switch to babybear by default

.github/workflows/integration.yml

@@ -55,17 +55,16 @@ jobs:
           MOCK_PROVING: 1
         run: cargo run --package ceno_zkvm --features sanity-check --bin e2e -- --platform=ceno --hints=10 --public-io=4191 examples/target/riscv32im-ceno-zkvm-elf/debug/examples/fibonacci
 
-      - name: Run fibonacci (debug) + feature u16limb_circuit
+      - name: Run fibonacci (release)
         env:
-          RUST_LOG: debug
           RUSTFLAGS: "-C opt-level=3"
-        run: cargo run --package ceno_zkvm --features u16limb_circuit --bin e2e -- --platform=ceno --hints=10 --public-io=4191 examples/target/riscv32im-ceno-zkvm-elf/debug/examples/fibonacci
-
+        run: cargo run --release --package ceno_zkvm --bin e2e -- --platform=ceno --hints=10 --public-io=4191 examples/target/riscv32im-ceno-zkvm-elf/release/examples/fibonacci
 
-      - name: Run fibonacci (release)
+      - name: Run fibonacci (release + goldilocks)
         env:
           RUSTFLAGS: "-C opt-level=3"
-        run: cargo run --release --package ceno_zkvm --bin e2e -- --platform=ceno --hints=10 --public-io=4191 examples/target/riscv32im-ceno-zkvm-elf/release/examples/fibonacci
+        run: cargo run --release --package ceno_zkvm --no-default-features --features goldilocks --bin e2e -- --field=goldilocks --platform=ceno --hints=10 --public-io=4191 examples/target/riscv32im-ceno-zkvm-elf/release/examples/fibonacci
+
 
       - name: Install cargo make
         run: |

ceno_zkvm/Cargo.toml

@@ -59,9 +59,10 @@ ceno-examples = { path = "../examples-builder" }
 glob = "0.3"
 
 [features]
-default = ["forbid_overflow"]
+default = ["forbid_overflow", "nightly-features", "u16limb_circuit"]
 flamegraph = ["pprof2/flamegraph", "pprof2/criterion"]
 forbid_overflow = []
+goldilocks = ["forbid_overflow", "nightly-features"]
 jemalloc = ["dep:tikv-jemallocator", "dep:tikv-jemalloc-ctl"]
 jemalloc-prof = ["jemalloc", "tikv-jemallocator?/profiling"]
 nightly-features = [

ceno_zkvm/benches/fibonacci.rs

@@ -10,7 +10,7 @@ use ceno_zkvm::{
 mod alloc;
 use criterion::*;
 
-use ff_ext::GoldilocksExt2;
+use ff_ext::BabyBearExt4;
 use gkr_iop::cpu::{CpuBackend, CpuProver};
 use mpcs::BasefoldDefault;
 use transcript::BasicTranscript;
@@ -26,7 +26,7 @@ criterion_main!(fibonacci_prove_group);
 const NUM_SAMPLES: usize = 10;
 
 type Pcs = BasefoldDefault<E>;
-type E = GoldilocksExt2;
+type E = BabyBearExt4;
 
 // Relevant init data for fibonacci run
 fn setup() -> (Program, Platform) {

ceno_zkvm/benches/fibonacci_witness.rs

@@ -7,7 +7,7 @@ use ceno_zkvm::{
 use std::{fs, path::PathBuf, rc::Rc, time::Duration};
 mod alloc;
 use criterion::*;
-use ff_ext::GoldilocksExt2;
+use ff_ext::BabyBearExt4;
 use gkr_iop::cpu::{CpuBackend, CpuProver};
 use mpcs::BasefoldDefault;
 
@@ -21,7 +21,7 @@ criterion_main!(fibonacci);
 
 const NUM_SAMPLES: usize = 10;
 type Pcs = BasefoldDefault<E>;
-type E = GoldilocksExt2;
+type E = BabyBearExt4;
 
 // Relevant init data for fibonacci run
 fn setup() -> (Program, Platform) {
@@ -79,5 +79,5 @@ fn fibonacci_witness(c: &mut Criterion) {
 
     group.finish();
 
-    type E = GoldilocksExt2;
+    type E = BabyBearExt4;
 }

ceno_zkvm/benches/is_prime.rs

@@ -8,7 +8,7 @@ use ceno_zkvm::{
 };
 mod alloc;
 use criterion::*;
-use ff_ext::GoldilocksExt2;
+use ff_ext::BabyBearExt4;
 use gkr_iop::cpu::{CpuBackend, CpuProver};
 use mpcs::BasefoldDefault;
 
@@ -22,7 +22,7 @@ criterion_main!(is_prime);
 
 const NUM_SAMPLES: usize = 10;
 type Pcs = BasefoldDefault<E>;
-type E = GoldilocksExt2;
+type E = BabyBearExt4;
 
 // Relevant init data for fibonacci run
 fn setup() -> (Program, Platform) {
@@ -75,6 +75,4 @@ fn is_prime_1(c: &mut Criterion) {
 
         group.finish();
     }
-
-    type E = GoldilocksExt2;
 }

ceno_zkvm/benches/keccak.rs

@@ -9,7 +9,7 @@ use ceno_zkvm::{
 mod alloc;
 use criterion::*;
 
-use ff_ext::GoldilocksExt2;
+use ff_ext::BabyBearExt4;
 use gkr_iop::cpu::{CpuBackend, CpuProver};
 use mpcs::BasefoldDefault;
 
@@ -24,7 +24,7 @@ criterion_main!(keccak_prove_group);
 const NUM_SAMPLES: usize = 10;
 
 type Pcs = BasefoldDefault<E>;
-type E = GoldilocksExt2;
+type E = BabyBearExt4;
 
 // Relevant init data for keccak run
 fn setup() -> (Program, Platform) {

ceno_zkvm/benches/lookup_keccakf.rs

@@ -2,7 +2,7 @@ use std::time::Duration;
 
 use ceno_zkvm::precompiles::{run_faster_keccakf, setup_lookup_keccak_gkr_circuit};
 use criterion::*;
-use ff_ext::GoldilocksExt2;
+use ff_ext::BabyBearExt4;
 
 use itertools::Itertools;
 use mpcs::BasefoldDefault;
@@ -41,14 +41,13 @@ fn keccak_f_fn(c: &mut Criterion) {
                         let circuit =
                             setup_lookup_keccak_gkr_circuit().expect("setup circuit error");
                         #[allow(clippy::unit_arg)]
-                        let _ =
-                            run_faster_keccakf::<GoldilocksExt2, BasefoldDefault<GoldilocksExt2>>(
-                                circuit,
-                                black_box(states),
-                                false,
-                                false,
-                            )
-                            .expect("unable to get proof");
+                        let _ = run_faster_keccakf::<BabyBearExt4, BasefoldDefault<BabyBearExt4>>(
+                            circuit,
+                            black_box(states),
+                            false,
+                            false,
+                        )
+                        .expect("unable to get proof");
                         let elapsed = instant.elapsed();
                         println!(
                             "keccak_f::create_proof, instances = {}, time = {}",

ceno_zkvm/benches/quadratic_sorting.rs

@@ -8,7 +8,7 @@ use ceno_zkvm::{
 };
 mod alloc;
 use criterion::*;
-use ff_ext::GoldilocksExt2;
+use ff_ext::BabyBearExt4;
 use gkr_iop::cpu::{CpuBackend, CpuProver};
 use mpcs::BasefoldDefault;
 use rand::{RngCore, SeedableRng};
@@ -23,7 +23,7 @@ criterion_main!(quadratic_sorting);
 
 const NUM_SAMPLES: usize = 10;
 type Pcs = BasefoldDefault<E>;
-type E = GoldilocksExt2;
+type E = BabyBearExt4;
 
 // Relevant init data for fibonacci run
 fn setup() -> (Program, Platform) {
@@ -77,6 +77,4 @@ fn quadratic_sorting_1(c: &mut Criterion) {
 
         group.finish();
     }
-
-    type E = GoldilocksExt2;
 }

ceno_zkvm/benches/riscv_add.rs

@@ -10,10 +10,10 @@ mod alloc;
 use criterion::*;
 
 use ceno_zkvm::scheme::constants::MAX_NUM_VARIABLES;
-use ff_ext::GoldilocksExt2;
 use gkr_iop::cpu::{CpuBackend, CpuProver};
 use mpcs::{BasefoldDefault, PolynomialCommitmentScheme, SecurityLevel};
 
+use ff_ext::BabyBearExt4;
 use rand::rngs::OsRng;
 use transcript::{BasicTranscript, Transcript};
 use witness::RowMajorMatrix;
@@ -121,5 +121,5 @@ fn bench_add(c: &mut Criterion) {
         group.finish();
     }
 
-    type E = GoldilocksExt2;
+    type E = BabyBearExt4;
 }

ceno_zkvm/src/bin/e2e.rs

@@ -271,7 +271,7 @@ fn main() {
                 max_steps,
                 args.proof_file,
                 args.vk_file,
-                Checkpoint::PrepVerify, // FIXME: when whir and babybear is ready
+                Checkpoint::Complete,
             )
         }
         (PcsKind::Whir, FieldType::Goldilocks) => {

ceno_zkvm/src/e2e.rs

@@ -71,8 +71,8 @@ pub enum PcsKind {
 )]
 pub enum FieldType {
     #[default]
-    Goldilocks,
     BabyBear,
+    Goldilocks,
 }
 
 pub struct FullMemState<Record> {

ceno_zkvm/src/instructions/riscv/im_insn.rs

@@ -23,6 +23,7 @@ pub struct IMInstructionConfig<E: ExtensionField> {
 }
 
 impl<E: ExtensionField> IMInstructionConfig<E> {
+    #[allow(clippy::too_many_arguments)]
     pub fn construct_circuit(
         circuit_builder: &mut CircuitBuilder<E>,
         insn_kind: InsnKind,

ceno_zkvm/src/instructions/riscv/rv32im.rs

@@ -6,6 +6,8 @@ use super::{
 use crate::instructions::riscv::auipc::AuipcInstruction;
 #[cfg(feature = "u16limb_circuit")]
 use crate::instructions::riscv::lui::LuiInstruction;
+#[cfg(not(feature = "u16limb_circuit"))]
+use crate::tables::PowTableCircuit;
 use crate::{
     error::ZKVMError,
     instructions::{
@@ -28,9 +30,8 @@ use crate::{
     },
     structs::{ZKVMConstraintSystem, ZKVMFixedTraces, ZKVMWitnesses},
     tables::{
-        AndTableCircuit, LtuTableCircuit, OrTableCircuit, PowTableCircuit, TableCircuit,
-        U5TableCircuit, U8TableCircuit, U14TableCircuit, U16TableCircuit, U18TableCircuit,
-        XorTableCircuit,
+        AndTableCircuit, LtuTableCircuit, OrTableCircuit, TableCircuit, U5TableCircuit,
+        U8TableCircuit, U14TableCircuit, U16TableCircuit, U18TableCircuit, XorTableCircuit,
     },
 };
 use ceno_emul::{
@@ -127,6 +128,7 @@ pub struct Rv32imConfig<E: ExtensionField> {
     pub or_table_config: <OrTableCircuit<E> as TableCircuit<E>>::TableConfig,
     pub xor_table_config: <XorTableCircuit<E> as TableCircuit<E>>::TableConfig,
     pub ltu_config: <LtuTableCircuit<E> as TableCircuit<E>>::TableConfig,
+    #[cfg(not(feature = "u16limb_circuit"))]
     pub pow_config: <PowTableCircuit<E> as TableCircuit<E>>::TableConfig,
 }
 
@@ -204,6 +206,7 @@ impl<E: ExtensionField> Rv32imConfig<E> {
         let or_table_config = cs.register_table_circuit::<OrTableCircuit<E>>();
         let xor_table_config = cs.register_table_circuit::<XorTableCircuit<E>>();
         let ltu_config = cs.register_table_circuit::<LtuTableCircuit<E>>();
+        #[cfg(not(feature = "u16limb_circuit"))]
         let pow_config = cs.register_table_circuit::<PowTableCircuit<E>>();
 
         Self {
@@ -272,6 +275,7 @@ impl<E: ExtensionField> Rv32imConfig<E> {
             or_table_config,
             xor_table_config,
             ltu_config,
+            #[cfg(not(feature = "u16limb_circuit"))]
             pow_config,
         }
     }
@@ -350,6 +354,7 @@ impl<E: ExtensionField> Rv32imConfig<E> {
         fixed.register_table_circuit::<OrTableCircuit<E>>(cs, &self.or_table_config, &());
         fixed.register_table_circuit::<XorTableCircuit<E>>(cs, &self.xor_table_config, &());
         fixed.register_table_circuit::<LtuTableCircuit<E>>(cs, &self.ltu_config, &());
+        #[cfg(not(feature = "u16limb_circuit"))]
         fixed.register_table_circuit::<PowTableCircuit<E>>(cs, &self.pow_config, &());
     }
 
@@ -480,6 +485,7 @@ impl<E: ExtensionField> Rv32imConfig<E> {
         witness.assign_table_circuit::<OrTableCircuit<E>>(cs, &self.or_table_config, &())?;
         witness.assign_table_circuit::<XorTableCircuit<E>>(cs, &self.xor_table_config, &())?;
         witness.assign_table_circuit::<LtuTableCircuit<E>>(cs, &self.ltu_config, &())?;
+        #[cfg(not(feature = "u16limb_circuit"))]
         witness.assign_table_circuit::<PowTableCircuit<E>>(cs, &self.pow_config, &())?;
 
         Ok(())

gkr_iop/src/cpu/mod.rs

@@ -35,7 +35,7 @@ impl<E: ExtensionField, PCS: PolynomialCommitmentScheme<E>> Default for CpuBacke
 
 impl<E: ExtensionField, PCS: PolynomialCommitmentScheme<E>> CpuBackend<E, PCS> {
     pub fn new(max_poly_size_log2: usize, security_level: SecurityLevel) -> Self {
-        let param = PCS::setup(E::BaseField::TWO_ADICITY, security_level).unwrap();
+        let param = PCS::setup(1 << E::BaseField::TWO_ADICITY, security_level).unwrap();
         let (pp, vp) = PCS::trim(param, 1 << max_poly_size_log2).unwrap();
         Self {
             pp,