diff --git a/Cargo.toml b/Cargo.toml
index 761c6c5..43d1f21 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -6,7 +6,8 @@ homepage = "https://github.com/matterinc/bellman"
 license = "MIT/Apache-2.0"
 name = "bellman"
 repository = "https://github.com/matterinc/bellman"
-version = "0.1.3"
+version = "0.2.0"
+edition = "2018"
 
 [lib]
 crate-type = ["cdylib", "lib", "staticlib"]
@@ -15,20 +16,25 @@ crate-type = ["cdylib", "lib", "staticlib"]
 rand = "0.4"
 bit-vec = "0.4.4"
 futures = "0.1"
-futures-cpupool = "0.1"
-num_cpus = "1"
-crossbeam = "0.7.1"
-pairing = { git = 'https://github.com/matterinc/pairing' }
-byteorder = "1"
-ff = { git = 'https://github.com/matterinc/ff', features = ["derive"] }
 
-tiny-keccak = "1.4.2"
+#pairing = { git = 'https://github.com/matterinc/pairing', tag = "0.16.2" }
+pairing = { path = "../pairing" }
+byteorder = "1"
+
+futures-cpupool = {version = "0.1", optional = true}
+num_cpus = {version = "1", optional = true}
+crossbeam = {version = "0.7.1", optional = true}
+
+tiny-keccak = {version = "1.4.2", optional = true}
 
 [dependencies.blake2-rfc]
 git = "https://github.com/gtank/blake2-rfc"
 rev = "7a5b5fc99ae483a0043db7547fb79a6fa44b88a9"
 
 [features]
-default = []
-
-singlecore = []
+default = ["multicore", "gm17", "sonic"]
+#default = ["singlecore"]
+multicore = ["futures-cpupool", "num_cpus", "crossbeam"]
+sonic = ["tiny-keccak"]
+gm17 = []
+singlecore = ["futures-cpupool"]
diff --git a/src/cs.rs b/src/cs.rs
new file mode 100644
index 0000000..a84e606
--- /dev/null
+++ b/src/cs.rs
@@ -0,0 +1,411 @@
+use pairing::{Engine};
+use pairing::ff::Field;
+
+use std::ops::{Add, Sub};
+use std::fmt;
+use std::error::Error;
+use std::io;
+use std::marker::PhantomData;
+
+/// Computations are expressed in terms of arithmetic circuits, in particular
+/// rank-1 quadratic constraint systems. The `Circuit` trait represents a
+/// circuit that can be synthesized. The `synthesize` method is called during
+/// CRS generation and during proving.
+pub trait Circuit<E: Engine> {
+    /// Synthesize the circuit into a rank-1 quadratic constraint system
+    fn synthesize<CS: ConstraintSystem<E>>(
+        self,
+        cs: &mut CS
+    ) -> Result<(), SynthesisError>;
+}
+
+/// Represents a variable in our constraint system.
+#[derive(Copy, Clone, Debug)]
+pub struct Variable(pub(crate) Index);
+
+impl Variable {
+    /// This constructs a variable with an arbitrary index.
+    /// Circuit implementations are not recommended to use this.
+    pub fn new_unchecked(idx: Index) -> Variable {
+        Variable(idx)
+    }
+
+    /// This returns the index underlying the variable.
+    /// Circuit implementations are not recommended to use this.
+    pub fn get_unchecked(&self) -> Index {
+        self.0
+    }
+}
+
+/// Represents the index of either an input variable or
+/// auxillary variable.
+#[derive(Copy, Clone, PartialEq, Debug)]
+pub enum Index {
+    Input(usize),
+    Aux(usize)
+}
+
+/// This represents a linear combination of some variables, with coefficients
+/// in the scalar field of a pairing-friendly elliptic curve group.
+#[derive(Clone)]
+pub struct LinearCombination<E: Engine>(pub(crate) Vec<(Variable, E::Fr)>);
+
+impl<E: Engine> AsRef<[(Variable, E::Fr)]> for LinearCombination<E> {
+    fn as_ref(&self) -> &[(Variable, E::Fr)] {
+        &self.0
+    }
+}
+
+impl<E: Engine> LinearCombination<E> {
+    pub fn zero() -> LinearCombination<E> {
+        LinearCombination(vec![])
+    }
+}
+
+impl<E: Engine> Add<(E::Fr, Variable)> for LinearCombination<E> {
+    type Output = LinearCombination<E>;
+
+    fn add(mut self, (coeff, var): (E::Fr, Variable)) -> LinearCombination<E> {
+        self.0.push((var, coeff));
+
+        self
+    }
+}
+
+impl<E: Engine> Sub<(E::Fr, Variable)> for LinearCombination<E> {    
+    type Output = LinearCombination<E>;
+
+    fn sub(self, (mut coeff, var): (E::Fr, Variable)) -> LinearCombination<E> {
+        coeff.negate();
+
+        self + (coeff, var)
+    }
+}
+
+impl<E: Engine> Add<Variable> for LinearCombination<E> {
+    type Output = LinearCombination<E>;
+
+    fn add(self, other: Variable) -> LinearCombination<E> {
+        self + (E::Fr::one(), other)
+    }
+}
+
+impl<E: Engine> Sub<Variable> for LinearCombination<E> {
+    type Output = LinearCombination<E>;
+
+    fn sub(self, other: Variable) -> LinearCombination<E> {
+        self - (E::Fr::one(), other)
+    }
+}
+
+impl<'a, E: Engine> Add<&'a LinearCombination<E>> for LinearCombination<E> {
+    type Output = LinearCombination<E>;
+
+    fn add(mut self, other: &'a LinearCombination<E>) -> LinearCombination<E> {
+        for s in &other.0 {
+            self = self + (s.1, s.0);
+        }
+
+        self
+    }
+}
+
+impl<'a, E: Engine> Sub<&'a LinearCombination<E>> for LinearCombination<E> {
+    type Output = LinearCombination<E>;
+
+    fn sub(mut self, other: &'a LinearCombination<E>) -> LinearCombination<E> {
+        for s in &other.0 {
+            self = self - (s.1, s.0);
+        }
+
+        self
+    }
+}
+
+impl<'a, E: Engine> Add<(E::Fr, &'a LinearCombination<E>)> for LinearCombination<E> {
+    type Output = LinearCombination<E>;
+
+    fn add(mut self, (coeff, other): (E::Fr, &'a LinearCombination<E>)) -> LinearCombination<E> {
+        for s in &other.0 {
+            let mut tmp = s.1;
+            tmp.mul_assign(&coeff);
+            self = self + (tmp, s.0);
+        }
+
+        self
+    }
+}
+
+impl<'a, E: Engine> Sub<(E::Fr, &'a LinearCombination<E>)> for LinearCombination<E> {
+    type Output = LinearCombination<E>;
+
+    fn sub(mut self, (coeff, other): (E::Fr, &'a LinearCombination<E>)) -> LinearCombination<E> {
+        for s in &other.0 {
+            let mut tmp = s.1;
+            tmp.mul_assign(&coeff);
+            self = self - (tmp, s.0);
+        }
+
+        self
+    }
+}
+
+/// This is an error that could occur during circuit synthesis contexts,
+/// such as CRS generation, proving or verification.
+#[derive(Debug)]
+pub enum SynthesisError {
+    /// During synthesis, we lacked knowledge of a variable assignment.
+    AssignmentMissing,
+    /// During synthesis, we divided by zero.
+    DivisionByZero,
+    /// During synthesis, we constructed an unsatisfiable constraint system.
+    Unsatisfiable,
+    /// During synthesis, our polynomials ended up being too high of degree
+    PolynomialDegreeTooLarge,
+    /// During proof generation, we encountered an identity in the CRS
+    UnexpectedIdentity,
+    /// During proof generation, we encountered an I/O error with the CRS
+    IoError(io::Error),
+    /// During verification, our verifying key was malformed.
+    MalformedVerifyingKey,
+    /// During CRS generation, we observed an unconstrained auxillary variable
+    UnconstrainedVariable
+}
+
+impl From<io::Error> for SynthesisError {
+    fn from(e: io::Error) -> SynthesisError {
+        SynthesisError::IoError(e)
+    }
+}
+
+impl Error for SynthesisError {
+    fn description(&self) -> &str {
+        match *self {
+            SynthesisError::AssignmentMissing => "an assignment for a variable could not be computed",
+            SynthesisError::DivisionByZero => "division by zero",
+            SynthesisError::Unsatisfiable => "unsatisfiable constraint system",
+            SynthesisError::PolynomialDegreeTooLarge => "polynomial degree is too large",
+            SynthesisError::UnexpectedIdentity => "encountered an identity element in the CRS",
+            SynthesisError::IoError(_) => "encountered an I/O error",
+            SynthesisError::MalformedVerifyingKey => "malformed verifying key",
+            SynthesisError::UnconstrainedVariable => "auxillary variable was unconstrained"
+        }
+    }
+}
+
+impl fmt::Display for SynthesisError {
+    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
+        if let &SynthesisError::IoError(ref e) = self {
+            write!(f, "I/O error: ")?;
+            e.fmt(f)
+        } else {
+            write!(f, "{}", self.description())
+        }
+    }
+}
+
+/// Represents a constraint system which can have new variables
+/// allocated and constrains between them formed.
+pub trait ConstraintSystem<E: Engine>: Sized {
+    /// Represents the type of the "root" of this constraint system
+    /// so that nested namespaces can minimize indirection.
+    type Root: ConstraintSystem<E>;
+
+    /// Return the "one" input variable
+    fn one() -> Variable {
+        Variable::new_unchecked(Index::Input(0))
+    }
+
+    /// Allocate a private variable in the constraint system. The provided function is used to
+    /// determine the assignment of the variable. The given `annotation` function is invoked
+    /// in testing contexts in order to derive a unique name for this variable in the current
+    /// namespace.
+    fn alloc<F, A, AR>(
+        &mut self,
+        annotation: A,
+        f: F
+    ) -> Result<Variable, SynthesisError>
+        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>;
+
+    /// Allocate a public variable in the constraint system. The provided function is used to
+    /// determine the assignment of the variable.
+    fn alloc_input<F, A, AR>(
+        &mut self,
+        annotation: A,
+        f: F
+    ) -> Result<Variable, SynthesisError>
+        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>;
+
+    /// Enforce that `A` * `B` = `C`. The `annotation` function is invoked in testing contexts
+    /// in order to derive a unique name for the constraint in the current namespace.
+    fn enforce<A, AR, LA, LB, LC>(
+        &mut self,
+        annotation: A,
+        a: LA,
+        b: LB,
+        c: LC
+    )
+        where A: FnOnce() -> AR, AR: Into<String>,
+              LA: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
+              LB: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
+              LC: FnOnce(LinearCombination<E>) -> LinearCombination<E>;
+
+    /// Create a new (sub)namespace and enter into it. Not intended
+    /// for downstream use; use `namespace` instead.
+    fn push_namespace<NR, N>(&mut self, name_fn: N)
+        where NR: Into<String>, N: FnOnce() -> NR;
+
+    /// Exit out of the existing namespace. Not intended for
+    /// downstream use; use `namespace` instead.
+    fn pop_namespace(&mut self);
+
+    /// Gets the "root" constraint system, bypassing the namespacing.
+    /// Not intended for downstream use; use `namespace` instead.
+    fn get_root(&mut self) -> &mut Self::Root;
+
+    /// Begin a namespace for this constraint system.
+    fn namespace<'a, NR, N>(
+        &'a mut self,
+        name_fn: N
+    ) -> Namespace<'a, E, Self::Root>
+        where NR: Into<String>, N: FnOnce() -> NR
+    {
+        self.get_root().push_namespace(name_fn);
+
+        Namespace(self.get_root(), PhantomData)
+    }
+}
+
+/// This is a "namespaced" constraint system which borrows a constraint system (pushing
+/// a namespace context) and, when dropped, pops out of the namespace context.
+pub struct Namespace<'a, E: Engine, CS: ConstraintSystem<E> + 'a>(&'a mut CS, PhantomData<E>);
+
+impl<'cs, E: Engine, CS: ConstraintSystem<E>> ConstraintSystem<E> for Namespace<'cs, E, CS> {
+    type Root = CS::Root;
+
+    fn one() -> Variable {
+        CS::one()
+    }
+
+    fn alloc<F, A, AR>(
+        &mut self,
+        annotation: A,
+        f: F
+    ) -> Result<Variable, SynthesisError>
+        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
+    {
+        self.0.alloc(annotation, f)
+    }
+
+    fn alloc_input<F, A, AR>(
+        &mut self,
+        annotation: A,
+        f: F
+    ) -> Result<Variable, SynthesisError>
+        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
+    {
+        self.0.alloc_input(annotation, f)
+    }
+
+    fn enforce<A, AR, LA, LB, LC>(
+        &mut self,
+        annotation: A,
+        a: LA,
+        b: LB,
+        c: LC
+    )
+        where A: FnOnce() -> AR, AR: Into<String>,
+              LA: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
+              LB: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
+              LC: FnOnce(LinearCombination<E>) -> LinearCombination<E>
+    {
+        self.0.enforce(annotation, a, b, c)
+    }
+
+    // Downstream users who use `namespace` will never interact with these
+    // functions and they will never be invoked because the namespace is
+    // never a root constraint system.
+
+    fn push_namespace<NR, N>(&mut self, _: N)
+        where NR: Into<String>, N: FnOnce() -> NR
+    {
+        panic!("only the root's push_namespace should be called");
+    }
+
+    fn pop_namespace(&mut self)
+    {
+        panic!("only the root's pop_namespace should be called");
+    }
+
+    fn get_root(&mut self) -> &mut Self::Root
+    {
+        self.0.get_root()
+    }
+}
+
+impl<'a, E: Engine, CS: ConstraintSystem<E>> Drop for Namespace<'a, E, CS> {
+    fn drop(&mut self) {
+        self.get_root().pop_namespace()
+    }
+}
+
+/// Convenience implementation of ConstraintSystem<E> for mutable references to
+/// constraint systems.
+impl<'cs, E: Engine, CS: ConstraintSystem<E>> ConstraintSystem<E> for &'cs mut CS {
+    type Root = CS::Root;
+
+    fn one() -> Variable {
+        CS::one()
+    }
+
+    fn alloc<F, A, AR>(
+        &mut self,
+        annotation: A,
+        f: F
+    ) -> Result<Variable, SynthesisError>
+        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
+    {
+        (**self).alloc(annotation, f)
+    }
+
+    fn alloc_input<F, A, AR>(
+        &mut self,
+        annotation: A,
+        f: F
+    ) -> Result<Variable, SynthesisError>
+        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
+    {
+        (**self).alloc_input(annotation, f)
+    }
+
+    fn enforce<A, AR, LA, LB, LC>(
+        &mut self,
+        annotation: A,
+        a: LA,
+        b: LB,
+        c: LC
+    )
+        where A: FnOnce() -> AR, AR: Into<String>,
+              LA: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
+              LB: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
+              LC: FnOnce(LinearCombination<E>) -> LinearCombination<E>
+    {
+        (**self).enforce(annotation, a, b, c)
+    }
+
+    fn push_namespace<NR, N>(&mut self, name_fn: N)
+        where NR: Into<String>, N: FnOnce() -> NR
+    {
+        (**self).push_namespace(name_fn)
+    }
+
+    fn pop_namespace(&mut self)
+    {
+        (**self).pop_namespace()
+    }
+
+    fn get_root(&mut self) -> &mut Self::Root
+    {
+        (**self).get_root()
+    }
+}
\ No newline at end of file
diff --git a/src/domain.rs b/src/domain.rs
index 1d0203a..9622219 100644
--- a/src/domain.rs
+++ b/src/domain.rs
@@ -15,7 +15,7 @@ use pairing::{
     CurveProjective
 };
 
-use ff::{
+use pairing::ff::{
     Field, 
     PrimeField
 };
@@ -24,13 +24,9 @@ use super::{
     SynthesisError
 };
 
-
-use super::multicore::Worker;
+use super::worker::Worker;
 pub use super::group::*;
 
-#[cfg(not(feature = "singlecore"))]
-use super::parallel_fft::*;
-
 pub struct EvaluationDomain<E: Engine, G: Group<E>> {
     coeffs: Vec<G>,
     exp: u32,
@@ -55,7 +51,7 @@ impl<E: Engine, G: Group<E>> EvaluationDomain<E, G> {
 
     pub fn from_coeffs(mut coeffs: Vec<G>) -> Result<EvaluationDomain<E, G>, SynthesisError>
     {
-        use ff::PrimeField;
+        use pairing::ff::PrimeField;
         // Compute the size of our evaluation domain
 
         let coeffs_len = coeffs.len();
@@ -105,7 +101,7 @@ impl<E: Engine, G: Group<E>> EvaluationDomain<E, G> {
     // this one does expect coefficients to be smaller than `num_roots_of_unity/2` as we expect multiplication
     pub fn from_coeffs_into_sized(mut coeffs: Vec<G>, size: usize) -> Result<EvaluationDomain<E, G>, SynthesisError>
     {
-        use ff::PrimeField;
+        use pairing::ff::PrimeField;
         // Compute the size of our evaluation domain
 
         assert!(size >= coeffs.len());
@@ -264,6 +260,121 @@ impl<E: Engine, G: Group<E>> EvaluationDomain<E, G> {
     }
 }
 
+pub(crate) fn best_fft<E: Engine, T: Group<E>>(a: &mut [T], worker: &Worker, omega: &E::Fr, log_n: u32)
+{
+    let log_cpus = worker.log_num_cpus();
+
+    if log_n <= log_cpus {
+        serial_fft(a, omega, log_n);
+    } else {
+        parallel_fft(a, worker, omega, log_n, log_cpus);
+    }
+}
+
+pub(crate) fn serial_fft<E: Engine, T: Group<E>>(a: &mut [T], omega: &E::Fr, log_n: u32)
+{
+    fn bitreverse(mut n: u32, l: u32) -> u32 {
+        let mut r = 0;
+        for _ in 0..l {
+            r = (r << 1) | (n & 1);
+            n >>= 1;
+        }
+        r
+    }
+
+    let n = a.len() as u32;
+    assert_eq!(n, 1 << log_n);
+
+    for k in 0..n {
+        let rk = bitreverse(k, log_n);
+        if k < rk {
+            a.swap(rk as usize, k as usize);
+        }
+    }
+
+    let mut m = 1;
+    for _ in 0..log_n {
+        let w_m = omega.pow(&[(n / (2*m)) as u64]);
+
+        let mut k = 0;
+        while k < n {
+            let mut w = E::Fr::one();
+            for j in 0..m {
+                let mut t = a[(k+j+m) as usize];
+                t.group_mul_assign(&w);
+                let mut tmp = a[(k+j) as usize];
+                tmp.group_sub_assign(&t);
+                a[(k+j+m) as usize] = tmp;
+                a[(k+j) as usize].group_add_assign(&t);
+                w.mul_assign(&w_m);
+            }
+
+            k += 2*m;
+        }
+
+        m *= 2;
+    }
+}
+
+pub(crate) fn parallel_fft<E: Engine, T: Group<E>>(
+    a: &mut [T],
+    worker: &Worker,
+    omega: &E::Fr,
+    log_n: u32,
+    log_cpus: u32
+)
+{
+    assert!(log_n >= log_cpus);
+
+    let num_cpus = 1 << log_cpus;
+    let log_new_n = log_n - log_cpus;
+    let mut tmp = vec![vec![T::group_zero(); 1 << log_new_n]; num_cpus];
+    let new_omega = omega.pow(&[num_cpus as u64]);
+
+    worker.scope(0, |scope, _| {
+        let a = &*a;
+
+        for (j, tmp) in tmp.iter_mut().enumerate() {
+            scope.spawn(move |_| {
+                // Shuffle into a sub-FFT
+                let omega_j = omega.pow(&[j as u64]);
+                let omega_step = omega.pow(&[(j as u64) << log_new_n]);
+
+                let mut elt = E::Fr::one();
+                for i in 0..(1 << log_new_n) {
+                    for s in 0..num_cpus {
+                        let idx = (i + (s << log_new_n)) % (1 << log_n);
+                        let mut t = a[idx];
+                        t.group_mul_assign(&elt);
+                        tmp[i].group_add_assign(&t);
+                        elt.mul_assign(&omega_step);
+                    }
+                    elt.mul_assign(&omega_j);
+                }
+
+                // Perform sub-FFT
+                serial_fft(tmp, &new_omega, log_new_n);
+            });
+        }
+    });
+
+    // TODO: does this hurt or help?
+    worker.scope(a.len(), |scope, chunk| {
+        let tmp = &tmp;
+
+        for (idx, a) in a.chunks_mut(chunk).enumerate() {
+            scope.spawn(move |_| {
+                let mut idx = idx * chunk;
+                let mask = (1 << log_cpus) - 1;
+                for a in a {
+                    *a = tmp[idx & mask][idx >> log_cpus];
+                    idx += 1;
+                }
+            });
+        }
+    });
+}
+
 // Test multiplying various (low degree) polynomials together and
 // comparing with naive evaluations.
 #[test]
diff --git a/src/gm17/generator.rs b/src/gm17/generator.rs
index 059a041..320ab07 100644
--- a/src/gm17/generator.rs
+++ b/src/gm17/generator.rs
@@ -11,7 +11,7 @@ use pairing::{
     CurveAffine
 };
 
-use ff::{    
+use pairing::ff::{    
     PrimeField,
     Field
 };
@@ -21,7 +21,7 @@ use super::{
     VerifyingKey
 };
 
-use ::{
+use crate::{
     SynthesisError,
     Circuit,
     ConstraintSystem,
@@ -30,12 +30,12 @@ use ::{
     Index
 };
 
-use ::domain::{
+use crate::domain::{
     EvaluationDomain,
     Scalar
 };
 
-use ::multicore::{
+use crate::worker::{
     Worker
 };
 
diff --git a/src/gm17/mod.rs b/src/gm17/mod.rs
index b8c5b7f..aee5de6 100644
--- a/src/gm17/mod.rs
+++ b/src/gm17/mod.rs
@@ -4,11 +4,11 @@ use pairing::{
     EncodedPoint
 };
 
-use ::{
+use crate::{
     SynthesisError
 };
 
-use multiexp::SourceBuilder;
+use crate::source::SourceBuilder;
 use std::io::{self, Read, Write};
 use std::sync::Arc;
 use byteorder::{BigEndian, WriteBytesExt, ReadBytesExt};
@@ -474,7 +474,7 @@ impl<E: Engine> PartialEq for Parameters<E> {
 //     use {Circuit, SynthesisError, ConstraintSystem};
 
 //     use rand::{Rand, thread_rng};
-//     use ff::{Field};
+//     use pairing::ff::{Field};
 //     use pairing::bls12_381::{Bls12, Fr};
 
 //     #[test]
diff --git a/src/gm17/tests/mod.rs b/src/gm17/tests/mod.rs
index 4440b19..ee34ec6 100644
--- a/src/gm17/tests/mod.rs
+++ b/src/gm17/tests/mod.rs
@@ -2,7 +2,7 @@ use pairing::{
     Engine
 };
 
-use ff:: {
+use pairing::ff:: {
     Field,
     PrimeField,
 };
@@ -12,7 +12,7 @@ use super::super::tests::XORDemo;
 
 use std::marker::PhantomData;
 
-use ::{
+use crate::{
     Circuit,
     ConstraintSystem,
     SynthesisError
diff --git a/src/groth16/generator.rs b/src/groth16/generator.rs
index 0d597a6..3d4cb1f 100644
--- a/src/groth16/generator.rs
+++ b/src/groth16/generator.rs
@@ -11,7 +11,7 @@ use pairing::{
     CurveAffine
 };
 
-use ff::{    
+use pairing::ff::{    
     PrimeField,
     Field
 };
@@ -21,7 +21,7 @@ use super::{
     VerifyingKey
 };
 
-use ::{
+use crate::{
     SynthesisError,
     Circuit,
     ConstraintSystem,
@@ -30,12 +30,12 @@ use ::{
     Index
 };
 
-use ::domain::{
+use crate::domain::{
     EvaluationDomain,
     Scalar
 };
 
-use ::multicore::{
+use crate::worker::{
     Worker
 };
 
diff --git a/src/groth16/mod.rs b/src/groth16/mod.rs
index 7dcc8ab..14fb642 100644
--- a/src/groth16/mod.rs
+++ b/src/groth16/mod.rs
@@ -4,11 +4,11 @@ use pairing::{
     EncodedPoint
 };
 
-use ::{
+use crate::{
     SynthesisError
 };
 
-use multiexp::SourceBuilder;
+use crate::source::SourceBuilder;
 use std::io::{self, Read, Write};
 use std::sync::Arc;
 use byteorder::{BigEndian, WriteBytesExt, ReadBytesExt};
@@ -484,10 +484,10 @@ impl<'a, E: Engine> ParameterSource<E> for &'a Parameters<E> {
 #[cfg(test)]
 mod test_with_bls12_381 {
     use super::*;
-    use {Circuit, SynthesisError, ConstraintSystem};
+    use crate::{Circuit, SynthesisError, ConstraintSystem};
 
     use rand::{Rand, thread_rng};
-    use ff::{Field};
+    use pairing::ff::{Field};
     use pairing::bls12_381::{Bls12, Fr};
 
     #[test]
diff --git a/src/groth16/prover.rs b/src/groth16/prover.rs
index 98f2b62..f66e251 100644
--- a/src/groth16/prover.rs
+++ b/src/groth16/prover.rs
@@ -12,7 +12,7 @@ use pairing::{
     CurveAffine
 };
 
-use ff::{
+use pairing::ff::{
     PrimeField,
     Field
 };
@@ -22,7 +22,7 @@ use super::{
     Proof
 };
 
-use ::{
+use crate::{
     SynthesisError,
     Circuit,
     ConstraintSystem,
@@ -31,18 +31,19 @@ use ::{
     Index
 };
 
-use ::domain::{
+use crate::domain::{
     EvaluationDomain,
     Scalar
 };
 
-use ::multiexp::{
+use crate::source::{
     DensityTracker,
-    FullDensity,
-    multiexp
+    FullDensity
 };
 
-use ::multicore::{
+use crate::multiexp::*;
+
+use crate::worker::{
     Worker
 };
 
diff --git a/src/groth16/tests/mod.rs b/src/groth16/tests/mod.rs
index 5a2bb2c..7442501 100644
--- a/src/groth16/tests/mod.rs
+++ b/src/groth16/tests/mod.rs
@@ -2,7 +2,7 @@ use pairing::{
     Engine
 };
 
-use ff:: {
+use pairing::ff:: {
     Field,
     PrimeField,
 };
@@ -12,7 +12,7 @@ use super::super::tests::XORDemo;
 
 use std::marker::PhantomData;
 
-use ::{
+use crate::{
     Circuit,
     ConstraintSystem,
     SynthesisError
diff --git a/src/groth16/verifier.rs b/src/groth16/verifier.rs
index 6d30915..ad1c210 100644
--- a/src/groth16/verifier.rs
+++ b/src/groth16/verifier.rs
@@ -4,7 +4,7 @@ use pairing::{
     CurveAffine
 };
 
-use ff::{PrimeField};
+use pairing::ff::{PrimeField};
 
 use super::{
     Proof,
@@ -12,7 +12,7 @@ use super::{
     PreparedVerifyingKey
 };
 
-use ::{
+use crate::{
     SynthesisError
 };
 
diff --git a/src/group.rs b/src/group.rs
index 7b4fe2e..f66b776 100644
--- a/src/group.rs
+++ b/src/group.rs
@@ -3,7 +3,7 @@ use pairing::{
     CurveProjective
 };
 
-use ff::{
+use pairing::ff::{
     Field, 
     PrimeField
 };
diff --git a/src/lib.rs b/src/lib.rs
index e7f17e3..690219a 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,458 +1,45 @@
 #![allow(unused_imports)]
 
-extern crate pairing;
+extern crate pairing as pairing_import;
 extern crate rand;
 extern crate bit_vec;
 extern crate byteorder;
-extern crate ff;
-
-#[cfg(not(feature = "singlecore"))]
-extern crate num_cpus;
-extern crate futures;
-extern crate futures_cpupool;
-extern crate crossbeam;
 
 pub mod domain;
 pub mod groth16;
+
+#[cfg(feature = "gm17")]
 pub mod gm17;
+#[cfg(feature = "sonic")]
 pub mod sonic;
 
 mod group;
 mod source;
-
-#[cfg(not(feature = "singlecore"))]
-mod parallel_fft;
-mod multicore;
-mod parallel_multiexp;
-
-#[cfg(feature = "singlecore")]
-mod serial_fft;
-mod serial_multiexp;
+mod multiexp;
 
 #[cfg(test)]
 mod tests;
 
-use pairing::{Engine};
-use ff::Field;
+#[cfg(feature = "multicore")]
+mod multicore;
 
-use std::ops::{Add, Sub};
-use std::fmt;
-use std::error::Error;
-use std::io;
-use std::marker::PhantomData;
+#[cfg(feature = "singlecore")]
+mod singlecore;
 
-pub mod multiexp {
-    pub use source::*;
-
-    #[cfg(not(feature = "singlecore"))]
-    pub use parallel_multiexp::*;
+mod worker {
+    #[cfg(feature = "multicore")]
+    pub use crate::multicore::*;
 
     #[cfg(feature = "singlecore")]
-    pub use serial_multiexp::*;
+    pub use crate::singlecore::*;
 }
 
-/// Computations are expressed in terms of arithmetic circuits, in particular
-/// rank-1 quadratic constraint systems. The `Circuit` trait represents a
-/// circuit that can be synthesized. The `synthesize` method is called during
-/// CRS generation and during proving.
-pub trait Circuit<E: Engine> {
-    /// Synthesize the circuit into a rank-1 quadratic constraint system
-    fn synthesize<CS: ConstraintSystem<E>>(
-        self,
-        cs: &mut CS
-    ) -> Result<(), SynthesisError>;
+pub mod pairing {
+    pub use pairing_import::*;
 }
 
-/// Represents a variable in our constraint system.
-#[derive(Copy, Clone, Debug)]
-pub struct Variable(Index);
-
-impl Variable {
-    /// This constructs a variable with an arbitrary index.
-    /// Circuit implementations are not recommended to use this.
-    pub fn new_unchecked(idx: Index) -> Variable {
-        Variable(idx)
-    }
-
-    /// This returns the index underlying the variable.
-    /// Circuit implementations are not recommended to use this.
-    pub fn get_unchecked(&self) -> Index {
-        self.0
-    }
-}
-
-/// Represents the index of either an input variable or
-/// auxillary variable.
-#[derive(Copy, Clone, PartialEq, Debug)]
-pub enum Index {
-    Input(usize),
-    Aux(usize)
-}
-
-/// This represents a linear combination of some variables, with coefficients
-/// in the scalar field of a pairing-friendly elliptic curve group.
-#[derive(Clone)]
-pub struct LinearCombination<E: Engine>(Vec<(Variable, E::Fr)>);
-
-impl<E: Engine> AsRef<[(Variable, E::Fr)]> for LinearCombination<E> {
-    fn as_ref(&self) -> &[(Variable, E::Fr)] {
-        &self.0
-    }
-}
-
-impl<E: Engine> LinearCombination<E> {
-    pub fn zero() -> LinearCombination<E> {
-        LinearCombination(vec![])
-    }
-}
-
-impl<E: Engine> Add<(E::Fr, Variable)> for LinearCombination<E> {
-    type Output = LinearCombination<E>;
-
-    fn add(mut self, (coeff, var): (E::Fr, Variable)) -> LinearCombination<E> {
-        self.0.push((var, coeff));
-
-        self
-    }
-}
-
-impl<E: Engine> Sub<(E::Fr, Variable)> for LinearCombination<E> {    
-    type Output = LinearCombination<E>;
-
-    fn sub(self, (mut coeff, var): (E::Fr, Variable)) -> LinearCombination<E> {
-        coeff.negate();
-
-        self + (coeff, var)
-    }
-}
-
-impl<E: Engine> Add<Variable> for LinearCombination<E> {
-    type Output = LinearCombination<E>;
-
-    fn add(self, other: Variable) -> LinearCombination<E> {
-        self + (E::Fr::one(), other)
-    }
-}
-
-impl<E: Engine> Sub<Variable> for LinearCombination<E> {
-    type Output = LinearCombination<E>;
-
-    fn sub(self, other: Variable) -> LinearCombination<E> {
-        self - (E::Fr::one(), other)
-    }
-}
-
-impl<'a, E: Engine> Add<&'a LinearCombination<E>> for LinearCombination<E> {
-    type Output = LinearCombination<E>;
-
-    fn add(mut self, other: &'a LinearCombination<E>) -> LinearCombination<E> {
-        for s in &other.0 {
-            self = self + (s.1, s.0);
-        }
-
-        self
-    }
-}
-
-impl<'a, E: Engine> Sub<&'a LinearCombination<E>> for LinearCombination<E> {
-    type Output = LinearCombination<E>;
-
-    fn sub(mut self, other: &'a LinearCombination<E>) -> LinearCombination<E> {
-        for s in &other.0 {
-            self = self - (s.1, s.0);
-        }
-
-        self
-    }
-}
-
-impl<'a, E: Engine> Add<(E::Fr, &'a LinearCombination<E>)> for LinearCombination<E> {
-    type Output = LinearCombination<E>;
-
-    fn add(mut self, (coeff, other): (E::Fr, &'a LinearCombination<E>)) -> LinearCombination<E> {
-        for s in &other.0 {
-            let mut tmp = s.1;
-            tmp.mul_assign(&coeff);
-            self = self + (tmp, s.0);
-        }
-
-        self
-    }
-}
-
-impl<'a, E: Engine> Sub<(E::Fr, &'a LinearCombination<E>)> for LinearCombination<E> {
-    type Output = LinearCombination<E>;
-
-    fn sub(mut self, (coeff, other): (E::Fr, &'a LinearCombination<E>)) -> LinearCombination<E> {
-        for s in &other.0 {
-            let mut tmp = s.1;
-            tmp.mul_assign(&coeff);
-            self = self - (tmp, s.0);
-        }
-
-        self
-    }
-}
-
-/// This is an error that could occur during circuit synthesis contexts,
-/// such as CRS generation, proving or verification.
-#[derive(Debug)]
-pub enum SynthesisError {
-    /// During synthesis, we lacked knowledge of a variable assignment.
-    AssignmentMissing,
-    /// During synthesis, we divided by zero.
-    DivisionByZero,
-    /// During synthesis, we constructed an unsatisfiable constraint system.
-    Unsatisfiable,
-    /// During synthesis, our polynomials ended up being too high of degree
-    PolynomialDegreeTooLarge,
-    /// During proof generation, we encountered an identity in the CRS
-    UnexpectedIdentity,
-    /// During proof generation, we encountered an I/O error with the CRS
-    IoError(io::Error),
-    /// During verification, our verifying key was malformed.
-    MalformedVerifyingKey,
-    /// During CRS generation, we observed an unconstrained auxillary variable
-    UnconstrainedVariable
-}
-
-impl From<io::Error> for SynthesisError {
-    fn from(e: io::Error) -> SynthesisError {
-        SynthesisError::IoError(e)
-    }
-}
-
-impl Error for SynthesisError {
-    fn description(&self) -> &str {
-        match *self {
-            SynthesisError::AssignmentMissing => "an assignment for a variable could not be computed",
-            SynthesisError::DivisionByZero => "division by zero",
-            SynthesisError::Unsatisfiable => "unsatisfiable constraint system",
-            SynthesisError::PolynomialDegreeTooLarge => "polynomial degree is too large",
-            SynthesisError::UnexpectedIdentity => "encountered an identity element in the CRS",
-            SynthesisError::IoError(_) => "encountered an I/O error",
-            SynthesisError::MalformedVerifyingKey => "malformed verifying key",
-            SynthesisError::UnconstrainedVariable => "auxillary variable was unconstrained"
-        }
-    }
-}
-
-impl fmt::Display for SynthesisError {
-    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
-        if let &SynthesisError::IoError(ref e) = self {
-            write!(f, "I/O error: ")?;
-            e.fmt(f)
-        } else {
-            write!(f, "{}", self.description())
-        }
-    }
-}
-
-/// Represents a constraint system which can have new variables
-/// allocated and constrains between them formed.
-pub trait ConstraintSystem<E: Engine>: Sized {
-    /// Represents the type of the "root" of this constraint system
-    /// so that nested namespaces can minimize indirection.
-    type Root: ConstraintSystem<E>;
-
-    /// Return the "one" input variable
-    fn one() -> Variable {
-        Variable::new_unchecked(Index::Input(0))
-    }
-
-    /// Allocate a private variable in the constraint system. The provided function is used to
-    /// determine the assignment of the variable. The given `annotation` function is invoked
-    /// in testing contexts in order to derive a unique name for this variable in the current
-    /// namespace.
-    fn alloc<F, A, AR>(
-        &mut self,
-        annotation: A,
-        f: F
-    ) -> Result<Variable, SynthesisError>
-        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>;
-
-    /// Allocate a public variable in the constraint system. The provided function is used to
-    /// determine the assignment of the variable.
-    fn alloc_input<F, A, AR>(
-        &mut self,
-        annotation: A,
-        f: F
-    ) -> Result<Variable, SynthesisError>
-        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>;
-
-    /// Enforce that `A` * `B` = `C`. The `annotation` function is invoked in testing contexts
-    /// in order to derive a unique name for the constraint in the current namespace.
-    fn enforce<A, AR, LA, LB, LC>(
-        &mut self,
-        annotation: A,
-        a: LA,
-        b: LB,
-        c: LC
-    )
-        where A: FnOnce() -> AR, AR: Into<String>,
-              LA: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
-              LB: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
-              LC: FnOnce(LinearCombination<E>) -> LinearCombination<E>;
-
-    /// Create a new (sub)namespace and enter into it. Not intended
-    /// for downstream use; use `namespace` instead.
-    fn push_namespace<NR, N>(&mut self, name_fn: N)
-        where NR: Into<String>, N: FnOnce() -> NR;
-
-    /// Exit out of the existing namespace. Not intended for
-    /// downstream use; use `namespace` instead.
-    fn pop_namespace(&mut self);
-
-    /// Gets the "root" constraint system, bypassing the namespacing.
-    /// Not intended for downstream use; use `namespace` instead.
-    fn get_root(&mut self) -> &mut Self::Root;
-
-    /// Begin a namespace for this constraint system.
-    fn namespace<'a, NR, N>(
-        &'a mut self,
-        name_fn: N
-    ) -> Namespace<'a, E, Self::Root>
-        where NR: Into<String>, N: FnOnce() -> NR
-    {
-        self.get_root().push_namespace(name_fn);
-
-        Namespace(self.get_root(), PhantomData)
-    }
-}
-
-/// This is a "namespaced" constraint system which borrows a constraint system (pushing
-/// a namespace context) and, when dropped, pops out of the namespace context.
-pub struct Namespace<'a, E: Engine, CS: ConstraintSystem<E> + 'a>(&'a mut CS, PhantomData<E>);
-
-impl<'cs, E: Engine, CS: ConstraintSystem<E>> ConstraintSystem<E> for Namespace<'cs, E, CS> {
-    type Root = CS::Root;
-
-    fn one() -> Variable {
-        CS::one()
-    }
-
-    fn alloc<F, A, AR>(
-        &mut self,
-        annotation: A,
-        f: F
-    ) -> Result<Variable, SynthesisError>
-        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
-    {
-        self.0.alloc(annotation, f)
-    }
-
-    fn alloc_input<F, A, AR>(
-        &mut self,
-        annotation: A,
-        f: F
-    ) -> Result<Variable, SynthesisError>
-        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
-    {
-        self.0.alloc_input(annotation, f)
-    }
-
-    fn enforce<A, AR, LA, LB, LC>(
-        &mut self,
-        annotation: A,
-        a: LA,
-        b: LB,
-        c: LC
-    )
-        where A: FnOnce() -> AR, AR: Into<String>,
-              LA: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
-              LB: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
-              LC: FnOnce(LinearCombination<E>) -> LinearCombination<E>
-    {
-        self.0.enforce(annotation, a, b, c)
-    }
-
-    // Downstream users who use `namespace` will never interact with these
-    // functions and they will never be invoked because the namespace is
-    // never a root constraint system.
-
-    fn push_namespace<NR, N>(&mut self, _: N)
-        where NR: Into<String>, N: FnOnce() -> NR
-    {
-        panic!("only the root's push_namespace should be called");
-    }
-
-    fn pop_namespace(&mut self)
-    {
-        panic!("only the root's pop_namespace should be called");
-    }
-
-    fn get_root(&mut self) -> &mut Self::Root
-    {
-        self.0.get_root()
-    }
-}
-
-impl<'a, E: Engine, CS: ConstraintSystem<E>> Drop for Namespace<'a, E, CS> {
-    fn drop(&mut self) {
-        self.get_root().pop_namespace()
-    }
-}
-
-/// Convenience implementation of ConstraintSystem<E> for mutable references to
-/// constraint systems.
-impl<'cs, E: Engine, CS: ConstraintSystem<E>> ConstraintSystem<E> for &'cs mut CS {
-    type Root = CS::Root;
-
-    fn one() -> Variable {
-        CS::one()
-    }
-
-    fn alloc<F, A, AR>(
-        &mut self,
-        annotation: A,
-        f: F
-    ) -> Result<Variable, SynthesisError>
-        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
-    {
-        (**self).alloc(annotation, f)
-    }
-
-    fn alloc_input<F, A, AR>(
-        &mut self,
-        annotation: A,
-        f: F
-    ) -> Result<Variable, SynthesisError>
-        where F: FnOnce() -> Result<E::Fr, SynthesisError>, A: FnOnce() -> AR, AR: Into<String>
-    {
-        (**self).alloc_input(annotation, f)
-    }
-
-    fn enforce<A, AR, LA, LB, LC>(
-        &mut self,
-        annotation: A,
-        a: LA,
-        b: LB,
-        c: LC
-    )
-        where A: FnOnce() -> AR, AR: Into<String>,
-              LA: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
-              LB: FnOnce(LinearCombination<E>) -> LinearCombination<E>,
-              LC: FnOnce(LinearCombination<E>) -> LinearCombination<E>
-    {
-        (**self).enforce(annotation, a, b, c)
-    }
-
-    fn push_namespace<NR, N>(&mut self, name_fn: N)
-        where NR: Into<String>, N: FnOnce() -> NR
-    {
-        (**self).push_namespace(name_fn)
-    }
-
-    fn pop_namespace(&mut self)
-    {
-        (**self).pop_namespace()
-    }
-
-    fn get_root(&mut self) -> &mut Self::Root
-    {
-        (**self).get_root()
-    }
-}
+mod cs;
+pub use self::cs::*;
 
 static mut VERBOSE_SWITCH: i8 = -1;
 
diff --git a/src/multicore.rs b/src/multicore.rs
index 84055d7..bb113c6 100644
--- a/src/multicore.rs
+++ b/src/multicore.rs
@@ -4,11 +4,14 @@
 //! crossbeam but may be extended in the future to
 //! allow for various parallelism strategies.
 
-use num_cpus;
-use futures::{Future, IntoFuture, Poll};
-use futures_cpupool::{CpuPool, CpuFuture};
-use crossbeam::{self};
-use crossbeam::thread::{Scope};
+extern crate num_cpus;
+extern crate futures;
+extern crate futures_cpupool;
+extern crate crossbeam;
+
+use self::futures::{Future, IntoFuture, Poll};
+use self::futures_cpupool::{CpuPool, CpuFuture};
+use self::crossbeam::thread::{Scope};
 
 #[derive(Clone)]
 pub struct Worker {
diff --git a/src/parallel_multiexp.rs b/src/multiexp.rs
similarity index 93%
rename from src/parallel_multiexp.rs
rename to src/multiexp.rs
index 8b3d3bb..1e114d9 100644
--- a/src/parallel_multiexp.rs
+++ b/src/multiexp.rs
@@ -4,7 +4,7 @@ use pairing::{
     Engine
 };
 
-use ff::{
+use pairing::ff::{
     PrimeField,
     Field,
     PrimeFieldRepr,
@@ -13,7 +13,7 @@ use ff::{
 use std::sync::Arc;
 use super::source::*;
 use futures::{Future};
-use super::multicore::Worker;
+use super::worker::Worker;
 
 use super::SynthesisError;
 
@@ -212,19 +212,16 @@ fn dense_multiexp_inner<G: CurveAffine>(
     handle_trivial: bool
 ) -> Result<<G as CurveAffine>::Projective, SynthesisError>
 {   
-    use crossbeam::channel::{unbounded, RecvError};
     // Perform this region of the multiexp. We use a different strategy - go over region in parallel,
     // then over another region, etc. No Arc required
     let this = {
         // let mask = (1u64 << c) - 1u64;
-        let (s, r) = unbounded();
-        // let this_region = 
+        let this_region = 
         pool.scope(bases.len(), |scope, chunk| {
-            // let mut handles = vec![];
-            // let mut this_acc = <G as CurveAffine>::Projective::zero();
+            let mut handles = vec![];
+            let mut this_acc = <G as CurveAffine>::Projective::zero();
             for (base, exp) in bases.chunks(chunk).zip(exponents.chunks(chunk)) {
-                let s = s.clone();
-                // let handle = 
+                let handle = 
                 scope.spawn(move |_| {
                     let mut buckets = vec![<G as CurveAffine>::Projective::zero(); (1 << c) - 1];
                     // Accumulate the result
@@ -265,32 +262,32 @@ fn dense_multiexp_inner<G: CurveAffine>(
                     }
 
                     // acc contains values over this region
-                    s.send(acc).expect("must send result");
+                    // s.send(acc).expect("must send result");
                     
-                    // acc
+                    acc
                 });
         
-                // handles.push(handle);
+                handles.push(handle);
             }
 
-            // // wait for all threads to finish
-            // for r in handles.into_iter() {
-            //     let thread_result = r.join().unwrap();
-            //     this_acc.add_assign(&thread_result);
-            // }
+            // wait for all threads to finish
+            for r in handles.into_iter() {
+                let thread_result = r.join().unwrap();
+                this_acc.add_assign(&thread_result);
+            }
 
 
-            // this_acc
+            this_acc
         });
 
-        let mut this_region = <G as CurveAffine>::Projective::zero();
-        loop {
-            if r.is_empty() {
-                break;
-            }
-            let value = r.recv().expect("must have value");
-            this_region.add_assign(&value);
-        }
+        // let mut this_region = <G as CurveAffine>::Projective::zero();
+        // loop {
+        //     if r.is_empty() {
+        //         break;
+        //     }
+        //     let value = r.recv().expect("must have value");
+        //     this_region.add_assign(&value);
+        // }
 
         this_region
     };
diff --git a/src/parallel_fft.rs b/src/parallel_fft.rs
deleted file mode 100644
index 687d146..0000000
--- a/src/parallel_fft.rs
+++ /dev/null
@@ -1,143 +0,0 @@
-//! This module contains an `EvaluationDomain` abstraction for
-//! performing various kinds of polynomial arithmetic on top of
-//! the scalar field.
-//!
-//! In pairing-based SNARKs like Groth16, we need to calculate
-//! a quotient polynomial over a target polynomial with roots
-//! at distinct points associated with each constraint of the
-//! constraint system. In order to be efficient, we choose these
-//! roots to be the powers of a 2^n root of unity in the field.
-//! This allows us to perform polynomial operations in O(n)
-//! by performing an O(n log n) FFT over such a domain.
-
-use pairing::{
-    Engine,
-    CurveProjective
-};
-
-use ff::{
-    Field, 
-    PrimeField
-};
-
-use super::{
-    SynthesisError
-};
-
-use super::multicore::Worker;
-use super::group::*;
-
-pub(crate) fn best_fft<E: Engine, T: Group<E>>(a: &mut [T], worker: &Worker, omega: &E::Fr, log_n: u32)
-{
-    let log_cpus = worker.log_num_cpus();
-
-    if log_n <= log_cpus {
-        serial_fft(a, omega, log_n);
-    } else {
-        parallel_fft(a, worker, omega, log_n, log_cpus);
-    }
-}
-
-pub(crate) fn serial_fft<E: Engine, T: Group<E>>(a: &mut [T], omega: &E::Fr, log_n: u32)
-{
-    fn bitreverse(mut n: u32, l: u32) -> u32 {
-        let mut r = 0;
-        for _ in 0..l {
-            r = (r << 1) | (n & 1);
-            n >>= 1;
-        }
-        r
-    }
-
-    let n = a.len() as u32;
-    assert_eq!(n, 1 << log_n);
-
-    for k in 0..n {
-        let rk = bitreverse(k, log_n);
-        if k < rk {
-            a.swap(rk as usize, k as usize);
-        }
-    }
-
-    let mut m = 1;
-    for _ in 0..log_n {
-        let w_m = omega.pow(&[(n / (2*m)) as u64]);
-
-        let mut k = 0;
-        while k < n {
-            let mut w = E::Fr::one();
-            for j in 0..m {
-                let mut t = a[(k+j+m) as usize];
-                t.group_mul_assign(&w);
-                let mut tmp = a[(k+j) as usize];
-                tmp.group_sub_assign(&t);
-                a[(k+j+m) as usize] = tmp;
-                a[(k+j) as usize].group_add_assign(&t);
-                w.mul_assign(&w_m);
-            }
-
-            k += 2*m;
-        }
-
-        m *= 2;
-    }
-}
-
-pub(crate) fn parallel_fft<E: Engine, T: Group<E>>(
-    a: &mut [T],
-    worker: &Worker,
-    omega: &E::Fr,
-    log_n: u32,
-    log_cpus: u32
-)
-{
-    assert!(log_n >= log_cpus);
-
-    let num_cpus = 1 << log_cpus;
-    let log_new_n = log_n - log_cpus;
-    let mut tmp = vec![vec![T::group_zero(); 1 << log_new_n]; num_cpus];
-    let new_omega = omega.pow(&[num_cpus as u64]);
-
-    worker.scope(0, |scope, _| {
-        let a = &*a;
-
-        for (j, tmp) in tmp.iter_mut().enumerate() {
-            scope.spawn(move |_| {
-                // Shuffle into a sub-FFT
-                let omega_j = omega.pow(&[j as u64]);
-                let omega_step = omega.pow(&[(j as u64) << log_new_n]);
-
-                let mut elt = E::Fr::one();
-                for i in 0..(1 << log_new_n) {
-                    for s in 0..num_cpus {
-                        let idx = (i + (s << log_new_n)) % (1 << log_n);
-                        let mut t = a[idx];
-                        t.group_mul_assign(&elt);
-                        tmp[i].group_add_assign(&t);
-                        elt.mul_assign(&omega_step);
-                    }
-                    elt.mul_assign(&omega_j);
-                }
-
-                // Perform sub-FFT
-                serial_fft(tmp, &new_omega, log_new_n);
-            });
-        }
-    });
-
-    // TODO: does this hurt or help?
-    worker.scope(a.len(), |scope, chunk| {
-        let tmp = &tmp;
-
-        for (idx, a) in a.chunks_mut(chunk).enumerate() {
-            scope.spawn(move |_| {
-                let mut idx = idx * chunk;
-                let mask = (1 << log_cpus) - 1;
-                for a in a {
-                    *a = tmp[idx & mask][idx >> log_cpus];
-                    idx += 1;
-                }
-            });
-        }
-    });
-}
\ No newline at end of file
diff --git a/src/serial_fft.rs b/src/serial_fft.rs
deleted file mode 100644
index e69de29..0000000
diff --git a/src/serial_multiexp.rs b/src/serial_multiexp.rs
deleted file mode 100644
index e69de29..0000000
diff --git a/src/singlecore.rs b/src/singlecore.rs
new file mode 100644
index 0000000..e669ac6
--- /dev/null
+++ b/src/singlecore.rs
@@ -0,0 +1,94 @@
+//! This is a dummy interface to substitute multicore worker
+//! in environments like WASM
+extern crate futures;
+extern crate futures_cpupool;
+
+use std::marker::PhantomData;
+
+use self::futures::{Future, IntoFuture, Poll};
+use self::futures_cpupool::{CpuFuture, CpuPool};
+
+#[derive(Clone)]
+pub struct Worker {
+    cpus: usize,
+    pool: CpuPool
+}
+
+impl Worker {
+    // We don't expose this outside the library so that
+    // all `Worker` instances have the same number of
+    // CPUs configured.
+    pub(crate) fn new_with_cpus(cpus: usize) -> Worker {
+        Worker {
+            cpus: 1,
+            pool: CpuPool::new(1)
+        }
+    }
+
+    pub fn new() -> Worker {
+        Self::new_with_cpus(1)
+    }
+
+    pub fn log_num_cpus(&self) -> u32 {
+        0u32
+    }
+
+    pub fn compute<F, R>(
+        &self, f: F
+    ) -> WorkerFuture<R::Item, R::Error>
+        where F: FnOnce() -> R + Send + 'static,
+              R: IntoFuture + 'static,
+              R::Future: Send + 'static,
+              R::Item: Send + 'static,
+              R::Error: Send + 'static
+    {
+        WorkerFuture {
+            future: self.pool.spawn_fn(f)
+        }
+    }
+
+    pub fn scope<'a, F, R>(
+        &self,
+        elements: usize,
+        f: F
+    ) -> R
+        where F: FnOnce(&Scope<'a>, usize) -> R
+    {
+        let chunk_size = elements;
+
+        let scope = Scope{
+            _marker: PhantomData
+        };
+
+        f(&scope, chunk_size)
+    }
+}
+#[derive(Clone)]
+pub struct Scope<'a> {
+    _marker: PhantomData<& 'a usize>
+}
+
+impl<'a> Scope<'a> {
+    pub fn spawn<F, R>(
+        &self,
+        f: F
+    ) -> R
+        where F: FnOnce(&Scope<'a>) -> R
+    {
+        f(&self)
+    }
+}
+
+pub struct WorkerFuture<T, E> {
+    future: CpuFuture<T, E>
+}
+
+impl<T: Send + 'static, E: Send + 'static> Future for WorkerFuture<T, E> {
+    type Item = T;
+    type Error = E;
+
+    fn poll(&mut self) -> Poll<Self::Item, Self::Error>
+    {
+        self.future.poll()
+    }
+}
\ No newline at end of file
diff --git a/src/sonic/cs/lc.rs b/src/sonic/cs/lc.rs
index 806eb61..8730473 100644
--- a/src/sonic/cs/lc.rs
+++ b/src/sonic/cs/lc.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine};
 use std::ops::{Add, Sub, Neg};
 
diff --git a/src/sonic/cs/mod.rs b/src/sonic/cs/mod.rs
index a2069e3..0a42dfa 100644
--- a/src/sonic/cs/mod.rs
+++ b/src/sonic/cs/mod.rs
@@ -1,7 +1,6 @@
-extern crate ff;
 extern crate pairing;
 
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine};
 
 use crate::{SynthesisError};
diff --git a/src/sonic/helped/adapted_helper.rs b/src/sonic/helped/adapted_helper.rs
index 87e99da..0243298 100644
--- a/src/sonic/helped/adapted_helper.rs
+++ b/src/sonic/helped/adapted_helper.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/helped/adapted_prover.rs b/src/sonic/helped/adapted_prover.rs
index 2811db4..68f9565 100644
--- a/src/sonic/helped/adapted_prover.rs
+++ b/src/sonic/helped/adapted_prover.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 
@@ -97,7 +97,7 @@ pub fn create_proof_on_srs<E: Engine, C: Circuit<E> + Clone>(
 
 // #[test]
 // fn my_fun_circuit_test() {
-//     use ff::PrimeField;
+//     use pairing::ff::PrimeField;
 //     use pairing::bls12_381::{Bls12, Fr};
 //     use super::*;
 //     use crate::sonic::cs::{Basic, ConstraintSystem, LinearCombination};
diff --git a/src/sonic/helped/adapted_verifier.rs b/src/sonic/helped/adapted_verifier.rs
index 0c4b0c8..0a4b159 100644
--- a/src/sonic/helped/adapted_verifier.rs
+++ b/src/sonic/helped/adapted_verifier.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 
@@ -53,7 +53,7 @@ pub fn verify_aggregate<E: Engine, C: Circuit<E> + Clone, R: Rng>(
 
 // #[test]
 // fn my_fun_circuit_test() {
-//     use ff::PrimeField;
+//     use pairing::ff::PrimeField;
 //     use pairing::bls12_381::{Bls12, Fr};
 //     use super::*;
 //     use crate::sonic::cs::{Basic, ConstraintSystem, LinearCombination};
diff --git a/src/sonic/helped/batch.rs b/src/sonic/helped/batch.rs
index 9af97c2..3cec29c 100644
--- a/src/sonic/helped/batch.rs
+++ b/src/sonic/helped/batch.rs
@@ -8,7 +8,7 @@
 //! This submodule contains the `Batch` abstraction for creating a
 //! context for batch verification.
 
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveAffine, CurveProjective};
 
 use crate::SynthesisError;
diff --git a/src/sonic/helped/generator.rs b/src/sonic/helped/generator.rs
index a1fb6de..652c501 100644
--- a/src/sonic/helped/generator.rs
+++ b/src/sonic/helped/generator.rs
@@ -9,7 +9,7 @@ use pairing::{
     CurveAffine
 };
 
-use ff::{    
+use pairing::ff::{    
     PrimeField,
     Field
 };
@@ -19,7 +19,7 @@ use super::{
     VerifyingKey
 };
 
-use ::{
+use crate::{
     SynthesisError,
     Circuit,
     ConstraintSystem,
@@ -32,7 +32,7 @@ use crate::domain::{
     Scalar
 };
 
-use ::multicore::{
+use crate::multicore::{
     Worker
 };
 
diff --git a/src/sonic/helped/helper.rs b/src/sonic/helped/helper.rs
index 5b11f11..a8a853a 100644
--- a/src/sonic/helped/helper.rs
+++ b/src/sonic/helped/helper.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/helped/mod.rs b/src/sonic/helped/mod.rs
index 3a0be60..0135b66 100644
--- a/src/sonic/helped/mod.rs
+++ b/src/sonic/helped/mod.rs
@@ -1,7 +1,6 @@
-extern crate ff;
 extern crate pairing;
 
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/helped/parameters.rs b/src/sonic/helped/parameters.rs
index 488c186..2e000e6 100644
--- a/src/sonic/helped/parameters.rs
+++ b/src/sonic/helped/parameters.rs
@@ -1,4 +1,4 @@
-use ff::{
+use pairing::ff::{
     Field,
     PrimeField, 
     PrimeFieldRepr
@@ -10,11 +10,11 @@ use pairing::{
     EncodedPoint
 };
 
-use ::{
+use crate::{
     SynthesisError
 };
 
-use multiexp::SourceBuilder;
+use crate::source::SourceBuilder;
 use std::io::{self, Read, Write};
 use std::sync::Arc;
 use byteorder::{BigEndian, WriteBytesExt, ReadBytesExt};
@@ -64,7 +64,7 @@ impl<E: Engine> Proof<E> {
         mut writer: W
     ) -> io::Result<()>
     {
-        use ff::{PrimeField, PrimeFieldRepr};
+        use pairing::ff::{PrimeField, PrimeFieldRepr};
         writer.write_all(self.r.into_compressed().as_ref())?;
         writer.write_all(self.t.into_compressed().as_ref())?;
         let mut buffer = vec![];
diff --git a/src/sonic/helped/poly.rs b/src/sonic/helped/poly.rs
index 6001061..22ce7e7 100644
--- a/src/sonic/helped/poly.rs
+++ b/src/sonic/helped/poly.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/helped/prover.rs b/src/sonic/helped/prover.rs
index 79cc0ec..f3f4645 100644
--- a/src/sonic/helped/prover.rs
+++ b/src/sonic/helped/prover.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 
@@ -397,7 +397,7 @@ pub fn create_proof_on_srs<E: Engine, C: Circuit<E>, S: SynthesisDriver>(
 
 #[test]
 fn my_fun_circuit_test() {
-    use ff::PrimeField;
+    use pairing::ff::PrimeField;
     use pairing::bls12_381::{Bls12, Fr};
     use super::*;
     use crate::sonic::cs::{Basic, ConstraintSystem, LinearCombination};
@@ -449,8 +449,8 @@ fn my_fun_circuit_test() {
 
 #[test]
 fn polynomial_commitment_test() {
-    use ff::PrimeField;
-    use ff::PrimeFieldRepr;
+    use pairing::ff::PrimeField;
+    use pairing::ff::PrimeFieldRepr;
     use pairing::bls12_381::{Bls12, Fr};
     use super::*;
     use crate::sonic::cs::{Basic, ConstraintSystem, LinearCombination};
diff --git a/src/sonic/helped/verifier.rs b/src/sonic/helped/verifier.rs
index 2bded1b..9a520b7 100644
--- a/src/sonic/helped/verifier.rs
+++ b/src/sonic/helped/verifier.rs
@@ -1,4 +1,4 @@
-use ff::{Field};
+use pairing::ff::{Field};
 use pairing::{Engine, CurveProjective};
 use std::marker::PhantomData;
 use rand::{Rand, Rng};
diff --git a/src/sonic/mod.rs b/src/sonic/mod.rs
index 6786a7f..2241a62 100644
--- a/src/sonic/mod.rs
+++ b/src/sonic/mod.rs
@@ -1,4 +1,3 @@
-extern crate ff;
 extern crate pairing;
 
 pub use crate::{SynthesisError};
diff --git a/src/sonic/sonic/adaptor.rs b/src/sonic/sonic/adaptor.rs
index f38eaed..87cd9e7 100644
--- a/src/sonic/sonic/adaptor.rs
+++ b/src/sonic/sonic/adaptor.rs
@@ -1,8 +1,7 @@
-extern crate ff;
 extern crate pairing;
 extern crate rand;
 
-use ff::{Field, PrimeField};
+use pairing::ff::{Field, PrimeField};
 use pairing::{Engine, CurveProjective};
 
 // this one is for all external interfaces
diff --git a/src/sonic/sonic/mod.rs b/src/sonic/sonic/mod.rs
index 8b3e220..aa16edb 100644
--- a/src/sonic/sonic/mod.rs
+++ b/src/sonic/sonic/mod.rs
@@ -1,4 +1,3 @@
-extern crate ff;
 extern crate pairing;
 
 mod adaptor;
diff --git a/src/sonic/srs/mod.rs b/src/sonic/srs/mod.rs
index d2eae6e..e93d24f 100644
--- a/src/sonic/srs/mod.rs
+++ b/src/sonic/srs/mod.rs
@@ -1,4 +1,3 @@
-extern crate ff;
 extern crate pairing;
 
 mod srs;
diff --git a/src/sonic/srs/srs.rs b/src/sonic/srs/srs.rs
index b8d5831..7833ab8 100644
--- a/src/sonic/srs/srs.rs
+++ b/src/sonic/srs/srs.rs
@@ -1,4 +1,4 @@
-use ff::{Field, PrimeField};
+use pairing::ff::{Field, PrimeField};
 use pairing::{CurveAffine, CurveProjective, Engine, Wnaf};
 
 use std::io::{self, Read, Write};
diff --git a/src/sonic/transcript/mod.rs b/src/sonic/transcript/mod.rs
index fc5035f..4525030 100644
--- a/src/sonic/transcript/mod.rs
+++ b/src/sonic/transcript/mod.rs
@@ -1,7 +1,6 @@
-extern crate ff;
 extern crate pairing;
 
-use ff::{Field, PrimeField, PrimeFieldRepr};
+use pairing::ff::{Field, PrimeField, PrimeFieldRepr};
 use pairing::{CurveAffine, CurveProjective, Engine};
 use std::io;
 
diff --git a/src/sonic/unhelped/grand_product_argument.rs b/src/sonic/unhelped/grand_product_argument.rs
index dad9505..61c914d 100644
--- a/src/sonic/unhelped/grand_product_argument.rs
+++ b/src/sonic/unhelped/grand_product_argument.rs
@@ -2,7 +2,7 @@
 /// in those two polynomials are equal (part of the permutation argument) with additional assumption that
 /// those coefficients are never equal to zero
 
-use ff::{Field, PrimeField, PrimeFieldRepr};
+use pairing::ff::{Field, PrimeField, PrimeFieldRepr};
 use pairing::{Engine, CurveProjective, CurveAffine};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/unhelped/permutation_argument.rs b/src/sonic/unhelped/permutation_argument.rs
index 96c6039..1813ae6 100644
--- a/src/sonic/unhelped/permutation_argument.rs
+++ b/src/sonic/unhelped/permutation_argument.rs
@@ -2,7 +2,7 @@
 /// actually a commitment to a vector of values that are equal to `(s^{perm})_i * y^{perm(i)}`
 /// for some fixed permutation `perm`
 
-use ff::{Field, PrimeField, PrimeFieldRepr, ScalarEngine};
+use pairing::ff::{Field, PrimeField, PrimeFieldRepr, ScalarEngine};
 use pairing::{Engine, CurveProjective, CurveAffine};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/unhelped/s2_proof.rs b/src/sonic/unhelped/s2_proof.rs
index 72f9eb7..77bb8a8 100644
--- a/src/sonic/unhelped/s2_proof.rs
+++ b/src/sonic/unhelped/s2_proof.rs
@@ -1,4 +1,4 @@
-use ff::{Field, PrimeField, PrimeFieldRepr};
+use pairing::ff::{Field, PrimeField, PrimeFieldRepr};
 use pairing::{Engine, CurveProjective, CurveAffine};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/unhelped/wellformed_argument.rs b/src/sonic/unhelped/wellformed_argument.rs
index 5e707f1..a321278 100644
--- a/src/sonic/unhelped/wellformed_argument.rs
+++ b/src/sonic/unhelped/wellformed_argument.rs
@@ -1,7 +1,7 @@
 /// Wellformedness argument allows to verify that some committment was to multivariate polynomial of degree n, 
 /// with no constant term and negative powers
 
-use ff::{Field, PrimeField, PrimeFieldRepr};
+use pairing::ff::{Field, PrimeField, PrimeFieldRepr};
 use pairing::{Engine, CurveProjective, CurveAffine};
 use std::marker::PhantomData;
 
diff --git a/src/sonic/util.rs b/src/sonic/util.rs
index 9e9f6b1..9f3e055 100644
--- a/src/sonic/util.rs
+++ b/src/sonic/util.rs
@@ -1,5 +1,5 @@
 use crate::SynthesisError;
-use ff::{Field, PrimeField, PrimeFieldRepr, ScalarEngine};
+use pairing::ff::{Field, PrimeField, PrimeFieldRepr, ScalarEngine};
 use pairing::{CurveAffine, CurveProjective, Engine};
 use super::srs::SRS;
 
@@ -437,7 +437,7 @@ pub fn check_polynomial_commitment<E: Engine>(
 
 #[test]
 fn laurent_division() {
-    use ff::PrimeField;
+    use pairing::ff::PrimeField;
     use pairing::bls12_381::{Fr};
 
     let mut poly = vec![
diff --git a/src/source.rs b/src/source.rs
index 5a77d8f..8fa8a94 100644
--- a/src/source.rs
+++ b/src/source.rs
@@ -4,7 +4,7 @@ use pairing::{
     Engine
 };
 
-use ff::{
+use pairing::ff::{
     PrimeField,
     Field,
     PrimeFieldRepr,
diff --git a/src/tests/dummy_engine.rs b/src/tests/dummy_engine.rs
index a436533..0cc5cdc 100644
--- a/src/tests/dummy_engine.rs
+++ b/src/tests/dummy_engine.rs
@@ -6,7 +6,7 @@ use pairing::{
     EncodedPoint
 };
 
-use ff::{
+use pairing::ff::{
     PrimeField,
     PrimeFieldRepr,
     Field,
diff --git a/src/tests/mod.rs b/src/tests/mod.rs
index f9f6ce9..e72f954 100644
--- a/src/tests/mod.rs
+++ b/src/tests/mod.rs
@@ -2,7 +2,7 @@ use pairing::{
     Engine
 };
 
-use ff:: {
+use pairing::ff:: {
     Field,
     PrimeField,
 };
@@ -12,7 +12,7 @@ use self::dummy_engine::*;
 
 use std::marker::PhantomData;
 
-use ::{
+use crate::{
     Circuit,
     ConstraintSystem,
     SynthesisError
diff --git a/tests/mimc.rs b/tests/mimc.rs
index c0dce24..9446145 100644
--- a/tests/mimc.rs
+++ b/tests/mimc.rs
@@ -1,7 +1,6 @@
 extern crate bellman;
 extern crate pairing;
 extern crate rand;
-extern crate ff;
 
 // For randomness (during paramgen and proof generation)
 use rand::{thread_rng, Rng};
@@ -14,7 +13,7 @@ use pairing::{
     Engine  
 };
 
-use ff::{
+use pairing::ff::{
     Field,
 };
 
@@ -442,7 +441,7 @@ impl<'a, E: Engine> Circuit<E> for MiMCDemoNoInputs<'a, E> {
 
 #[test]
 fn test_sonic_mimc() {
-    use ff::{Field, PrimeField};
+    use pairing::ff::{Field, PrimeField};
     use pairing::{Engine, CurveAffine, CurveProjective};
     use pairing::bls12_381::{Bls12, Fr};
     use std::time::{Instant};
@@ -546,7 +545,7 @@ fn test_sonic_mimc() {
 
 #[test]
 fn test_inputs_into_sonic_mimc() {
-    use ff::{Field, PrimeField};
+    use pairing::ff::{Field, PrimeField};
     use pairing::{Engine, CurveAffine, CurveProjective};
     use pairing::bn256::{Bn256, Fr};
     // use pairing::bls12_381::{Bls12, Fr};