Skip to content
This repository has been archived by the owner on Jul 5, 2024. It is now read-only.

Commit

Permalink
feat: use peak memory estimation from Han
Browse files Browse the repository at this point in the history
  • Loading branch information
ed255 committed Feb 9, 2024
1 parent 627afbc commit c391b0e
Show file tree
Hide file tree
Showing 2 changed files with 394 additions and 172 deletions.
373 changes: 373 additions & 0 deletions zkevm-circuits/src/bin/stats/halo2_stats.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,373 @@
//! Utility functions and types to get circuit stats from any halo2 circuit

use eth_types::Field;
use halo2_proofs::plonk::ConstraintSystem;
use std::collections::BTreeSet;

// From Scroll https://github.com/scroll-tech/zkevm-circuits/blob/7d9bc181953cfc6e7baf82ff0ce651281fd70a8a/zkevm-circuits/src/util.rs#L275
#[allow(dead_code)]
#[derive(Debug, Default)]
pub(crate) struct CircuitStats {
pub num_constraints: usize,
pub num_fixed_columns: usize,
pub num_lookups: usize,
pub num_shuffles: usize,
pub num_advice_columns: usize,
pub num_instance_columns: usize,
pub num_selectors: usize,
// num_simple_selectors: usize,
pub num_permutation_columns: usize,
pub degree: usize,
pub blinding_factors: usize,
pub num_challenges: usize,
pub max_phase: u8,
pub num_rotation: usize,
pub min_rotation: i32,
pub max_rotation: i32,
pub num_verification_ecmul: usize,
// Aux data to diff between records
num_advice_queries: usize,
num_gates: usize,
}

impl CircuitStats {
// Peak memory analysis by Han:
//
// fn create_proof(params: &KZGParams, pk: &ProvingKey, circuit: &Circuit, instances: &[&[F]]) {
// Let:
//
// - k: log 2 of number of rows
// - n: `1 << k`
// - d: Degree of circuit
// - e: Extension magnitude, equal to `(d - 1).next_power_of_two()`
// - c_f: number of fixed columns
// - c_a: number of advice columns
// - c_i: number of instance columns
// - c_p: number of columns enabled with copy constraint
// - c_pg: number of grand product in permutation argument, equal to `div_ceil(c_p, d - 2)`
// - c_l: number of lookup argument
//
// The memory usage M.C and M.S stands for:
//
// - M.C: number of "n elliptic curve points" (with symbol ◯)
// - M.S: number of "n field elements" (with symbol △)
// - M.E: number of "e * n field elements" (with symbol ⬡)
//
// So the actual memory usage in terms of bytes will be:
//
// M = 32 * n * (2 * M.C + M.S + e * M.E)
//
// We'll ignore other values with sublinear amount to n.
//
//
// 0. In the beginning:
//
// `params` has:
// ◯ powers_of_tau
// ◯ ifft(powers_of_tau)
//
// M.C = 2 (+= 2)
// M.S = 0
// M.E = 0
//
// `pk` has:
// ⬡ l0
// ⬡ l_last
// ⬡ l_active_row
// △ fixed_lagranges (c_f)
// △ fixed_monomials (c_f)
// ⬡ fixed_extended_lagranges (c_f)
// △ permutation_lagranges (c_p)
// △ permutation_monomials (c_p)
// ⬡ permutation_extended_lagranges (c_p)
//
// M.C = 2
// M.S = 2 * c_f + 2 * c_p (+= 2 * c_f + 2 * c_p)
// M.E = 3 + c_f + c_p (+= 3 + c_f + c_p)
//
// And let's ignore `circuit`
//
//
// ### 1. Pad instances as lagrange form and compute its monomial form.
//
// M.C = 2
// M.S = 2 * c_f + 2 * c_p + 2 * c_i (+= 2 * c_i)
// M.E = 3 + c_f + c_p
// ```
// let instance_lagranges = instances.to_lagranges();
// let instance_monomials = instance_lagranges.to_monomials();
// ```
//
//
// ### 2. Synthesize circuit and collect advice column values.
//
// M.C = 2
// M.S = 2 * c_f + 2 * c_p + 2 * c_i + c_a (+= c_a)
// M.E = 3 + c_f + c_p
// ```
// let advice_lagranges = circuit.synthesize_all_phases();
// ```
//
//
// ### 3. Generate permuted input and table of lookup argument.
// For each lookup argument, we have:
//
// △ compressed_input_lagranges - cached for later computation
// △ permuted_input_lagranges
// △ permuted_input_monomials
// △ compressed_table_lagranges - cached for later computation
// △ permuted_table_lagranges
// △ permuted_table_monomials
//
// M.C = 2
// M.S = 2 * c_f + 2 * c_p + 2 * c_i + c_a + 6 * c_l (+= 6 * c_l)
// M.E = 3 + c_f + c_p
// ```
// let (
// compressed_input_lagranges,
// permuted_input_lagranges,
// permuted_input_monomials,
// compressed_table_lagranges,
// permuted_table_lagranges,
// permuted_table_monomials,
// ) = lookup_permuted()
// ```
//
//
// ### 4. Generate grand products of permutation argument.
//
// M.C = 2
// M.S = 2 * c_f + 2 * c_p + 2 * c_i + c_a + 6 * c_l + c_pg (+= c_pg)
// M.E = 3 + c_f + c_p + c_pg (+= c_pg)
// ```
// let (
// perm_grand_product_monomials,
// perm_grand_product_extended_lagranges,
// ) = permutation_grand_products();
// ```
//
//
// ### 5. Generate grand products of lookup argument.
// And then drops unnecessary lagranges values.
//
// M.C = 2
// M.S = 2 * c_f + 2 * c_p + 2 * c_i + c_a + 3 * c_l + c_pg (-= 3 * c_l)
// M.E = 3 + c_f + c_p + c_pg
// > let lookup_product_monomials = lookup_grand_products();
// > drop(compressed_input_lagranges);
// > drop(permuted_input_lagranges);
// > drop(compressed_table_lagranges);
// > drop(permuted_table_lagranges);
//
//
// ### 6. Generate random polynomial.
//
// M.C = 2
// M.S = 1 + 2 * c_f + 2 * c_p + 2 * c_i + c_a + 3 * c_l + c_pg (+= 1)
// M.E = 3 + c_f + c_p + c_pg
// ```
// let random_monomial = random();
// ```
//
//
// ### 7. Turn advice_lagranges into advice_monomials.
// ```
// let advice_monomials = advice_lagranges.to_monomials();
// drop(advice_lagranges);
// ```
//
//
// ### 8. Generate necessary extended lagranges.
//
// M.C = 2
// M.S = 1 + 2 * c_f + 2 * c_p + 2 * c_i + c_a + 3 * c_l + c_pg
// M.E = 3 + c_f + c_p + c_pg + c_i + c_a (+= c_i + c_a)
// ```
// let instances_extended_lagrnages = instances_monomials.to_extended_lagranges();
// let advice_extended_lagrnages = advice_monomials.to_extended_lagranges();
// ```
//
//
// ### 9. While computing the quotient, these extended lagranges:
//
// ⬡ permuted_input_extended_lagranges
// ⬡ permuted_table_extended_lagranges
// ⬡ lookup_product_extended_lagranges
//
// of each lookup argument are generated on the fly and drop before next.
//
// And 1 extra quotient_extended_lagrange is created. So the peak memory:
//
// M.C = 2
// M.S = 1 + 2 * c_f + 2 * c_p + 2 * c_i + c_a + 3 * c_l + c_pg
// M.E = 4 + c_f + c_p + c_pg + c_i + c_a + 3 * (c_l > 0) (+= 3 * (c_l > 0) + 1)
// ```
// let quotient_extended_lagrange = quotient_extended_lagrange();
// ```
//
//
// ### 10. After quotient is comuputed, drop all the other extended lagranges.
//
// M.C = 2
// M.S = 1 + 2 * c_f + 2 * c_p + 2 * c_i + c_a + 3 * c_l + c_pg
// M.E = 4 + c_f + c_p (-= c_pg + c_i + c_a + 3 * (c_l > 0))
// drop(instances_extended_lagrnages)
// drop(advice_extended_lagrnages)
// drop(perm_grand_product_extended_lagranges)
//
//
// ### 11. Turn quotient_extended_lagrange into monomial form.
// And then cut int `d - 1` pieces.
//
// M.C = 2
// M.S = 2 * c_f + 2 * c_p + 2 * c_i + c_a + 3 * c_l + c_pg + d (+= d - 1)
// M.E = 3 + c_f + c_p (-= 1)
// ```
// let quotient_monomials = quotient_monomials()
// drop(quotient_extended_lagrange)
// ```
//
//
// ### 12. Evaluate and open all polynomial except instance ones.
// }
pub(crate) fn estimate_peak_mem(&self, k: u32) -> usize {
let field_bytes = 32;
let c_f = self.num_fixed_columns;
let c_a = self.num_advice_columns;
let c_i = self.num_instance_columns;
let c_p = self.num_permutation_columns;
let c_l = self.num_lookups;
let c_pg = (c_p + self.degree - 3) / (self.degree - 2);
let e = (self.degree - 1).next_power_of_two();
// number of "n elliptic curve points"
let m_c = 2;
// number of "n field elements"
let m_s = 1 + 2 * c_f + 2 * c_p + 2 * c_i + c_a + 3 * c_l + c_pg;
// number of "e * n field elements"
let m_e = 4 + c_f + c_p + c_pg + c_i + c_a + 3 * (c_l > 0) as usize;
let unit = m_c + m_s + e * m_e;
unit * 2usize.pow(k) * field_bytes
}
}

// Return the stats in `meta`, accounting only for the circuit delta from the last aggregated stats
// in `agg`.
// Adaptaed from Scroll https://github.com/scroll-tech/zkevm-circuits/blob/7d9bc181953cfc6e7baf82ff0ce651281fd70a8a/zkevm-circuits/src/util.rs#L294
pub(crate) fn circuit_stats<F: Field>(
agg: &CircuitStats,
meta: &ConstraintSystem<F>,
) -> CircuitStats {
let max_phase = meta
.advice_column_phase()
.iter()
.skip(agg.num_advice_columns)
.max()
.copied()
.unwrap_or_default();

let rotations = meta
.advice_queries()
.iter()
.skip(agg.num_advice_queries)
.map(|(_, q)| q.0)
.collect::<BTreeSet<i32>>();

let num_fixed_columns = meta.num_fixed_columns() - agg.num_fixed_columns;
let num_lookups = meta.lookups().len() - agg.num_lookups;
let num_shuffles = meta.shuffles().len() - agg.num_shuffles;
let num_advice_columns = meta.num_advice_columns() - agg.num_advice_columns;
let num_instance_columns = meta.num_instance_columns() - agg.num_instance_columns;
let num_selectors = meta.num_selectors() - agg.num_selectors;
let num_permutation_columns =
meta.permutation().get_columns().len() - agg.num_permutation_columns;

CircuitStats {
num_constraints: meta
.gates()
.iter()
.skip(agg.num_gates)
.map(|g| g.polynomials().len())
.sum::<usize>(),
num_fixed_columns,
num_lookups,
num_shuffles,
num_advice_columns,
num_instance_columns,
num_selectors,
// num_simple_selectors: meta.num_simple_selectors(),
num_permutation_columns,
degree: meta.degree(),
blinding_factors: meta.blinding_factors(),
num_challenges: meta.num_challenges() - agg.num_challenges,
max_phase,
num_rotation: rotations.len(),
min_rotation: rotations.first().cloned().unwrap_or_default(),
max_rotation: rotations.last().cloned().unwrap_or_default(),
num_verification_ecmul: num_advice_columns
+ num_instance_columns
+ num_permutation_columns
+ num_shuffles
+ num_selectors
+ num_fixed_columns
+ 3 * num_lookups
+ rotations.len(),
num_advice_queries: meta.advice_queries().len() - agg.num_advice_queries,
num_gates: meta.gates().len() - agg.num_gates,
}
}

pub(crate) struct StatsCollection<F: Field> {
aggregate: bool,
shared_cs: ConstraintSystem<F>,
pub(crate) agg: CircuitStats,
pub(crate) list: Vec<(String, CircuitStats)>,
}

impl<F: Field> StatsCollection<F> {
// With aggregate=true, all records are overwritten each time, leading to a single
// aggregate stats that represents the final circuit.
// With aggregate=false, each record is stored in a different entry with a name, and the
// ConstraintSystem is reset so that each entry is independent.
pub(crate) fn new(aggregate: bool) -> Self {
Self {
aggregate,
shared_cs: ConstraintSystem::default(),
agg: CircuitStats::default(),
list: Vec::new(),
}
}

// Record a shared table
pub(crate) fn record_shared(&mut self, name: &str, meta: &mut ConstraintSystem<F>) {
// Shared tables should only add columns, and nothing more
assert_eq!(meta.lookups().len(), 0);
assert_eq!(meta.shuffles().len(), 0);
assert_eq!(meta.permutation().get_columns().len(), 0);
assert_eq!(meta.degree(), 3); // 3 comes from the permutation argument
assert_eq!(meta.blinding_factors(), 5); // 5 is the minimum blinding factor
assert_eq!(meta.advice_queries().len(), 0);
assert_eq!(meta.gates().len(), 0);

if self.aggregate {
self.agg = circuit_stats(&CircuitStats::default(), meta);
} else {
let stats = circuit_stats(&self.agg, meta);
self.agg = circuit_stats(&CircuitStats::default(), meta);
self.list.push((name.to_string(), stats));
// Keep the ConstraintSystem with all the tables
self.shared_cs = meta.clone();
}
}

// Record a subcircuit
pub(crate) fn record(&mut self, name: &str, meta: &mut ConstraintSystem<F>) {
if self.aggregate {
self.agg = circuit_stats(&CircuitStats::default(), meta);
} else {
let stats = circuit_stats(&self.agg, meta);
self.list.push((name.to_string(), stats));
// Revert meta to the ConstraintSystem just with the tables
*meta = self.shared_cs.clone();
}
}
}
Loading

0 comments on commit c391b0e

Please sign in to comment.