Skip to content

Enable parallel only on non-zkVM targets in pcs, executor, and machine #584

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
eigmax merged 4 commits into from
Nov 2, 2026
Merged

Enable parallel only on non-zkVM targets in pcs, executor, and machine #584

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
8a21b0a
Enable parallel only on non-zkVM targets in pcs, executor, and machine
Nov 2, 2026
1a4d24f
Remove duplicate target-specific feature activation from executor and
Nov 2, 2026
1673550
refactor: streamline parallel processing in prove and jagged modules
Nov 2, 2026
de270ad
doc: update comments for compute_global_cumulative_sum
Nov 2, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion crates/core/executor/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ zkm-core-jit = { path = "../jit" }

# p3
p3-field = { workspace = true }
p3-maybe-rayon = { workspace = true, features = ["parallel"] }
p3-maybe-rayon = { workspace = true }
p3-koala-bear = { workspace = true }
p3-symmetric = { workspace = true }

Expand Down
18 changes: 9 additions & 9 deletions crates/core/executor/src/program.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,7 @@ use elf::{endian::LittleEndian, file::Class, ElfBytes};

use p3_field::BasedVectorSpace;
use p3_field::PrimeField32;
use p3_maybe_rayon::prelude::IndexedParallelIterator;
use p3_maybe_rayon::prelude::IntoParallelIterator;
use p3_maybe_rayon::prelude::IntoParallelRefIterator;
use p3_maybe_rayon::prelude::{ParallelBridge, ParallelIterator};
use p3_maybe_rayon::prelude::*;
use serde::{Deserialize, Serialize};
use zkm_pcs::air::MachineProgram;
use zkm_pcs::septic_curve::{SepticCurve, SepticCurveComplete};
Expand Down Expand Up @@ -382,10 +379,10 @@ impl<F: PrimeField32> MachineProgram<F> for Program {
}

fn initial_global_cumulative_sum(&self) -> SepticDigest<F> {
let mut digests: Vec<SepticCurveComplete<F>> = self
.image
.iter()
.par_bridge()
let image = self.image.iter();
#[cfg(not(target_os = "zkvm"))]
let image = image.par_bridge();
let mut digests: Vec<SepticCurveComplete<F>> = image
.map(|(&addr, &word)| {
let values = [
(LookupKind::Memory as u32) << 16,
Expand All @@ -404,7 +401,10 @@ impl<F: PrimeField32> MachineProgram<F> for Program {
.collect();
digests.push(SepticCurveComplete::Affine(SepticDigest::<F>::zero().0));
SepticDigest(
digests.into_par_iter().reduce(|| SepticCurveComplete::Infinity, |a, b| a + b).point(),
digests
.into_par_iter()
.par_fold_reduce(|| SepticCurveComplete::Infinity, |a, b| a + b, |a, b| a + b)
.point(),
)
}
}
Expand Down
6 changes: 2 additions & 4 deletions crates/core/executor/src/record.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -749,9 +749,7 @@ impl MachineRecord for ExecutionRecord {
fn compute_global_cumulative_sum(events: &[GlobalLookupEvent]) -> SepticDigest<u32> {
use p3_field::BasedVectorSpace;
use p3_koala_bear::KoalaBear;
use p3_maybe_rayon::prelude::{
IndexedParallelIterator, IntoParallelRefIterator, ParallelIterator,
};
use p3_maybe_rayon::prelude::*;
type F = KoalaBear;

let acc = events
Expand All @@ -765,7 +763,7 @@ fn compute_global_cumulative_sum(events: &[GlobalLookupEvent]) -> SepticDigest<u
let point = if event.is_receive { point } else { point.neg() };
SepticCurveComplete::Affine(point)
})
.reduce(|| SepticCurveComplete::Infinity, |a, b| a + b);
.par_fold_reduce(|| SepticCurveComplete::Infinity, |a, b| a + b, |a, b| a + b);
let acc = SepticCurveComplete::Affine(SepticDigest::<F>::zero().0) + acc;
let final_digest = acc.point();
SepticDigest(SepticCurve::convert(final_digest, |x: F| x.as_canonical_u32()))
Expand Down
2 changes: 1 addition & 1 deletion crates/core/machine/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,7 @@ p3-commit = { workspace = true }
p3-field = { workspace = true }
p3-keccak-air = { workspace = true }
p3-matrix = { workspace = true }
p3-maybe-rayon = { workspace = true, features = ["parallel"] }
p3-maybe-rayon = { workspace = true }
p3-poseidon2 = { workspace = true }
p3-uni-stark = { workspace = true }
p3-util = { workspace = true }
Expand Down
43 changes: 20 additions & 23 deletions crates/core/machine/src/alu/add_sub/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ use itertools::Itertools;
use p3_air::{Air, AirBuilder, BaseAir, WindowAccess};
use p3_field::{Field, PrimeCharacteristicRing, PrimeField32};
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{ParallelBridge, ParallelIterator};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::{
events::{AluEvent, ByteRecord},
ExecutionRecord, Opcode, Program,
Expand Down Expand Up @@ -100,26 +100,24 @@ impl<F: PrimeField32> MachineAir<F> for AddSubChip {
let padded_nb_rows = <AddSubChip as MachineAir<F>>::num_rows(self, input).unwrap();
let mut values = zeroed_f_vec(padded_nb_rows * NUM_ADD_SUB_COLS);

values.chunks_mut(chunk_size * NUM_ADD_SUB_COLS).enumerate().par_bridge().for_each(
|(i, rows)| {
rows.chunks_mut(NUM_ADD_SUB_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut AddSubCols<F> = row.borrow_mut();

if idx < input.add_sub_events.len() {
let mut byte_lookup_events = Vec::new();
let event = &input.add_sub_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
},
);
values.par_chunks_mut(chunk_size * NUM_ADD_SUB_COLS).enumerate().for_each(|(i, rows)| {
rows.chunks_mut(NUM_ADD_SUB_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut AddSubCols<F> = row.borrow_mut();

if idx < input.add_sub_events.len() {
let mut byte_lookup_events = Vec::new();
let event = &input.add_sub_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
});

Ok(RowMajorMatrix::new(values, NUM_ADD_SUB_COLS))
}
Expand All @@ -133,8 +131,7 @@ impl<F: PrimeField32> MachineAir<F> for AddSubChip {

let blu_batches = input
.add_sub_events
.chunks(chunk_size)
.par_bridge()
.par_chunks(chunk_size)
.map(|events| {
let mut blu: zkm_core_executor::events::ByteLookupMap = Default::default();
events.iter().for_each(|event| {
Expand Down
7 changes: 3 additions & 4 deletions crates/core/machine/src/alu/add_sub_imm/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ use itertools::Itertools;
use p3_air::{Air, AirBuilder, BaseAir, WindowAccess};
use p3_field::{Field, PrimeCharacteristicRing, PrimeField32};
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{ParallelBridge, ParallelIterator};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::{
events::{AluEvent, ByteRecord},
ExecutionRecord, Opcode, Program,
Expand Down Expand Up @@ -99,7 +99,7 @@ impl<F: PrimeField32> MachineAir<F> for AddSubImmChip {
let padded_nb_rows = <AddSubImmChip as MachineAir<F>>::num_rows(self, input).unwrap();
let mut values = zeroed_f_vec(padded_nb_rows * NUM_ADD_SUB_IMM_COLS);

values.chunks_mut(chunk_size * NUM_ADD_SUB_IMM_COLS).enumerate().par_bridge().for_each(
values.par_chunks_mut(chunk_size * NUM_ADD_SUB_IMM_COLS).enumerate().for_each(
|(i, rows)| {
rows.chunks_mut(NUM_ADD_SUB_IMM_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
Expand Down Expand Up @@ -132,8 +132,7 @@ impl<F: PrimeField32> MachineAir<F> for AddSubImmChip {

let blu_batches = input
.add_sub_imm_events
.chunks(chunk_size)
.par_bridge()
.par_chunks(chunk_size)
.map(|events| {
let mut blu: zkm_core_executor::events::ByteLookupMap = Default::default();
events.iter().for_each(|event| {
Expand Down
38 changes: 18 additions & 20 deletions crates/core/machine/src/alu/lt/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -116,26 +116,24 @@ impl<F: PrimeField32> MachineAir<F> for LtChip {
let mut values = zeroed_f_vec(padded_nb_rows * NUM_LT_COLS);
let chunk_size = std::cmp::max((input.lt_events.len() + 1) / num_cpus::get(), 1);

values.chunks_mut(chunk_size * NUM_LT_COLS).enumerate().par_bridge().for_each(
|(i, rows)| {
rows.chunks_mut(NUM_LT_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut LtCols<F> = row.borrow_mut();

if idx < input.lt_events.len() {
let mut byte_lookup_events = Vec::new();
let event = &input.lt_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
},
);
values.par_chunks_mut(chunk_size * NUM_LT_COLS).enumerate().for_each(|(i, rows)| {
rows.chunks_mut(NUM_LT_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut LtCols<F> = row.borrow_mut();

if idx < input.lt_events.len() {
let mut byte_lookup_events = Vec::new();
let event = &input.lt_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
});

Ok(RowMajorMatrix::new(values, NUM_LT_COLS))
}
Expand Down
38 changes: 18 additions & 20 deletions crates/core/machine/src/alu/lt_imm/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -116,26 +116,24 @@ impl<F: PrimeField32> MachineAir<F> for LtImmChip {
let mut values = zeroed_f_vec(padded_nb_rows * NUM_LT_IMM_COLS);
let chunk_size = std::cmp::max((input.lt_imm_events.len() + 1) / num_cpus::get(), 1);

values.chunks_mut(chunk_size * NUM_LT_IMM_COLS).enumerate().par_bridge().for_each(
|(i, rows)| {
rows.chunks_mut(NUM_LT_IMM_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut LtImmCols<F> = row.borrow_mut();

if idx < input.lt_imm_events.len() {
let mut byte_lookup_events = Vec::new();
let event = &input.lt_imm_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
},
);
values.par_chunks_mut(chunk_size * NUM_LT_IMM_COLS).enumerate().for_each(|(i, rows)| {
rows.chunks_mut(NUM_LT_IMM_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut LtImmCols<F> = row.borrow_mut();

if idx < input.lt_imm_events.len() {
let mut byte_lookup_events = Vec::new();
let event = &input.lt_imm_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
});

Ok(RowMajorMatrix::new(values, NUM_LT_IMM_COLS))
}
Expand Down
38 changes: 18 additions & 20 deletions crates/core/machine/src/alu/mul/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,7 @@ use zkm_pcs::air::BaseAirBuilder;
use p3_air::{Air, AirBuilder, BaseAir, WindowAccess};
use p3_field::{PrimeCharacteristicRing, PrimeField32};
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{ParallelBridge, ParallelIterator, ParallelSlice};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::{
events::{ByteLookupEvent, ByteRecord, CompAluEvent, MemoryAccessPosition, MemoryRecordEnum},
ByteOpcode, ExecutionRecord, Opcode, Program,
Expand Down Expand Up @@ -166,26 +166,24 @@ impl<F: PrimeField32> MachineAir<F> for MulChip {
let nb_rows = input.mul_events.len();
let chunk_size = std::cmp::max((nb_rows + 1) / num_cpus::get(), 1);

values.chunks_mut(chunk_size * NUM_MUL_COLS).enumerate().par_bridge().for_each(
|(i, rows)| {
rows.chunks_mut(NUM_MUL_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut MulCols<F> = row.borrow_mut();
values.par_chunks_mut(chunk_size * NUM_MUL_COLS).enumerate().for_each(|(i, rows)| {
rows.chunks_mut(NUM_MUL_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
let cols: &mut MulCols<F> = row.borrow_mut();

if idx < nb_rows {
let mut byte_lookup_events = Vec::new();
let event = &input.mul_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
},
);
if idx < nb_rows {
let mut byte_lookup_events = Vec::new();
let event = &input.mul_events[idx];
self.event_to_row(
event,
cols,
&mut byte_lookup_events,
&input.program,
input.public_values.execution_shard,
);
}
});
});

Ok(RowMajorMatrix::new(values, NUM_MUL_COLS))
}
Expand Down
4 changes: 2 additions & 2 deletions crates/core/machine/src/alu/sr/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,7 +52,7 @@ use itertools::Itertools;
use p3_air::{Air, AirBuilder, BaseAir, WindowAccess};
use p3_field::{PrimeCharacteristicRing, PrimeField32};
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{ParallelBridge, ParallelIterator, ParallelSlice};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::{
events::{AluEvent, ByteLookupEvent, ByteRecord},
ByteOpcode, ExecutionRecord, Opcode, Program,
Expand Down Expand Up @@ -171,7 +171,7 @@ impl<F: PrimeField32> MachineAir<F> for ShiftRightChip {
let mut values = zeroed_f_vec(padded_nb_rows * NUM_SHIFT_RIGHT_COLS);
let chunk_size = std::cmp::max((nb_rows + 1) / num_cpus::get(), 1);

values.chunks_mut(chunk_size * NUM_SHIFT_RIGHT_COLS).enumerate().par_bridge().for_each(
values.par_chunks_mut(chunk_size * NUM_SHIFT_RIGHT_COLS).enumerate().for_each(
|(i, rows)| {
rows.chunks_mut(NUM_SHIFT_RIGHT_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
Expand Down
4 changes: 2 additions & 2 deletions crates/core/machine/src/alu/sr_imm/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,7 +50,7 @@ use itertools::Itertools;
use p3_air::{Air, AirBuilder, BaseAir, WindowAccess};
use p3_field::{PrimeCharacteristicRing, PrimeField32};
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{ParallelBridge, ParallelIterator, ParallelSlice};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::{
events::{AluEvent, ByteLookupEvent, ByteRecord},
ByteOpcode, ExecutionRecord, Opcode, Program,
Expand Down Expand Up @@ -169,7 +169,7 @@ impl<F: PrimeField32> MachineAir<F> for ShiftRightImmChip {
let mut values = zeroed_f_vec(padded_nb_rows * NUM_SHIFT_RIGHT_IMM_COLS);
let chunk_size = std::cmp::max((nb_rows + 1) / num_cpus::get(), 1);

values.chunks_mut(chunk_size * NUM_SHIFT_RIGHT_IMM_COLS).enumerate().par_bridge().for_each(
values.par_chunks_mut(chunk_size * NUM_SHIFT_RIGHT_IMM_COLS).enumerate().for_each(
|(i, rows)| {
rows.chunks_mut(NUM_SHIFT_RIGHT_IMM_COLS).enumerate().for_each(|(j, row)| {
let idx = i * chunk_size + j;
Expand Down
2 changes: 1 addition & 1 deletion crates/core/machine/src/memory/bump.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ use p3_air::{Air, BaseAir, WindowAccess};
use p3_field::PrimeCharacteristicRing;
use p3_field::PrimeField32;
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{IndexedParallelIterator, ParallelIterator, ParallelSliceMut};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::events::{ByteLookupEvent, ByteRecord, MemoryBumpEvent, MemoryReadRecord};
use zkm_core_executor::{ByteOpcode, ExecutionRecord, Program, NUM_REGISTERS};
use zkm_derive::AlignedBorrow;
Expand Down
4 changes: 1 addition & 3 deletions crates/core/machine/src/memory/local.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,9 +9,7 @@ use p3_air::{Air, AirBuilder, BaseAir, WindowAccess};
use p3_field::PrimeCharacteristicRing;
use p3_field::PrimeField32;
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{
IndexedParallelIterator, IntoParallelRefMutIterator, ParallelIterator,
};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::events::{ByteRecord, GlobalLookupEvent, MemoryLocalEvent};
use zkm_core_executor::{ExecutionRecord, Program};
use zkm_derive::AlignedBorrow;
Expand Down
5 changes: 2 additions & 3 deletions crates/core/machine/src/misc/mov_cond/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ use itertools::Itertools;
use p3_air::{Air, AirBuilder, BaseAir, WindowAccess};
use p3_field::{PrimeCharacteristicRing, PrimeField32};
use p3_matrix::dense::RowMajorMatrix;
use p3_maybe_rayon::prelude::{ParallelBridge, ParallelIterator};
use p3_maybe_rayon::prelude::*;
use zkm_core_executor::{
events::{ByteRecord, MovCondEvent},
ExecutionRecord, Opcode, Program,
Expand Down Expand Up @@ -104,9 +104,8 @@ impl<F: PrimeField32> MachineAir<F> for MovCondChip {
let mut values = zeroed_f_vec(padded_nb_rows * NUM_MOV_COND_COLS);

let blu_events = values
.chunks_mut(chunk_size * NUM_MOV_COND_COLS)
.par_chunks_mut(chunk_size * NUM_MOV_COND_COLS)
.enumerate()
.par_bridge()
.map(|(i, rows)| {
let mut blu: zkm_core_executor::events::ByteLookupMap = Default::default();
rows.chunks_mut(NUM_MOV_COND_COLS).enumerate().for_each(|(j, row)| {
Expand Down
Loading
Loading