M-Labs
/
nalgebra
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

put back checked kernels and refactor upper layer

This commit is contained in:
Saurabh 2022-03-21 16:55:46 -06:00
parent 6d26f4f32c
commit c6f832d1d8
3 changed files with 196 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

68
nalgebra-sparse/src/ops/serial/cs.rs
View File

@ -1,5 +1,3 @@
//use std::collections::HashSet;
use crate::cs::CsMatrix;
use crate::ops::serial::{OperationError, OperationErrorKind};
use crate::ops::Op;
@ -7,12 +5,12 @@ use crate::SparseEntryMut;
use nalgebra::{ClosedAdd, ClosedMul, DMatrixSlice, DMatrixSliceMut, Scalar};
use num_traits::{One, Zero};
//fn spmm_cs_unexpected_entry() -> OperationError {
// OperationError::from_kind_and_message(
// OperationErrorKind::InvalidPattern,
// String::from("Found unexpected entry that is not present in `c`."),
// )
//}
fn spmm_cs_unexpected_entry() -> OperationError {
OperationError::from_kind_and_message(
OperationErrorKind::InvalidPattern,
String::from("Found unexpected entry that is not present in `c`."),
)
}
/// Helper functionality for implementing CSR/CSC SPMM.
///
@ -22,7 +20,7 @@ use num_traits::{One, Zero};
/// reversed (since transpose(AB) = transpose(B) * transpose(A) and CSC(A) = transpose(CSR(A)).
///
/// We assume here that the matrices have already been verified to be dimensionally compatible.
pub fn spmm_cs_prealloc<T>(
pub fn spmm_cs_prealloc_unchecked<T>(
beta: T,
c: &mut CsMatrix<T>,
alpha: T,
@ -43,8 +41,10 @@ where
let b_lane_k = b.get_lane(k).unwrap();
let alpha_aik = alpha.clone() * a_ik.clone();
for (j, b_kj) in b_lane_k.minor_indices().iter().zip(b_lane_k.values()) {
// Determine the location in C to append the value
scratchpad_values[*j] += alpha_aik.clone() * b_kj.clone();
// use a dense scatter vector to accumulate non-zeros quickly
unsafe {
*scratchpad_values.get_unchecked_mut(*j) += alpha_aik.clone() * b_kj.clone();
}
}
}
@ -53,15 +53,55 @@ where
values
.iter_mut()
.zip(indices)
.for_each(|(output_ref, index)| {
*output_ref = beta.clone() * output_ref.clone() + scratchpad_values[*index].clone();
scratchpad_values[*index] = Zero::zero();
.for_each(|(output_ref, index)| unsafe {
*output_ref = beta.clone() * output_ref.clone()
+ scratchpad_values.get_unchecked(*index).clone();
*scratchpad_values.get_unchecked_mut(*index) = Zero::zero();
});
}
Ok(())
}
pub fn spmm_cs_prealloc_checked<T>(
beta: T,
c: &mut CsMatrix<T>,
alpha: T,
a: &CsMatrix<T>,
b: &CsMatrix<T>,
) -> Result<(), OperationError>
where
T: Scalar + ClosedAdd + ClosedMul + Zero + One,
{
for i in 0..c.pattern().major_dim() {
let a_lane_i = a.get_lane(i).unwrap();
let mut c_lane_i = c.get_lane_mut(i).unwrap();
for c_ij in c_lane_i.values_mut() {
*c_ij = beta.clone() * c_ij.clone();
}
for (&k, a_ik) in a_lane_i.minor_indices().iter().zip(a_lane_i.values()) {
let b_lane_k = b.get_lane(k).unwrap();
let (mut c_lane_i_cols, mut c_lane_i_values) = c_lane_i.indices_and_values_mut();
let alpha_aik = alpha.clone() * a_ik.clone();
for (j, b_kj) in b_lane_k.minor_indices().iter().zip(b_lane_k.values()) {
// Determine the location in C to append the value
let (c_local_idx, _) = c_lane_i_cols
.iter()
.enumerate()
.find(|(_, c_col)| *c_col == j)
.ok_or_else(spmm_cs_unexpected_entry)?;
c_lane_i_values[c_local_idx] += alpha_aik.clone() * b_kj.clone();
c_lane_i_cols = &c_lane_i_cols[c_local_idx..];
c_lane_i_values = &mut c_lane_i_values[c_local_idx..];
}
}
}
Ok(())
}
fn spadd_cs_unexpected_entry() -> OperationError {
OperationError::from_kind_and_message(
OperationErrorKind::InvalidPattern,

94
nalgebra-sparse/src/ops/serial/csc.rs
View File

@ -1,5 +1,7 @@
use crate::csc::CscMatrix;
use crate::ops::serial::cs::{spadd_cs_prealloc, spmm_cs_dense, spmm_cs_prealloc};
use crate::ops::serial::cs::{
spadd_cs_prealloc, spmm_cs_dense, spmm_cs_prealloc_checked, spmm_cs_prealloc_unchecked,
};
use crate::ops::serial::{OperationError, OperationErrorKind};
use crate::ops::Op;
use nalgebra::{ClosedAdd, ClosedMul, DMatrixSlice, DMatrixSliceMut, RealField, Scalar};
@ -71,7 +73,7 @@ where
/// # Panics
///
/// Panics if the dimensions of the matrices involved are not compatible with the expression.
pub fn spmm_csc_prealloc<T>(
pub fn spmm_csc_prealloc_checked<T>(
beta: T,
c: &mut CscMatrix<T>,
alpha: T,
@ -83,35 +85,81 @@ where
{
assert_compatible_spmm_dims!(c, a, b);
use Op::{NoOp, Transpose};
use Op::NoOp;
match (&a, &b) {
(NoOp(ref a), NoOp(ref b)) => {
// Note: We have to reverse the order for CSC matrices
spmm_cs_prealloc(beta, &mut c.cs, alpha, &b.cs, &a.cs)
}
_ => {
// Currently we handle transposition by explicitly precomputing transposed matrices
// and calling the operation again without transposition
let a_ref: &CscMatrix<T> = a.inner_ref();
let b_ref: &CscMatrix<T> = b.inner_ref();
let (a, b) = {
use Cow::*;
match (&a, &b) {
(NoOp(_), NoOp(_)) => unreachable!(),
(Transpose(ref a), NoOp(_)) => (Owned(a.transpose()), Borrowed(b_ref)),
(NoOp(_), Transpose(ref b)) => (Borrowed(a_ref), Owned(b.transpose())),
(Transpose(ref a), Transpose(ref b)) => {
(Owned(a.transpose()), Owned(b.transpose()))
}
}
};
spmm_csc_prealloc(beta, c, alpha, NoOp(a.as_ref()), NoOp(b.as_ref()))
spmm_cs_prealloc_checked(beta, &mut c.cs, alpha, &b.cs, &a.cs)
}
_ => do_transposes(beta, c, alpha, a, b, spmm_csc_prealloc_checked),
}
}
/// Faster sparse-sparse matrix multiplication, `C <- beta * C + alpha * op(A) * op(B)`.
/// This will not return an error even if the patterns don't match.
/// Should be used for situations where pattern creation immediately preceeds multiplication.
///
/// Panics if the dimensions of the matrices involved are not compatible with the expression.
pub(crate) fn spmm_csc_prealloc_unchecked<T>(
beta: T,
c: &mut CscMatrix<T>,
alpha: T,
a: Op<&CscMatrix<T>>,
b: Op<&CscMatrix<T>>,
) -> Result<(), OperationError>
where
T: Scalar + ClosedAdd + ClosedMul + Zero + One,
{
assert_compatible_spmm_dims!(c, a, b);
use Op::NoOp;
match (&a, &b) {
(NoOp(ref a), NoOp(ref b)) => {
// Note: We have to reverse the order for CSC matrices
spmm_cs_prealloc_unchecked(beta, &mut c.cs, alpha, &b.cs, &a.cs)
}
_ => do_transposes(beta, c, alpha, a, b, spmm_csc_prealloc_unchecked),
}
}
fn do_transposes<T, F>(
beta: T,
c: &mut CscMatrix<T>,
alpha: T,
a: Op<&CscMatrix<T>>,
b: Op<&CscMatrix<T>>,
caller: F,
) -> Result<(), OperationError>
where
T: Scalar + ClosedAdd + ClosedMul + Zero + One,
F: Fn(
T,
&mut CscMatrix<T>,
T,
Op<&CscMatrix<T>>,
Op<&CscMatrix<T>>,
) -> Result<(), OperationError>,
{
use Op::{NoOp, Transpose};
// Currently we handle transposition by explicitly precomputing transposed matrices
// and calling the operation again without transposition
let a_ref: &CscMatrix<T> = a.inner_ref();
let b_ref: &CscMatrix<T> = b.inner_ref();
let (a, b) = {
use Cow::*;
match (&a, &b) {
(NoOp(_), NoOp(_)) => unreachable!(),
(Transpose(ref a), NoOp(_)) => (Owned(a.transpose()), Borrowed(b_ref)),
(NoOp(_), Transpose(ref b)) => (Borrowed(a_ref), Owned(b.transpose())),
(Transpose(ref a), Transpose(ref b)) => (Owned(a.transpose()), Owned(b.transpose())),
}
};
caller(beta, c, alpha, NoOp(a.as_ref()), NoOp(b.as_ref()))
}
/// Solve the lower triangular system `op(L) X = B`.
///
/// Only the lower triangular part of L is read, and the result is stored in B.

95
nalgebra-sparse/src/ops/serial/csr.rs
View File

@ -1,5 +1,7 @@
use crate::csr::CsrMatrix;
use crate::ops::serial::cs::{spadd_cs_prealloc, spmm_cs_dense, spmm_cs_prealloc};
use crate::ops::serial::cs::{
spadd_cs_prealloc, spmm_cs_dense, spmm_cs_prealloc_checked, spmm_cs_prealloc_unchecked,
};
use crate::ops::serial::OperationError;
use crate::ops::Op;
use nalgebra::{ClosedAdd, ClosedMul, DMatrixSlice, DMatrixSliceMut, Scalar};
@ -65,7 +67,7 @@ where
/// # Panics
///
/// Panics if the dimensions of the matrices involved are not compatible with the expression.
pub fn spmm_csr_prealloc<T>(
pub fn spmm_csr_prealloc_checked<T>(
beta: T,
c: &mut CsrMatrix<T>,
alpha: T,
@ -77,30 +79,75 @@ where
{
assert_compatible_spmm_dims!(c, a, b);
use Op::{NoOp, Transpose};
use Op::NoOp;
match (&a, &b) {
(NoOp(ref a), NoOp(ref b)) => spmm_cs_prealloc(beta, &mut c.cs, alpha, &a.cs, &b.cs),
_ => {
// Currently we handle transposition by explicitly precomputing transposed matrices
// and calling the operation again without transposition
// TODO: At least use workspaces to allow control of allocations. Maybe
// consider implementing certain patterns (like A^T * B) explicitly
let a_ref: &CsrMatrix<T> = a.inner_ref();
let b_ref: &CsrMatrix<T> = b.inner_ref();
let (a, b) = {
use Cow::*;
match (&a, &b) {
(NoOp(_), NoOp(_)) => unreachable!(),
(Transpose(ref a), NoOp(_)) => (Owned(a.transpose()), Borrowed(b_ref)),
(NoOp(_), Transpose(ref b)) => (Borrowed(a_ref), Owned(b.transpose())),
(Transpose(ref a), Transpose(ref b)) => {
(Owned(a.transpose()), Owned(b.transpose()))
}
}
};
spmm_csr_prealloc(beta, c, alpha, NoOp(a.as_ref()), NoOp(b.as_ref()))
(NoOp(ref a), NoOp(ref b)) => {
spmm_cs_prealloc_checked(beta, &mut c.cs, alpha, &a.cs, &b.cs)
}
_ => do_transposes(beta, c, alpha, a, b, spmm_csr_prealloc_checked),
}
}
/// Faster sparse-sparse matrix multiplication, `C <- beta * C + alpha * op(A) * op(B)`.
/// This will not return an error even if the patterns don't match.
/// Should be used for situations where pattern creation immediately preceeds multiplication.
///
/// Panics if the dimensions of the matrices involved are not compatible with the expression.
pub(crate) fn spmm_csr_prealloc_unchecked<T>(
beta: T,
c: &mut CsrMatrix<T>,
alpha: T,
a: Op<&CsrMatrix<T>>,
b: Op<&CsrMatrix<T>>,
) -> Result<(), OperationError>
where
T: Scalar + ClosedAdd + ClosedMul + Zero + One,
{
assert_compatible_spmm_dims!(c, a, b);
use Op::NoOp;
match (&a, &b) {
(NoOp(ref a), NoOp(ref b)) => {
spmm_cs_prealloc_unchecked(beta, &mut c.cs, alpha, &a.cs, &b.cs)
}
_ => do_transposes(beta, c, alpha, a, b, spmm_csr_prealloc_unchecked),
}
}
fn do_transposes<T, F>(
beta: T,
c: &mut CsrMatrix<T>,
alpha: T,
a: Op<&CsrMatrix<T>>,
b: Op<&CsrMatrix<T>>,
caller: F,
) -> Result<(), OperationError>
where
T: Scalar + ClosedAdd + ClosedMul + Zero + One,
F: Fn(
T,
&mut CsrMatrix<T>,
T,
Op<&CsrMatrix<T>>,
Op<&CsrMatrix<T>>,
) -> Result<(), OperationError>,
{
use Op::{NoOp, Transpose};
// Currently we handle transposition by explicitly precomputing transposed matrices
// and calling the operation again without transposition
let a_ref: &CsrMatrix<T> = a.inner_ref();
let b_ref: &CsrMatrix<T> = b.inner_ref();
let (a, b) = {
use Cow::*;
match (&a, &b) {
(NoOp(_), NoOp(_)) => unreachable!(),
(Transpose(ref a), NoOp(_)) => (Owned(a.transpose()), Borrowed(b_ref)),
(NoOp(_), Transpose(ref b)) => (Borrowed(a_ref), Owned(b.transpose())),
(Transpose(ref a), Transpose(ref b)) => (Owned(a.transpose()), Owned(b.transpose())),
}
};
caller(beta, c, alpha, NoOp(a.as_ref()), NoOp(b.as_ref()))
}