize, n: usize) {
(0..cols).for_each(|i| println!("{}: {:?}", i, &self.at(i)[..n]))
}
}
impl VecZnxCommon for VecZnx {}
impl VecZnxApi for VecZnx {
type Owned = VecZnx;
impl VecZnx {
/// Returns a new struct implementing [VecZnx] with the provided data as backing array.
///
/// The struct will take ownership of buf[..[VecZnx::bytes_of]]
///
/// User must ensure that data is properly alligned and that
/// the size of data is at least equal to [VecZnx::bytes_of].
fn from_bytes(n: usize, cols: usize, bytes: &mut [u8]) -> Self::Owned {
let size = Self::bytes_of(n, cols);
debug_assert!(
bytes.len() >= size,
"invalid bytes: bytes.len()={} < self.bytes_of(n={}, cols={})={}",
bytes.len(),
n,
cols,
size
);
pub fn from_bytes(n: usize, cols: usize, bytes: &mut [u8]) -> Self {
#[cfg(debug_assertions)]
{
assert_alignement(bytes.as_ptr())
assert_eq!(bytes.len(), Self::bytes_of(n, cols));
assert_alignement(bytes.as_ptr());
}
unsafe {
let bytes_i64: &mut [i64] = cast_mut::<u8, i64>(bytes);
let ptr: *mut i64 = bytes_i64.as_mut_ptr();
VecZnx {
n: n,
cols: cols,
data: Vec::from_raw_parts(bytes_i64.as_mut_ptr(), bytes.len(), bytes.len()),
ptr: ptr,
}
}
}
pub fn from_bytes_borrow(n: usize, cols: usize, bytes: &mut [u8]) -> Self {
#[cfg(debug_assertions)]
{
assert!(bytes.len() >= Self::bytes_of(n, cols));
assert_alignement(bytes.as_ptr());
}
VecZnx {
n: n,
data: alias_mut_slice_to_vec(cast_mut(&mut bytes[..size])),
cols: cols,
data: Vec::new(),
ptr: bytes.as_mut_ptr() as *mut i64,
}
}
fn bytes_of(n: usize, cols: usize) -> usize {
pub fn bytes_of(n: usize, cols: usize) -> usize {
bytes_of_vec_znx(n, cols)
}
fn copy_from<A: VecZnxCommon, B: VecZnxCommon>(&mut self, a: &A)
where
Self: AsMut<B>,
{
copy_vec_znx_from(self.as_mut(), a);
pub fn copy_from(&mut self, a: &VecZnx) {
copy_vec_znx_from(self, a);
}
fn raw(&self) -> &[i64] {
&self.data
pub fn raw(&self) -> &[i64] {
unsafe { std::slice::from_raw_parts(self.ptr, self.n * self.cols) }
}
fn raw_mut(&mut self) -> &mut [i64] {
&mut self.data
pub fn borrowing(&self) -> bool {
self.data.len() == 0
}
fn as_ptr(&self) -> *const i64 {
self.data.as_ptr()
pub fn raw_mut(&mut self) -> &mut [i64] {
unsafe { std::slice::from_raw_parts_mut(self.ptr, self.n * self.cols) }
}
fn as_mut_ptr(&mut self) -> *mut i64 {
self.data.as_mut_ptr()
pub fn as_ptr(&self) -> *const i64 {
self.ptr
}
fn at(&self, i: usize) -> &[i64] {
pub fn as_mut_ptr(&mut self) -> *mut i64 {
self.ptr
}
pub fn at(&self, i: usize) -> &[i64] {
let n: usize = self.n();
&self.raw()[n * i..n * (i + 1)]
}
fn at_mut(&mut self, i: usize) -> &mut [i64] {
pub fn at_mut(&mut self, i: usize) -> &mut [i64] {
let n: usize = self.n();
&mut self.raw_mut()[n * i..n * (i + 1)]
}
fn at_ptr(&self, i: usize) -> *const i64 {
&self.data[i * self.n] as *const i64
pub fn at_ptr(&self, i: usize) -> *const i64 {
self.ptr.wrapping_add(i * self.n)
}
fn at_mut_ptr(&mut self, i: usize) -> *mut i64 {
&mut self.data[i * self.n] as *mut i64
pub fn at_mut_ptr(&mut self, i: usize) -> *mut i64 {
self.ptr.wrapping_add(i * self.n)
}
fn zero(&mut self) {
unsafe { znx::znx_zero_i64_ref(self.data.len() as u64, self.data.as_mut_ptr()) }
pub fn zero(&mut self) {
unsafe { znx::znx_zero_i64_ref((self.n * self.cols) as u64, self.ptr) }
}
fn normalize(&mut self, log_base2k: usize, carry: &mut [u8]) {
pub fn normalize(&mut self, log_base2k: usize, carry: &mut [u8]) {
normalize(log_base2k, self, carry)
}
fn rsh(&mut self, log_base2k: usize, k: usize, carry: &mut [u8]) {
pub fn rsh(&mut self, log_base2k: usize, k: usize, carry: &mut [u8]) {
rsh(log_base2k, self, k, carry)
}
fn switch_degree<A: VecZnxCommon, B: VecZnxCommon>(&self, a: &mut A)
where
Self: AsRef<B>,
{
switch_degree(a, self.as_ref())
pub fn switch_degree(&self, a: &mut VecZnx) {
switch_degree(a, self)
}
fn print(&self, cols: usize, n: usize) {
pub fn print(&self, cols: usize, n: usize) {
(0..cols).for_each(|i| println!("{}: {:?}", i, &self.at(i)[..n]))
}
}
/// [VecZnx] represents a vector of small norm polynomials of Zn\[X\] with [i64] coefficients.
/// A [VecZnx] is composed of multiple Zn\[X\] polynomials stored in a single contiguous array
/// in the memory.
#[derive(Clone)]
pub struct VecZnx {
/// Polynomial degree.
pub n: usize,
/// Polynomial coefficients, as a contiguous array. Each col is equally spaced by n.
pub data: Vec<i64>,
}
impl AsMut<VecZnx> for VecZnx {
fn as_mut(&mut self) -> &mut VecZnx {
self
impl Infos for VecZnx {
/// Returns the base 2 logarithm of the [VecZnx] degree.
fn log_n(&self) -> usize {
(usize::BITS - (self.n - 1).leading_zeros()) as _
}
}
impl AsRef<VecZnx> for VecZnx {
fn as_ref(&self) -> &VecZnx {
self
/// Returns the [VecZnx] degree.
fn n(&self) -> usize {
self.n
}
/// Returns the number of cols of the [VecZnx].
fn cols(&self) -> usize {
self.cols
}
/// Returns the number of rows of the [VecZnx].
fn rows(&self) -> usize {
1
}
}
/// Copies the coefficients of `a` on the receiver.
/// Copy is done with the minimum size matching both backing arrays.
pub fn copy_vec_znx_from<A: VecZnxCommon, B: VecZnxCommon>(b: &mut B, a: &A) {
pub fn copy_vec_znx_from(b: &mut VecZnx, a: &VecZnx) {
let data_a: &[i64] = a.raw();
let data_b: &mut [i64] = b.raw_mut();
let size = min(data_b.len(), data_a.len());
@@ -353,9 +183,13 @@ pub fn copy_vec_znx_from<A: VecZnxCommon, B: VecZnxCommon>(b: &mut B, a: &A) {
impl VecZnx {
/// Allocates a new [VecZnx] composed of #cols polynomials of Z\[X\].
pub fn new(n: usize, cols: usize) -> Self {
let mut data: Vec<i64> = alloc_aligned::<i64>(n * cols);
let ptr: *mut i64 = data.as_mut_ptr();
Self {
n: n,
data: alloc_aligned::<i64>(n * cols),
cols: cols,
data: data,
ptr: ptr,
}
}
@@ -370,8 +204,12 @@ impl VecZnx {
return;
}
self.data
.truncate((self.cols() - k / log_base2k) * self.n());
if !self.borrowing() {
self.data
.truncate((self.cols() - k / log_base2k) * self.n());
}
self.cols -= k / log_base2k;
let k_rem: usize = k % log_base2k;
@@ -384,7 +222,7 @@ impl VecZnx {
}
}
pub fn switch_degree<A: VecZnxCommon, B: VecZnxCommon>(b: &mut B, a: &A) {
pub fn switch_degree(b: &mut VecZnx, a: &VecZnx) {
let (n_in, n_out) = (a.n(), b.n());
let (gap_in, gap_out): (usize, usize);
@@ -406,7 +244,7 @@ pub fn switch_degree<A: VecZnxCommon, B: VecZnxCommon>(b: &mut B, a: &A) {
});
}
fn normalize<T: VecZnxCommon>(log_base2k: usize, a: &mut T, tmp_bytes: &mut [u8]) {
fn normalize(log_base2k: usize, a: &mut VecZnx, tmp_bytes: &mut [u8]) {
let n: usize = a.n();
debug_assert!(
@@ -437,7 +275,7 @@ fn normalize<T: VecZnxCommon>(log_base2k: usize, a: &mut T, tmp_bytes: &mut [u8]
}
}
pub fn rsh<T: VecZnxCommon>(log_base2k: usize, a: &mut T, k: usize, tmp_bytes: &mut [u8]) {
pub fn rsh(log_base2k: usize, a: &mut VecZnx, k: usize, tmp_bytes: &mut [u8]) {
let n: usize = a.n();
debug_assert!(
@@ -469,26 +307,23 @@ pub fn rsh<T: VecZnxCommon>(log_base2k: usize, a: &mut T, k: usize, tmp_bytes: &
znx::znx_zero_i64_ref(n as u64, carry_i64.as_mut_ptr());
}
let mask: i64 = (1 << k_rem) - 1;
let log_base2k: usize = log_base2k;
(cols_steps..cols).for_each(|i| {
izip!(carry_i64.iter_mut(), a.at_mut(i).iter_mut()).for_each(|(ci, xi)| {
*xi += *ci << log_base2k;
*ci = get_base_k_carry(*xi, k_rem);
*xi = (*xi-*ci)>>k_rem;
*xi = (*xi - *ci) >> k_rem;
});
})
}
}
#[inline(always)]
fn get_base_k_carry(x: i64, k: usize) -> i64{
(x<<64-k) >> (64-k)
fn get_base_k_carry(x: i64, k: usize) -> i64 {
(x << 64 - k) >> (64 - k)
}
pub trait VecZnxCommon: VecZnxApi + Infos {}
pub trait VecZnxOps {
/// Allocates a new [VecZnx].
///
@@ -504,50 +339,34 @@ pub trait VecZnxOps {
fn vec_znx_normalize_tmp_bytes(&self) -> usize;
/// c <- a + b.
fn vec_znx_add<A: VecZnxCommon, B: VecZnxCommon, C: VecZnxCommon>(
&self,
c: &mut C,
a: &A,
b: &B,
);
fn vec_znx_add(&self, c: &mut VecZnx, a: &VecZnx, b: &VecZnx);
/// b <- b + a.
fn vec_znx_add_inplace<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &A);
fn vec_znx_add_inplace(&self, b: &mut VecZnx, a: &VecZnx);
/// c <- a - b.
fn vec_znx_sub<A: VecZnxCommon, B: VecZnxCommon, C: VecZnxCommon>(
&self,
c: &mut C,
a: &A,
b: &B,
);
fn vec_znx_sub(&self, c: &mut VecZnx, a: &VecZnx, b: &VecZnx);
/// b <- b - a.
fn vec_znx_sub_inplace<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &A);
fn vec_znx_sub_inplace(&self, b: &mut VecZnx, a: &VecZnx);
/// b <- -a.
fn vec_znx_negate<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &A);
fn vec_znx_negate(&self, b: &mut VecZnx, a: &VecZnx);
/// b <- -b.
fn vec_znx_negate_inplace<A: VecZnxCommon>(&self, a: &mut A);
fn vec_znx_negate_inplace(&self, a: &mut VecZnx);
/// b <- a * X^k (mod X^{n} + 1)
fn vec_znx_rotate<A: VecZnxCommon, B: VecZnxCommon>(&self, k: i64, b: &mut B, a: &A);
fn vec_znx_rotate(&self, k: i64, b: &mut VecZnx, a: &VecZnx);
/// a <- a * X^k (mod X^{n} + 1)
fn vec_znx_rotate_inplace<A: VecZnxCommon>(&self, k: i64, a: &mut A);
fn vec_znx_rotate_inplace(&self, k: i64, a: &mut VecZnx);
/// b <- phi_k(a) where phi_k: X^i -> X^{i*k} (mod (X^{n} + 1))
fn vec_znx_automorphism<A: VecZnxCommon, B: VecZnxCommon>(
&self,
k: i64,
b: &mut B,
a: &A,
a_cols: usize,
);
fn vec_znx_automorphism(&self, k: i64, b: &mut VecZnx, a: &VecZnx, a_cols: usize);
/// a <- phi_k(a) where phi_k: X^i -> X^{i*k} (mod (X^{n} + 1))
fn vec_znx_automorphism_inplace<A: VecZnxCommon>(&self, k: i64, a: &mut A, a_cols: usize);
fn vec_znx_automorphism_inplace(&self, k: i64, a: &mut VecZnx, a_cols: usize);
/// Splits b into subrings and copies them them into a.
///
@@ -555,12 +374,7 @@ pub trait VecZnxOps {
///
/// This method requires that all [VecZnx] of b have the same ring degree
/// and that b.n() * b.len() <= a.n()
fn vec_znx_split<A: VecZnxCommon, B: VecZnxCommon, C: VecZnxCommon>(
&self,
b: &mut Vec<B>,
a: &A,
buf: &mut C,
);
fn vec_znx_split(&self, b: &mut Vec<VecZnx>, a: &VecZnx, buf: &mut VecZnx);
/// Merges the subrings a into b.
///
@@ -568,7 +382,7 @@ pub trait VecZnxOps {
///
/// This method requires that all [VecZnx] of a have the same ring degree
/// and that a.n() * a.len() <= b.n()
fn vec_znx_merge<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &Vec<A>);
fn vec_znx_merge(&self, b: &mut VecZnx, a: &Vec<VecZnx>);
}
impl VecZnxOps for Module {
@@ -581,19 +395,14 @@ impl VecZnxOps for Module {
}
fn vec_znx_normalize_tmp_bytes(&self) -> usize {
unsafe { vec_znx::vec_znx_normalize_base2k_tmp_bytes(self.0) as usize }
unsafe { vec_znx::vec_znx_normalize_base2k_tmp_bytes(self.ptr) as usize }
}
// c <- a + b
fn vec_znx_add<A: VecZnxCommon, B: VecZnxCommon, C: VecZnxCommon>(
&self,
c: &mut C,
a: &A,
b: &B,
) {
fn vec_znx_add(&self, c: &mut VecZnx, a: &VecZnx, b: &VecZnx) {
unsafe {
vec_znx::vec_znx_add(
self.0,
self.ptr,
c.as_mut_ptr(),
c.cols() as u64,
c.n() as u64,
@@ -608,10 +417,10 @@ impl VecZnxOps for Module {
}
// b <- a + b
fn vec_znx_add_inplace<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &A) {
fn vec_znx_add_inplace(&self, b: &mut VecZnx, a: &VecZnx) {
unsafe {
vec_znx::vec_znx_add(
self.0,
self.ptr,
b.as_mut_ptr(),
b.cols() as u64,
b.n() as u64,
@@ -626,15 +435,10 @@ impl VecZnxOps for Module {
}
// c <- a + b
fn vec_znx_sub<A: VecZnxCommon, B: VecZnxCommon, C: VecZnxCommon>(
&self,
c: &mut C,
a: &A,
b: &B,
) {
fn vec_znx_sub(&self, c: &mut VecZnx, a: &VecZnx, b: &VecZnx) {
unsafe {
vec_znx::vec_znx_sub(
self.0,
self.ptr,
c.as_mut_ptr(),
c.cols() as u64,
c.n() as u64,
@@ -649,10 +453,10 @@ impl VecZnxOps for Module {
}
// b <- a + b
fn vec_znx_sub_inplace<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &A) {
fn vec_znx_sub_inplace(&self, b: &mut VecZnx, a: &VecZnx) {
unsafe {
vec_znx::vec_znx_sub(
self.0,
self.ptr,
b.as_mut_ptr(),
b.cols() as u64,
b.n() as u64,
@@ -666,10 +470,10 @@ impl VecZnxOps for Module {
}
}
fn vec_znx_negate<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &A) {
fn vec_znx_negate(&self, b: &mut VecZnx, a: &VecZnx) {
unsafe {
vec_znx::vec_znx_negate(
self.0,
self.ptr,
b.as_mut_ptr(),
b.cols() as u64,
b.n() as u64,
@@ -680,10 +484,10 @@ impl VecZnxOps for Module {
}
}
fn vec_znx_negate_inplace<A: VecZnxCommon>(&self, a: &mut A) {
fn vec_znx_negate_inplace(&self, a: &mut VecZnx) {
unsafe {
vec_znx::vec_znx_negate(
self.0,
self.ptr,
a.as_mut_ptr(),
a.cols() as u64,
a.n() as u64,
@@ -694,10 +498,10 @@ impl VecZnxOps for Module {
}
}
fn vec_znx_rotate<A: VecZnxCommon, B: VecZnxCommon>(&self, k: i64, b: &mut B, a: &A) {
fn vec_znx_rotate(&self, k: i64, b: &mut VecZnx, a: &VecZnx) {
unsafe {
vec_znx::vec_znx_rotate(
self.0,
self.ptr,
k,
b.as_mut_ptr(),
b.cols() as u64,
@@ -709,10 +513,10 @@ impl VecZnxOps for Module {
}
}
fn vec_znx_rotate_inplace<A: VecZnxCommon>(&self, k: i64, a: &mut A) {
fn vec_znx_rotate_inplace(&self, k: i64, a: &mut VecZnx) {
unsafe {
vec_znx::vec_znx_rotate(
self.0,
self.ptr,
k,
a.as_mut_ptr(),
a.cols() as u64,
@@ -739,11 +543,11 @@ impl VecZnxOps for Module {
///
/// # Example
/// ```
/// use base2k::{Module, FFT64, VecZnx, Encoding, Infos, VecZnxApi, VecZnxOps};
/// use base2k::{Module, MODULETYPE, VecZnx, Encoding, Infos, VecZnxOps};
/// use itertools::izip;
///
/// let n: usize = 8; // polynomial degree
/// let module = Module::new::<FFT64>(n);
/// let module = Module::new(n, MODULETYPE::FFT64);
/// let mut a: VecZnx = module.new_vec_znx(2);
/// let mut b: VecZnx = module.new_vec_znx(2);
/// let mut c: VecZnx = module.new_vec_znx(2);
@@ -759,21 +563,15 @@ impl VecZnxOps for Module {
/// (1..col.len()).for_each(|i|{
/// col[n-i] = -(i as i64)
/// });
/// izip!(b.data.iter(), c.data.iter()).for_each(|(a, b)| assert_eq!(a, b, "{} != {}", a, b));
/// izip!(b.raw().iter(), c.raw().iter()).for_each(|(a, b)| assert_eq!(a, b, "{} != {}", a, b));
/// ```
fn vec_znx_automorphism<A: VecZnxCommon, B: VecZnxCommon>(
&self,
k: i64,
b: &mut B,
a: &A,
a_cols: usize,
) {
fn vec_znx_automorphism(&self, k: i64, b: &mut VecZnx, a: &VecZnx, a_cols: usize) {
debug_assert_eq!(a.n(), self.n());
debug_assert_eq!(b.n(), self.n());
debug_assert!(a.cols() >= a_cols);
unsafe {
vec_znx::vec_znx_automorphism(
self.0,
self.ptr,
k,
b.as_mut_ptr(),
b.cols() as u64,
@@ -799,11 +597,11 @@ impl VecZnxOps for Module {
///
/// # Example
/// ```
/// use base2k::{Module, FFT64, VecZnx, Encoding, Infos, VecZnxApi, VecZnxOps};
/// use base2k::{Module, MODULETYPE, VecZnx, Encoding, Infos, VecZnxOps};
/// use itertools::izip;
///
/// let n: usize = 8; // polynomial degree
/// let module = Module::new::<FFT64>(n);
/// let module = Module::new(n, MODULETYPE::FFT64);
/// let mut a: VecZnx = VecZnx::new(n, 2);
/// let mut b: VecZnx = VecZnx::new(n, 2);
///
@@ -818,14 +616,14 @@ impl VecZnxOps for Module {
/// (1..col.len()).for_each(|i|{
/// col[n-i] = -(i as i64)
/// });
/// izip!(a.data.iter(), b.data.iter()).for_each(|(a, b)| assert_eq!(a, b, "{} != {}", a, b));
/// izip!(a.raw().iter(), b.raw().iter()).for_each(|(a, b)| assert_eq!(a, b, "{} != {}", a, b));
/// ```
fn vec_znx_automorphism_inplace<A: VecZnxCommon>(&self, k: i64, a: &mut A, a_cols: usize) {
fn vec_znx_automorphism_inplace(&self, k: i64, a: &mut VecZnx, a_cols: usize) {
debug_assert_eq!(a.n(), self.n());
debug_assert!(a.cols() >= a_cols);
unsafe {
vec_znx::vec_znx_automorphism(
self.0,
self.ptr,
k,
a.as_mut_ptr(),
a.cols() as u64,
@@ -837,12 +635,7 @@ impl VecZnxOps for Module {
}
}
fn vec_znx_split<A: VecZnxCommon, B: VecZnxCommon, C: VecZnxCommon>(
&self,
b: &mut Vec<B>,
a: &A,
buf: &mut C,
) {
fn vec_znx_split(&self, b: &mut Vec<VecZnx>, a: &VecZnx, buf: &mut VecZnx) {
let (n_in, n_out) = (a.n(), b[0].n());
debug_assert!(
@@ -868,7 +661,7 @@ impl VecZnxOps for Module {
})
}
fn vec_znx_merge<A: VecZnxCommon, B: VecZnxCommon>(&self, b: &mut B, a: &Vec<A>) {
fn vec_znx_merge(&self, b: &mut VecZnx, a: &Vec<VecZnx>) {
let (n_in, n_out) = (b.n(), a[0].n());
debug_assert!(

View File

@@ -1,29 +1,69 @@
use crate::ffi::vec_znx_big;
use crate::ffi::vec_znx_dft;
use crate::{assert_alignement, Infos, Module, VecZnxApi, VecZnxDft};
use crate::ffi::vec_znx_big::{self, vec_znx_bigcoeff_t};
use crate::{alloc_aligned, assert_alignement, Infos, Module, VecZnx, VecZnxDft, MODULETYPE};
pub struct VecZnxBig(pub *mut vec_znx_big::vec_znx_bigcoeff_t, pub usize);
pub struct VecZnxBig {
pub data: Vec<u8>,
pub ptr: *mut u8,
pub n: usize,
pub cols: usize,
pub backend: MODULETYPE,
}
impl VecZnxBig {
/// Returns a new [VecZnxBig] with the provided data as backing array.
/// User must ensure that data is properly alligned and that
/// the size of data is at least equal to [Module::bytes_of_vec_znx_big].
pub fn from_bytes(cols: usize, bytes: &mut [u8]) -> VecZnxBig {
pub fn from_bytes(module: &Module, cols: usize, bytes: &mut [u8]) -> Self {
#[cfg(debug_assertions)]
{
assert_alignement(bytes.as_ptr())
};
VecZnxBig(
bytes.as_mut_ptr() as *mut vec_znx_big::vec_znx_bigcoeff_t,
cols,
)
unsafe {
Self {
data: Vec::from_raw_parts(bytes.as_mut_ptr(), bytes.len(), bytes.len()),
ptr: bytes.as_mut_ptr(),
n: module.n(),
cols: cols,
backend: module.backend,
}
}
}
pub fn from_bytes_borrow(module: &Module, cols: usize, bytes: &mut [u8]) -> Self {
#[cfg(debug_assertions)]
{
assert_eq!(bytes.len(), module.bytes_of_vec_znx_big(cols));
assert_alignement(bytes.as_ptr());
}
Self {
data: Vec::new(),
ptr: bytes.as_mut_ptr(),
n: module.n(),
cols: cols,
backend: module.backend,
}
}
pub fn as_vec_znx_dft(&mut self) -> VecZnxDft {
VecZnxDft(self.0 as *mut vec_znx_dft::vec_znx_dft_t, self.1)
VecZnxDft {
data: Vec::new(),
ptr: self.ptr,
n: self.n,
cols: self.cols,
backend: self.backend,
}
}
pub fn n(&self) -> usize {
self.n
}
pub fn cols(&self) -> usize {
self.1
self.cols
}
pub fn backend(&self) -> MODULETYPE {
self.backend
}
}
@@ -47,39 +87,34 @@ pub trait VecZnxBigOps {
fn bytes_of_vec_znx_big(&self, cols: usize) -> usize;
/// b <- b - a
fn vec_znx_big_sub_small_a_inplace<T: VecZnxApi + Infos>(&self, b: &mut VecZnxBig, a: &T);
fn vec_znx_big_sub_small_a_inplace(&self, b: &mut VecZnxBig, a: &VecZnx);
/// c <- b - a
fn vec_znx_big_sub_small_a<T: VecZnxApi + Infos>(
&self,
c: &mut VecZnxBig,
a: &T,
b: &VecZnxBig,
);
fn vec_znx_big_sub_small_a(&self, c: &mut VecZnxBig, a: &VecZnx, b: &VecZnxBig);
/// c <- b + a
fn vec_znx_big_add_small<T: VecZnxApi + Infos>(&self, c: &mut VecZnxBig, a: &T, b: &VecZnxBig);
fn vec_znx_big_add_small(&self, c: &mut VecZnxBig, a: &VecZnx, b: &VecZnxBig);
/// b <- b + a
fn vec_znx_big_add_small_inplace<T: VecZnxApi + Infos>(&self, b: &mut VecZnxBig, a: &T);
fn vec_znx_big_add_small_inplace(&self, b: &mut VecZnxBig, a: &VecZnx);
fn vec_znx_big_normalize_tmp_bytes(&self) -> usize;
/// b <- normalize(a)
fn vec_znx_big_normalize<T: VecZnxApi + Infos>(
fn vec_znx_big_normalize(
&self,
log_base2k: usize,
b: &mut T,
b: &mut VecZnx,
a: &VecZnxBig,
tmp_bytes: &mut [u8],
);
fn vec_znx_big_range_normalize_base2k_tmp_bytes(&self) -> usize;
fn vec_znx_big_range_normalize_base2k<T: VecZnxApi + Infos>(
fn vec_znx_big_range_normalize_base2k(
&self,
log_base2k: usize,
res: &mut T,
res: &mut VecZnx,
a: &VecZnxBig,
a_range_begin: usize,
a_range_xend: usize,
@@ -94,7 +129,15 @@ pub trait VecZnxBigOps {
impl VecZnxBigOps for Module {
fn new_vec_znx_big(&self, cols: usize) -> VecZnxBig {
unsafe { VecZnxBig(vec_znx_big::new_vec_znx_big(self.0, cols as u64), cols) }
let mut data: Vec<u8> = alloc_aligned::<u8>(self.bytes_of_vec_znx_big(cols));
let ptr: *mut u8 = data.as_mut_ptr();
VecZnxBig {
data: data,
ptr: ptr,
n: self.n(),
cols: cols,
backend: self.backend(),
}
}
fn new_vec_znx_big_from_bytes(&self, cols: usize, bytes: &mut [u8]) -> VecZnxBig {
@@ -108,55 +151,50 @@ impl VecZnxBigOps for Module {
{
assert_alignement(bytes.as_ptr())
}
VecZnxBig::from_bytes(cols, bytes)
VecZnxBig::from_bytes(self, cols, bytes)
}
fn bytes_of_vec_znx_big(&self, cols: usize) -> usize {
unsafe { vec_znx_big::bytes_of_vec_znx_big(self.0, cols as u64) as usize }
unsafe { vec_znx_big::bytes_of_vec_znx_big(self.ptr, cols as u64) as usize }
}
fn vec_znx_big_sub_small_a_inplace<T: VecZnxApi + Infos>(&self, b: &mut VecZnxBig, a: &T) {
fn vec_znx_big_sub_small_a_inplace(&self, b: &mut VecZnxBig, a: &VecZnx) {
unsafe {
vec_znx_big::vec_znx_big_sub_small_a(
self.0,
b.0,
self.ptr,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
a.as_ptr(),
a.cols() as u64,
a.n() as u64,
b.0,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
)
}
}
fn vec_znx_big_sub_small_a<T: VecZnxApi + Infos>(
&self,
c: &mut VecZnxBig,
a: &T,
b: &VecZnxBig,
) {
fn vec_znx_big_sub_small_a(&self, c: &mut VecZnxBig, a: &VecZnx, b: &VecZnxBig) {
unsafe {
vec_znx_big::vec_znx_big_sub_small_a(
self.0,
c.0,
self.ptr,
c.ptr as *mut vec_znx_bigcoeff_t,
c.cols() as u64,
a.as_ptr(),
a.cols() as u64,
a.n() as u64,
b.0,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
)
}
}
fn vec_znx_big_add_small<T: VecZnxApi + Infos>(&self, c: &mut VecZnxBig, a: &T, b: &VecZnxBig) {
fn vec_znx_big_add_small(&self, c: &mut VecZnxBig, a: &VecZnx, b: &VecZnxBig) {
unsafe {
vec_znx_big::vec_znx_big_add_small(
self.0,
c.0,
self.ptr,
c.ptr as *mut vec_znx_bigcoeff_t,
c.cols() as u64,
b.0,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
a.as_ptr(),
a.cols() as u64,
@@ -165,13 +203,13 @@ impl VecZnxBigOps for Module {
}
}
fn vec_znx_big_add_small_inplace<T: VecZnxApi + Infos>(&self, b: &mut VecZnxBig, a: &T) {
fn vec_znx_big_add_small_inplace(&self, b: &mut VecZnxBig, a: &VecZnx) {
unsafe {
vec_znx_big::vec_znx_big_add_small(
self.0,
b.0,
self.ptr,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
b.0,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
a.as_ptr(),
a.cols() as u64,
@@ -181,13 +219,13 @@ impl VecZnxBigOps for Module {
}
fn vec_znx_big_normalize_tmp_bytes(&self) -> usize {
unsafe { vec_znx_big::vec_znx_big_normalize_base2k_tmp_bytes(self.0) as usize }
unsafe { vec_znx_big::vec_znx_big_normalize_base2k_tmp_bytes(self.ptr) as usize }
}
fn vec_znx_big_normalize<T: VecZnxApi + Infos>(
fn vec_znx_big_normalize(
&self,
log_base2k: usize,
b: &mut T,
b: &mut VecZnx,
a: &VecZnxBig,
tmp_bytes: &mut [u8],
) {
@@ -203,12 +241,12 @@ impl VecZnxBigOps for Module {
}
unsafe {
vec_znx_big::vec_znx_big_normalize_base2k(
self.0,
self.ptr,
log_base2k as u64,
b.as_mut_ptr(),
b.cols() as u64,
b.n() as u64,
a.0,
a.ptr as *mut vec_znx_bigcoeff_t,
a.cols() as u64,
tmp_bytes.as_mut_ptr(),
)
@@ -216,13 +254,13 @@ impl VecZnxBigOps for Module {
}
fn vec_znx_big_range_normalize_base2k_tmp_bytes(&self) -> usize {
unsafe { vec_znx_big::vec_znx_big_range_normalize_base2k_tmp_bytes(self.0) as usize }
unsafe { vec_znx_big::vec_znx_big_range_normalize_base2k_tmp_bytes(self.ptr) as usize }
}
fn vec_znx_big_range_normalize_base2k<T: VecZnxApi + Infos>(
fn vec_znx_big_range_normalize_base2k(
&self,
log_base2k: usize,
res: &mut T,
res: &mut VecZnx,
a: &VecZnxBig,
a_range_begin: usize,
a_range_xend: usize,
@@ -241,12 +279,12 @@ impl VecZnxBigOps for Module {
}
unsafe {
vec_znx_big::vec_znx_big_range_normalize_base2k(
self.0,
self.ptr,
log_base2k as u64,
res.as_mut_ptr(),
res.cols() as u64,
res.n() as u64,
a.0,
a.ptr as *mut vec_znx_bigcoeff_t,
a_range_begin as u64,
a_range_xend as u64,
a_range_step as u64,
@@ -258,11 +296,11 @@ impl VecZnxBigOps for Module {
fn vec_znx_big_automorphism(&self, gal_el: i64, b: &mut VecZnxBig, a: &VecZnxBig) {
unsafe {
vec_znx_big::vec_znx_big_automorphism(
self.0,
self.ptr,
gal_el,
b.0,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
a.0,
a.ptr as *mut vec_znx_bigcoeff_t,
a.cols() as u64,
);
}
@@ -271,11 +309,11 @@ impl VecZnxBigOps for Module {
fn vec_znx_big_automorphism_inplace(&self, gal_el: i64, a: &mut VecZnxBig) {
unsafe {
vec_znx_big::vec_znx_big_automorphism(
self.0,
self.ptr,
gal_el,
a.0,
a.ptr as *mut vec_znx_bigcoeff_t,
a.cols() as u64,
a.0,
a.ptr as *mut vec_znx_bigcoeff_t,
a.cols() as u64,
);
}

View File

@@ -1,33 +1,104 @@
use crate::ffi::vec_znx_big;
use crate::ffi::vec_znx_big::vec_znx_bigcoeff_t;
use crate::ffi::vec_znx_dft;
use crate::ffi::vec_znx_dft::bytes_of_vec_znx_dft;
use crate::{assert_alignement, Infos, Module, VecZnxApi, VecZnxBig};
use crate::ffi::vec_znx_dft::{bytes_of_vec_znx_dft, vec_znx_dft_t};
use crate::{alloc_aligned, VecZnx};
use crate::{assert_alignement, Infos, Module, VecZnxBig, MODULETYPE};
pub struct VecZnxDft(pub *mut vec_znx_dft::vec_znx_dft_t, pub usize);
pub struct VecZnxDft {
pub data: Vec<u8>,
pub ptr: *mut u8,
pub n: usize,
pub cols: usize,
pub backend: MODULETYPE,
}
impl VecZnxDft {
/// Returns a new [VecZnxDft] with the provided data as backing array.
/// User must ensure that data is properly alligned and that
/// the size of data is at least equal to [Module::bytes_of_vec_znx_dft].
pub fn from_bytes(cols: usize, tmp_bytes: &mut [u8]) -> VecZnxDft {
pub fn from_bytes(module: &Module, cols: usize, bytes: &mut [u8]) -> VecZnxDft {
#[cfg(debug_assertions)]
{
assert_alignement(tmp_bytes.as_ptr())
assert_eq!(bytes.len(), module.bytes_of_vec_znx_dft(cols));
assert_alignement(bytes.as_ptr())
}
unsafe {
VecZnxDft {
data: Vec::from_raw_parts(bytes.as_mut_ptr(), bytes.len(), bytes.len()),
ptr: bytes.as_mut_ptr(),
n: module.n(),
cols: cols,
backend: module.backend,
}
}
}
pub fn from_bytes_borrow(module: &Module, cols: usize, bytes: &mut [u8]) -> VecZnxDft {
#[cfg(debug_assertions)]
{
assert_eq!(bytes.len(), module.bytes_of_vec_znx_dft(cols));
assert_alignement(bytes.as_ptr());
}
VecZnxDft {
data: Vec::new(),
ptr: bytes.as_mut_ptr(),
n: module.n(),
cols: cols,
backend: module.backend,
}
VecZnxDft(
tmp_bytes.as_mut_ptr() as *mut vec_znx_dft::vec_znx_dft_t,
cols,
)
}
/// Cast a [VecZnxDft] into a [VecZnxBig].
/// The returned [VecZnxBig] shares the backing array
/// with the original [VecZnxDft].
pub fn as_vec_znx_big(&mut self) -> VecZnxBig {
VecZnxBig(self.0 as *mut vec_znx_big::vec_znx_bigcoeff_t, self.1)
VecZnxBig {
data: Vec::new(),
ptr: self.ptr,
n: self.n,
cols: self.cols,
backend: self.backend,
}
}
pub fn n(&self) -> usize {
self.n
}
pub fn cols(&self) -> usize {
self.1
self.cols
}
pub fn backend(&self) -> MODULETYPE {
self.backend
}
/// Returns a non-mutable reference of `T` of the entire contiguous array of the [VecZnxDft].
/// When using [`crate::FFT64`] as backend, `T` should be [f64].
/// When using [`crate::NTT120`] as backend, `T` should be [i64].
/// The length of the returned array is cols * n.
pub fn raw<T>(&self, module: &Module) -> &[T] {
let ptr: *const T = self.ptr as *const T;
let len: usize = (self.cols() * module.n() * 8) / std::mem::size_of::<T>();
unsafe { &std::slice::from_raw_parts(ptr, len) }
}
pub fn at<T>(&self, module: &Module, col_i: usize) -> &[T] {
&self.raw::<T>(module)[col_i * module.n()..(col_i + 1) * module.n()]
}
/// Returns a mutable reference of `T` of the entire contiguous array of the [VecZnxDft].
/// When using [`crate::FFT64`] as backend, `T` should be [f64].
/// When using [`crate::NTT120`] as backend, `T` should be [i64].
/// The length of the returned array is cols * n.
pub fn raw_mut<T>(&self, module: &Module) -> &mut [T] {
let ptr: *mut T = self.ptr as *mut T;
let len: usize = (self.cols() * module.n() * 8) / std::mem::size_of::<T>();
unsafe { std::slice::from_raw_parts_mut(ptr, len) }
}
pub fn at_mut<T>(&self, module: &Module, col_i: usize) -> &mut [T] {
&mut self.raw_mut::<T>(module)[col_i * module.n()..(col_i + 1) * module.n()]
}
}
@@ -72,12 +143,20 @@ pub trait VecZnxDftOps {
tmp_bytes: &mut [u8],
);
fn vec_znx_dft<T: VecZnxApi + Infos>(&self, b: &mut VecZnxDft, a: &T, a_limbs: usize);
fn vec_znx_dft(&self, b: &mut VecZnxDft, a: &VecZnx, a_limbs: usize);
}
impl VecZnxDftOps for Module {
fn new_vec_znx_dft(&self, cols: usize) -> VecZnxDft {
unsafe { VecZnxDft(vec_znx_dft::new_vec_znx_dft(self.0, cols as u64), cols) }
let mut data: Vec<u8> = alloc_aligned::<u8>(self.bytes_of_vec_znx_dft(cols));
let ptr: *mut u8 = data.as_mut_ptr();
VecZnxDft {
data: data,
ptr: ptr,
n: self.n(),
cols: cols,
backend: self.backend(),
}
}
fn new_vec_znx_dft_from_bytes(&self, cols: usize, tmp_bytes: &mut [u8]) -> VecZnxDft {
@@ -91,11 +170,11 @@ impl VecZnxDftOps for Module {
{
assert_alignement(tmp_bytes.as_ptr())
}
VecZnxDft::from_bytes(cols, tmp_bytes)
VecZnxDft::from_bytes(self, cols, tmp_bytes)
}
fn bytes_of_vec_znx_dft(&self, cols: usize) -> usize {
unsafe { bytes_of_vec_znx_dft(self.0, cols as u64) as usize }
unsafe { bytes_of_vec_znx_dft(self.ptr, cols as u64) as usize }
}
fn vec_znx_idft_tmp_a(&self, b: &mut VecZnxBig, a: &mut VecZnxDft, a_limbs: usize) {
@@ -106,19 +185,25 @@ impl VecZnxDftOps for Module {
a_limbs
);
unsafe {
vec_znx_dft::vec_znx_idft_tmp_a(self.0, b.0, b.cols() as u64, a.0, a_limbs as u64)
vec_znx_dft::vec_znx_idft_tmp_a(
self.ptr,
b.ptr as *mut vec_znx_bigcoeff_t,
b.cols() as u64,
a.ptr as *mut vec_znx_dft_t,
a_limbs as u64,
)
}
}
fn vec_znx_idft_tmp_bytes(&self) -> usize {
unsafe { vec_znx_dft::vec_znx_idft_tmp_bytes(self.0) as usize }
unsafe { vec_znx_dft::vec_znx_idft_tmp_bytes(self.ptr) as usize }
}
/// b <- DFT(a)
///
/// # Panics
/// If b.cols < a_cols
fn vec_znx_dft<T: VecZnxApi + Infos>(&self, b: &mut VecZnxDft, a: &T, a_cols: usize) {
fn vec_znx_dft(&self, b: &mut VecZnxDft, a: &VecZnx, a_cols: usize) {
debug_assert!(
b.cols() >= a_cols,
"invalid a_cols: b.cols()={} < a_cols={}",
@@ -127,8 +212,8 @@ impl VecZnxDftOps for Module {
);
unsafe {
vec_znx_dft::vec_znx_dft(
self.0,
b.0,
self.ptr,
b.ptr as *mut vec_znx_dft_t,
b.cols() as u64,
a.as_ptr(),
a_cols as u64,
@@ -169,10 +254,10 @@ impl VecZnxDftOps for Module {
}
unsafe {
vec_znx_dft::vec_znx_idft(
self.0,
b.0,
self.ptr,
b.ptr as *mut vec_znx_bigcoeff_t,
a.cols() as u64,
a.0,
a.ptr as *mut vec_znx_dft_t,
a_cols as u64,
tmp_bytes.as_mut_ptr(),
)

View File

@@ -1,5 +1,6 @@
use crate::ffi::vmp;
use crate::{assert_alignement, Infos, Module, VecZnxApi, VecZnxDft};
use crate::ffi::vec_znx_dft::vec_znx_dft_t;
use crate::ffi::vmp::{self, vmp_pmat_t};
use crate::{alloc_aligned, assert_alignement, Infos, Module, VecZnx, VecZnxDft, MODULETYPE};
/// Vector Matrix Product Prepared Matrix: a vector of [VecZnx],
/// stored as a 3D matrix in the DFT domain in a single contiguous array.
@@ -11,20 +12,75 @@ use crate::{assert_alignement, Infos, Module, VecZnxApi, VecZnxDft};
/// [VmpPMat] is used to permform a vector matrix product between a [VecZnx] and a [VmpPMat].
/// See the trait [VmpPMatOps] for additional information.
pub struct VmpPMat {
/// The pointer to the C memory.
pub data: *mut vmp::vmp_pmat_t,
/// Raw data, is empty if borrowing scratch space.
data: Vec<u8>,
/// Pointer to data. Can point to scratch space.
ptr: *mut u8,
/// The number of [VecZnxDft].
pub rows: usize,
rows: usize,
/// The number of cols in each [VecZnxDft].
pub cols: usize,
cols: usize,
/// The ring degree of each [VecZnxDft].
pub n: usize,
n: usize,
backend: MODULETYPE,
}
impl Infos for VmpPMat {
/// Returns the ring dimension of the [VmpPMat].
fn n(&self) -> usize {
self.n
}
fn log_n(&self) -> usize {
(usize::BITS - (self.n() - 1).leading_zeros()) as _
}
/// Returns the number of rows (i.e. of [VecZnxDft]) of the [VmpPMat]
fn rows(&self) -> usize {
self.rows
}
/// Returns the number of cols of the [VmpPMat].
/// The number of cols refers to the number of cols
/// of each [VecZnxDft].
/// This method is equivalent to [Self::cols].
fn cols(&self) -> usize {
self.cols
}
}
impl VmpPMat {
/// Returns the pointer to the [vmp_pmat_t].
pub fn data(&self) -> *mut vmp::vmp_pmat_t {
self.data
pub fn as_ptr(&self) -> *const u8 {
self.ptr
}
pub fn as_mut_ptr(&self) -> *mut u8 {
self.ptr
}
pub fn borrowed(&self) -> bool{
self.data.len() == 0
}
/// Returns a non-mutable reference of `T` of the entire contiguous array of the [VmpPMat].
/// When using [`crate::FFT64`] as backend, `T` should be [f64].
/// When using [`crate::NTT120`] as backend, `T` should be [i64].
/// The length of the returned array is rows * cols * n.
pub fn raw<T>(&self) -> &[T] {
let ptr: *const T = self.ptr as *const T;
let len: usize = (self.rows() * self.cols() * self.n() * 8) / std::mem::size_of::<T>();
unsafe { &std::slice::from_raw_parts(ptr, len) }
}
/// Returns a non-mutable reference of `T` of the entire contiguous array of the [VmpPMat].
/// When using [`crate::FFT64`] as backend, `T` should be [f64].
/// When using [`crate::NTT120`] as backend, `T` should be [i64].
/// The length of the returned array is rows * cols * n.
pub fn raw_mut<T>(&self) -> &mut [T] {
let ptr: *mut T = self.ptr as *mut T;
let len: usize = (self.rows() * self.cols() * self.n() * 8) / std::mem::size_of::<T>();
unsafe { std::slice::from_raw_parts_mut(ptr, len) }
}
/// Returns a copy of the backend array at index (i, j) of the [VmpPMat].
@@ -36,16 +92,16 @@ impl VmpPMat {
/// * `row`: row index (i).
/// * `col`: col index (j).
pub fn at<T: Default + Copy>(&self, row: usize, col: usize) -> Vec<T> {
let mut res: Vec<T> = vec![T::default(); self.n];
let mut res: Vec<T> = alloc_aligned(self.n);
if self.n < 8 {
res.copy_from_slice(
&self.get_backend_array::<T>()[(row + col * self.rows()) * self.n()
&self.raw::<T>()[(row + col * self.rows()) * self.n()
..(row + col * self.rows()) * (self.n() + 1)],
);
} else {
(0..self.n >> 3).for_each(|blk| {
res[blk * 8..(blk + 1) * 8].copy_from_slice(&self.get_array(row, col, blk)[..8]);
res[blk * 8..(blk + 1) * 8].copy_from_slice(&self.at_block(row, col, blk)[..8]);
});
}
@@ -54,33 +110,25 @@ impl VmpPMat {
/// When using [`crate::FFT64`] as backend, `T` should be [f64].
/// When using [`crate::NTT120`] as backend, `T` should be [i64].
fn get_array<T>(&self, row: usize, col: usize, blk: usize) -> &[T] {
fn at_block<T>(&self, row: usize, col: usize, blk: usize) -> &[T] {
let nrows: usize = self.rows();
let ncols: usize = self.cols();
if col == (ncols - 1) && (ncols & 1 == 1) {
&self.get_backend_array::<T>()[blk * nrows * ncols * 8 + col * nrows * 8 + row * 8..]
&self.raw::<T>()[blk * nrows * ncols * 8 + col * nrows * 8 + row * 8..]
} else {
&self.get_backend_array::<T>()[blk * nrows * ncols * 8
&self.raw::<T>()[blk * nrows * ncols * 8
+ (col / 2) * (2 * nrows) * 8
+ row * 2 * 8
+ (col % 2) * 8..]
}
}
/// Returns a non-mutable reference of `T` of the entire contiguous array of the [VmpPMat].
/// When using [`crate::FFT64`] as backend, `T` should be [f64].
/// When using [`crate::NTT120`] as backend, `T` should be [i64].
/// The length of the returned array is rows * cols * n.
pub fn get_backend_array<T>(&self) -> &[T] {
let ptr: *const T = self.data as *const T;
let len: usize = (self.rows() * self.cols() * self.n() * 8) / std::mem::size_of::<T>();
unsafe { &std::slice::from_raw_parts(ptr, len) }
}
}
/// This trait implements methods for vector matrix product,
/// that is, multiplying a [VecZnx] with a [VmpPMat].
pub trait VmpPMatOps {
fn bytes_of_vmp_pmat(&self, rows: usize, cols: usize) -> usize;
/// Allocates a new [VmpPMat] with the given number of rows and columns.
///
/// # Arguments
@@ -106,26 +154,6 @@ pub trait VmpPMatOps {
/// * `b`: [VmpPMat] on which the values are encoded.
/// * `a`: the contiguous array of [i64] of the 3D matrix to encode on the [VmpPMat].
/// * `buf`: scratch space, the size of buf can be obtained with [VmpPMatOps::vmp_prepare_tmp_bytes].
///
/// # Example
/// ```
/// use base2k::{Module, VmpPMat, VmpPMatOps, FFT64, Free, alloc_aligned};
/// use std::cmp::min;
///
/// let n: usize = 1024;
/// let module = Module::new::<FFT64>(n);
/// let rows = 5;
/// let cols = 6;
///
/// let mut b_mat: Vec<i64> = vec![0i64;n * cols * rows];
///
/// let mut buf: Vec<u8> = alloc_aligned(module.vmp_prepare_tmp_bytes(rows, cols));
///
/// let mut vmp_pmat: VmpPMat = module.new_vmp_pmat(rows, cols);
/// module.vmp_prepare_contiguous(&mut vmp_pmat, &b_mat, &mut buf);
///
/// vmp_pmat.free() // don't forget to free the memory once vmp_pmat is not needed anymore.
/// ```
fn vmp_prepare_contiguous(&self, b: &mut VmpPMat, a: &[i64], buf: &mut [u8]);
/// Prepares a [VmpPMat] from a vector of [VecZnx].
@@ -137,32 +165,6 @@ pub trait VmpPMatOps {
/// * `buf`: scratch space, the size of buf can be obtained with [VmpPMatOps::vmp_prepare_tmp_bytes].
///
/// The size of buf can be obtained with [VmpPMatOps::vmp_prepare_tmp_bytes].
///
/// # Example
/// ```
/// use base2k::{Module, FFT64, VmpPMat, VmpPMatOps, VecZnx, VecZnxApi, VecZnxOps, Free, alloc_aligned};
/// use std::cmp::min;
///
/// let n: usize = 1024;
/// let module: Module = Module::new::<FFT64>(n);
/// let rows: usize = 5;
/// let cols: usize = 6;
///
/// let mut vecznx: Vec<VecZnx>= Vec::new();
/// (0..rows).for_each(|_|{
/// vecznx.push(module.new_vec_znx(cols));
/// });
///
/// let slices: Vec<&[i64]> = vecznx.iter().map(|v| v.data.as_slice()).collect();
///
/// let mut buf: Vec<u8> = alloc_aligned(module.vmp_prepare_tmp_bytes(rows, cols));
///
/// let mut vmp_pmat: VmpPMat = module.new_vmp_pmat(rows, cols);
/// module.vmp_prepare_dblptr(&mut vmp_pmat, &slices, &mut buf);
///
/// vmp_pmat.free();
/// module.free();
/// ```
fn vmp_prepare_dblptr(&self, b: &mut VmpPMat, a: &[&[i64]], buf: &mut [u8]);
/// Prepares the ith-row of [VmpPMat] from a vector of [VecZnx].
@@ -175,26 +177,6 @@ pub trait VmpPMatOps {
/// * `buf`: scratch space, the size of buf can be obtained with [VmpPMatOps::vmp_prepare_tmp_bytes].
///
/// The size of buf can be obtained with [VmpPMatOps::vmp_prepare_tmp_bytes].
/// /// # Example
/// ```
/// use base2k::{Module, FFT64, VmpPMat, VmpPMatOps, VecZnx, VecZnxApi, VecZnxOps, Free, alloc_aligned};
/// use std::cmp::min;
///
/// let n: usize = 1024;
/// let module: Module = Module::new::<FFT64>(n);
/// let rows: usize = 5;
/// let cols: usize = 6;
///
/// let vecznx = module.new_vec_znx(cols);
///
/// let mut buf: Vec<u8> = alloc_aligned(module.vmp_prepare_tmp_bytes(rows, cols));
///
/// let mut vmp_pmat: VmpPMat = module.new_vmp_pmat(rows, cols);
/// module.vmp_prepare_row(&mut vmp_pmat, vecznx.raw(), 0, &mut buf);
///
/// vmp_pmat.free();
/// module.free();
/// ```
fn vmp_prepare_row(&self, b: &mut VmpPMat, a: &[i64], row_i: usize, tmp_bytes: &mut [u8]);
/// Returns the size of the stratch space necessary for [VmpPMatOps::vmp_apply_dft].
@@ -237,38 +219,7 @@ pub trait VmpPMatOps {
/// * `a`: the left operand [VecZnx] of the vector matrix product.
/// * `b`: the right operand [VmpPMat] of the vector matrix product.
/// * `buf`: scratch space, the size can be obtained with [VmpPMatOps::vmp_apply_dft_tmp_bytes].
///
/// # Example
/// ```
/// use base2k::{Module, VecZnx, VecZnxOps, VecZnxDft, VecZnxDftOps, VmpPMat, VmpPMatOps, FFT64, Free, VecZnxApi, alloc_aligned};
///
/// let n = 1024;
///
/// let module: Module = Module::new::<FFT64>(n);
/// let cols: usize = 5;
///
/// let rows: usize = cols;
/// let cols: usize = cols + 1;
/// let c_cols: usize = cols;
/// let a_cols: usize = cols;
/// let mut buf: Vec<u8> = alloc_aligned(module.vmp_apply_dft_tmp_bytes(c_cols, a_cols, rows, cols));
/// let mut vmp_pmat: VmpPMat = module.new_vmp_pmat(rows, cols);
///
/// let a: VecZnx = module.new_vec_znx(cols);
/// let mut c_dft: VecZnxDft = module.new_vec_znx_dft(cols);
/// module.vmp_apply_dft(&mut c_dft, &a, &vmp_pmat, &mut buf);
///
/// c_dft.free();
/// vmp_pmat.free();
/// module.free();
/// ```
fn vmp_apply_dft<T: VecZnxApi + Infos>(
&self,
c: &mut VecZnxDft,
a: &T,
b: &VmpPMat,
buf: &mut [u8],
);
fn vmp_apply_dft(&self, c: &mut VecZnxDft, a: &VecZnx, b: &VmpPMat, buf: &mut [u8]);
/// Returns the size of the stratch space necessary for [VmpPMatOps::vmp_apply_dft_to_dft].
///
@@ -311,32 +262,6 @@ pub trait VmpPMatOps {
/// * `a`: the left operand [VecZnxDft] of the vector matrix product.
/// * `b`: the right operand [VmpPMat] of the vector matrix product.
/// * `buf`: scratch space, the size can be obtained with [VmpPMatOps::vmp_apply_dft_to_dft_tmp_bytes].
///
/// # Example
/// ```
/// use base2k::{Module, VecZnx, VecZnxDft, VecZnxDftOps, VmpPMat, VmpPMatOps, FFT64, Free, alloc_aligned};
///
/// let n = 1024;
///
/// let module: Module = Module::new::<FFT64>(n);
/// let cols: usize = 5;
///
/// let rows: usize = cols;
/// let cols: usize = cols + 1;
/// let c_cols: usize = cols;
/// let a_cols: usize = cols;
/// let mut tmp_bytes: Vec<u8> = alloc_aligned(module.vmp_apply_dft_to_dft_tmp_bytes(c_cols, a_cols, rows, cols));
/// let mut vmp_pmat: VmpPMat = module.new_vmp_pmat(rows, cols);
///
/// let a_dft: VecZnxDft = module.new_vec_znx_dft(cols);
/// let mut c_dft: VecZnxDft = module.new_vec_znx_dft(cols);
/// module.vmp_apply_dft_to_dft(&mut c_dft, &a_dft, &vmp_pmat, &mut tmp_bytes);
///
/// a_dft.free();
/// c_dft.free();
/// vmp_pmat.free();
/// module.free();
/// ```
fn vmp_apply_dft_to_dft(&self, c: &mut VecZnxDft, a: &VecZnxDft, b: &VmpPMat, buf: &mut [u8]);
/// Applies the vector matrix product [VecZnxDft] x [VmpPMat] in place.
@@ -363,46 +288,29 @@ pub trait VmpPMatOps {
/// * `b`: the input and output of the vector matrix product, as a [VecZnxDft].
/// * `a`: the right operand [VmpPMat] of the vector matrix product.
/// * `buf`: scratch space, the size can be obtained with [VmpPMatOps::vmp_apply_dft_to_dft_tmp_bytes].
///
/// # Example
/// ```rust
/// use base2k::{Module, VecZnx, VecZnxOps, VecZnxDft, VmpPMat, VmpPMatOps, FFT64, Free, VecZnxApi, VecZnxDftOps,alloc_aligned};
///
/// let n = 1024;
///
/// let module: Module = Module::new::<FFT64>(n);
/// let cols: usize = 5;
///
/// let rows: usize = cols;
/// let cols: usize = cols + 1;
/// let mut tmp_bytes: Vec<u8> = alloc_aligned(module.vmp_apply_dft_to_dft_tmp_bytes(cols, cols, rows, cols));
/// let a: VecZnx = module.new_vec_znx(cols);
/// let mut vmp_pmat: VmpPMat = module.new_vmp_pmat(rows, cols);
///
/// let mut c_dft: VecZnxDft = module.new_vec_znx_dft(cols);
/// module.vmp_apply_dft_to_dft_inplace(&mut c_dft, &vmp_pmat, &mut tmp_bytes);
///
/// c_dft.free();
/// vmp_pmat.free();
/// module.free();
/// ```
fn vmp_apply_dft_to_dft_inplace(&self, b: &mut VecZnxDft, a: &VmpPMat, buf: &mut [u8]);
}
impl VmpPMatOps for Module {
fn bytes_of_vmp_pmat(&self, rows: usize, cols: usize) -> usize {
unsafe { vmp::bytes_of_vmp_pmat(self.ptr, rows as u64, cols as u64) as usize }
}
fn new_vmp_pmat(&self, rows: usize, cols: usize) -> VmpPMat {
unsafe {
VmpPMat {
data: vmp::new_vmp_pmat(self.0, rows as u64, cols as u64),
rows,
cols,
n: self.n(),
}
let mut data: Vec<u8> = alloc_aligned::<u8>(self.bytes_of_vmp_pmat(rows, cols));
let ptr: *mut u8 = data.as_mut_ptr();
VmpPMat {
data: data,
ptr: ptr,
n: self.n(),
cols: cols,
rows: rows,
backend: self.backend(),
}
}
fn vmp_prepare_tmp_bytes(&self, rows: usize, cols: usize) -> usize {
unsafe { vmp::vmp_prepare_tmp_bytes(self.0, rows as u64, cols as u64) as usize }
unsafe { vmp::vmp_prepare_tmp_bytes(self.ptr, rows as u64, cols as u64) as usize }
}
fn vmp_prepare_contiguous(&self, b: &mut VmpPMat, a: &[i64], tmp_bytes: &mut [u8]) {
@@ -414,8 +322,8 @@ impl VmpPMatOps for Module {
}
unsafe {
vmp::vmp_prepare_contiguous(
self.0,
b.data(),
self.ptr,
b.as_mut_ptr() as *mut vmp_pmat_t,
a.as_ptr(),
b.rows() as u64,
b.cols() as u64,
@@ -437,8 +345,8 @@ impl VmpPMatOps for Module {
}
unsafe {
vmp::vmp_prepare_dblptr(
self.0,
b.data(),
self.ptr,
b.as_mut_ptr() as *mut vmp_pmat_t,
ptrs.as_ptr(),
b.rows() as u64,
b.cols() as u64,
@@ -456,8 +364,8 @@ impl VmpPMatOps for Module {
}
unsafe {
vmp::vmp_prepare_row(
self.0,
b.data(),
self.ptr,
b.as_mut_ptr() as *mut vmp_pmat_t,
a.as_ptr(),
row_i as u64,
b.rows() as u64,
@@ -476,7 +384,7 @@ impl VmpPMatOps for Module {
) -> usize {
unsafe {
vmp::vmp_apply_dft_tmp_bytes(
self.0,
self.ptr,
res_cols as u64,
a_cols as u64,
gct_rows as u64,
@@ -485,13 +393,7 @@ impl VmpPMatOps for Module {
}
}
fn vmp_apply_dft<T: VecZnxApi + Infos>(
&self,
c: &mut VecZnxDft,
a: &T,
b: &VmpPMat,
tmp_bytes: &mut [u8],
) {
fn vmp_apply_dft(&self, c: &mut VecZnxDft, a: &VecZnx, b: &VmpPMat, tmp_bytes: &mut [u8]) {
debug_assert!(
tmp_bytes.len() >= self.vmp_apply_dft_tmp_bytes(c.cols(), a.cols(), b.rows(), b.cols())
);
@@ -501,13 +403,13 @@ impl VmpPMatOps for Module {
}
unsafe {
vmp::vmp_apply_dft(
self.0,
c.0,
self.ptr,
c.ptr as *mut vec_znx_dft_t,
c.cols() as u64,
a.as_ptr(),
a.cols() as u64,
a.n() as u64,
b.data(),
b.as_ptr() as *const vmp_pmat_t,
b.rows() as u64,
b.cols() as u64,
tmp_bytes.as_mut_ptr(),
@@ -524,7 +426,7 @@ impl VmpPMatOps for Module {
) -> usize {
unsafe {
vmp::vmp_apply_dft_to_dft_tmp_bytes(
self.0,
self.ptr,
res_cols as u64,
a_cols as u64,
gct_rows as u64,
@@ -550,12 +452,12 @@ impl VmpPMatOps for Module {
}
unsafe {
vmp::vmp_apply_dft_to_dft(
self.0,
c.0,
self.ptr,
c.ptr as *mut vec_znx_dft_t,
c.cols() as u64,
a.0,
a.ptr as *const vec_znx_dft_t,
a.cols() as u64,
b.data(),
b.as_ptr() as *const vmp_pmat_t,
b.rows() as u64,
b.cols() as u64,
tmp_bytes.as_mut_ptr(),
@@ -574,12 +476,12 @@ impl VmpPMatOps for Module {
}
unsafe {
vmp::vmp_apply_dft_to_dft(
self.0,
b.0,
self.ptr,
b.ptr as *mut vec_znx_dft_t,
b.cols() as u64,
b.0,
b.ptr as *mut vec_znx_dft_t,
b.cols() as u64,
a.data(),
a.as_ptr() as *const vmp_pmat_t,
a.rows() as u64,
a.cols() as u64,
tmp_bytes.as_mut_ptr(),