perf: add inplace decimal operations (#11004)

closes #9046

* add mutable decimal methods

* fix root

* gofmt

* add Clone(), rm assignment quo

* add SetInt64

* fix SetInt64

* More mut speedups

Co-authored-by: mconcat <monoidconcat@gmail.com>
Co-authored-by: ValarDragon <dojha12@gmail.com>




---

### Author Checklist

*All items are required. Please add a note to the item if the item is not applicable and
please add links to any relevant follow up issues.*

I have...

- [ ] included the correct [type prefix](https://github.com/commitizen/conventional-commit-types/blob/v3.0.0/index.json) in the PR title
- [ ] added `!` to the type prefix if API or client breaking change
- [ ] targeted the correct branch (see [PR Targeting](https://github.com/cosmos/cosmos-sdk/blob/master/CONTRIBUTING.md#pr-targeting))
- [ ] provided a link to the relevant issue or specification
- [ ] followed the guidelines for [building modules](https://github.com/cosmos/cosmos-sdk/blob/master/docs/building-modules)
- [ ] included the necessary unit and integration [tests](https://github.com/cosmos/cosmos-sdk/blob/master/CONTRIBUTING.md#testing)
- [ ] added a changelog entry to `CHANGELOG.md`
- [ ] included comments for [documenting Go code](https://blog.golang.org/godoc)
- [ ] updated the relevant documentation or specification
- [ ] reviewed "Files changed" and left comments if necessary
- [ ] confirmed all CI checks have passed

### Reviewers Checklist

*All items are required. Please add a note if the item is not applicable and please add
your handle next to the items reviewed if you only reviewed selected items.*

I have...

- [ ] confirmed the correct [type prefix](https://github.com/commitizen/conventional-commit-types/blob/v3.0.0/index.json) in the PR title
- [ ] confirmed `!` in the type prefix if API or client breaking change
- [ ] confirmed all author checklist items have been addressed 
- [ ] reviewed state machine logic
- [ ] reviewed API design and naming
- [ ] reviewed documentation is accurate
- [ ] reviewed tests and test coverage
- [ ] manually tested (if applicable)
This commit is contained in:
Sunny Aggarwal 2022-02-11 01:55:41 -08:00 committed by GitHub
parent 2c184d2056
commit 2e1c32834e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 158 additions and 63 deletions

View File

@ -160,6 +160,7 @@ Ref: https://keepachangelog.com/en/1.0.0/
* (deps) [\#10706](https://github.com/cosmos/cosmos-sdk/issues/10706) Bump rosetta-sdk-go to v0.7.2 and rosetta-cli to v0.7.3
* (types/errors) [\#10779](https://github.com/cosmos/cosmos-sdk/pull/10779) Move most functionality in `types/errors` to a standalone `errors` go module, except the `RootCodespace` errors and ABCI response helpers. All functions and types that used to live in `types/errors` are now aliased so this is not a breaking change.
* (gov) [\#10854](https://github.com/cosmos/cosmos-sdk/pull/10854) v1beta2's vote doesn't include the deprecate `option VoteOption` anymore. Instead, it only uses `WeightedVoteOption`.
* (types) [\#11004](https://github.com/cosmos/cosmos-sdk/pull/11004) Added mutable versions of many of the sdk.Dec types operations. This improves performance when used by avoiding reallocating a new bigint for each operation.
### Bug Fixes

View File

@ -208,7 +208,10 @@ func (d Dec) GTE(d2 Dec) bool { return (d.i).Cmp(d2.i) >= 0 } // greater
func (d Dec) LT(d2 Dec) bool { return (d.i).Cmp(d2.i) < 0 } // less than
func (d Dec) LTE(d2 Dec) bool { return (d.i).Cmp(d2.i) <= 0 } // less than or equal
func (d Dec) Neg() Dec { return Dec{new(big.Int).Neg(d.i)} } // reverse the decimal sign
func (d Dec) NegMut() Dec { d.i.Neg(d.i); return d } // reverse the decimal sign, mutable
func (d Dec) Abs() Dec { return Dec{new(big.Int).Abs(d.i)} } // absolute value
func (d Dec) Set(d2 Dec) Dec { d.i.Set(d2.i); return d } // set to existing dec value
func (d Dec) Clone() Dec { return Dec{new(big.Int).Set(d.i)} } // clone new dec
// BigInt returns a copy of the underlying big.Int.
func (d Dec) BigInt() *big.Int {
@ -220,123 +223,192 @@ func (d Dec) BigInt() *big.Int {
return cp.Set(d.i)
}
func (d Dec) ImmutOp(op func(Dec, Dec) Dec, d2 Dec) Dec {
return op(d.Clone(), d2)
}
func (d Dec) ImmutOpInt(op func(Dec, Int) Dec, d2 Int) Dec {
return op(d.Clone(), d2)
}
func (d Dec) ImmutOpInt64(op func(Dec, int64) Dec, d2 int64) Dec {
// TODO: use already allocated operand bigint to avoid
// newint each time, add mutex for race condition
// Issue: https://github.com/cosmos/cosmos-sdk/issues/11166
return op(d.Clone(), d2)
}
func (d Dec) SetInt64(i int64) Dec {
d.i.SetInt64(i)
d.i.Mul(d.i, precisionReuse)
return d
}
// addition
func (d Dec) Add(d2 Dec) Dec {
res := new(big.Int).Add(d.i, d2.i)
return d.ImmutOp(Dec.AddMut, d2)
}
if res.BitLen() > maxDecBitLen {
// mutable addition
func (d Dec) AddMut(d2 Dec) Dec {
d.i.Add(d.i, d2.i)
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{res}
return d
}
// subtraction
func (d Dec) Sub(d2 Dec) Dec {
res := new(big.Int).Sub(d.i, d2.i)
return d.ImmutOp(Dec.SubMut, d2)
}
if res.BitLen() > maxDecBitLen {
// mutable subtraction
func (d Dec) SubMut(d2 Dec) Dec {
d.i.Sub(d.i, d2.i)
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{res}
return d
}
// multiplication
func (d Dec) Mul(d2 Dec) Dec {
mul := new(big.Int).Mul(d.i, d2.i)
chopped := chopPrecisionAndRound(mul)
return d.ImmutOp(Dec.MulMut, d2)
}
// mutable multiplication
func (d Dec) MulMut(d2 Dec) Dec {
d.i.Mul(d.i, d2.i)
chopped := chopPrecisionAndRound(d.i)
if chopped.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{chopped}
*d.i = *chopped
return d
}
// multiplication truncate
func (d Dec) MulTruncate(d2 Dec) Dec {
mul := new(big.Int).Mul(d.i, d2.i)
chopped := chopPrecisionAndTruncate(mul)
return d.ImmutOp(Dec.MulTruncateMut, d2)
}
if chopped.BitLen() > maxDecBitLen {
// mutable multiplication truncage
func (d Dec) MulTruncateMut(d2 Dec) Dec {
d.i.Mul(d.i, d2.i)
chopPrecisionAndTruncate(d.i)
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{chopped}
return d
}
// multiplication
func (d Dec) MulInt(i Int) Dec {
mul := new(big.Int).Mul(d.i, i.i)
return d.ImmutOpInt(Dec.MulIntMut, i)
}
if mul.BitLen() > maxDecBitLen {
func (d Dec) MulIntMut(i Int) Dec {
d.i.Mul(d.i, i.i)
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{mul}
return d
}
// MulInt64 - multiplication with int64
func (d Dec) MulInt64(i int64) Dec {
mul := new(big.Int).Mul(d.i, big.NewInt(i))
return d.ImmutOpInt64(Dec.MulInt64Mut, i)
}
if mul.BitLen() > maxDecBitLen {
func (d Dec) MulInt64Mut(i int64) Dec {
d.i.Mul(d.i, big.NewInt(i))
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{mul}
return d
}
// quotient
func (d Dec) Quo(d2 Dec) Dec {
return d.ImmutOp(Dec.QuoMut, d2)
}
// mutable quotient
func (d Dec) QuoMut(d2 Dec) Dec {
// multiply precision twice
mul := new(big.Int).Mul(d.i, precisionReuse)
mul.Mul(mul, precisionReuse)
d.i.Mul(d.i, precisionReuse)
d.i.Mul(d.i, precisionReuse)
d.i.Quo(d.i, d2.i)
quo := new(big.Int).Quo(mul, d2.i)
chopped := chopPrecisionAndRound(quo)
if chopped.BitLen() > maxDecBitLen {
chopPrecisionAndRound(d.i)
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{chopped}
return d
}
// quotient truncate
func (d Dec) QuoTruncate(d2 Dec) Dec {
return d.ImmutOp(Dec.QuoTruncateMut, d2)
}
// mutable quotient truncate
func (d Dec) QuoTruncateMut(d2 Dec) Dec {
// multiply precision twice
mul := new(big.Int).Mul(d.i, precisionReuse)
mul.Mul(mul, precisionReuse)
d.i.Mul(d.i, precisionReuse)
d.i.Mul(d.i, precisionReuse)
d.i.Quo(d.i, d2.i)
quo := mul.Quo(mul, d2.i)
chopped := chopPrecisionAndTruncate(quo)
if chopped.BitLen() > maxDecBitLen {
chopPrecisionAndTruncate(d.i)
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{chopped}
return d
}
// quotient, round up
func (d Dec) QuoRoundUp(d2 Dec) Dec {
return d.ImmutOp(Dec.QuoRoundupMut, d2)
}
// mutable quotient, round up
func (d Dec) QuoRoundupMut(d2 Dec) Dec {
// multiply precision twice
mul := new(big.Int).Mul(d.i, precisionReuse)
mul.Mul(mul, precisionReuse)
d.i.Mul(d.i, precisionReuse)
d.i.Mul(d.i, precisionReuse)
d.i.Quo(d.i, d2.i)
quo := new(big.Int).Quo(mul, d2.i)
chopped := chopPrecisionAndRoundUp(quo)
if chopped.BitLen() > maxDecBitLen {
chopPrecisionAndRoundUp(d.i)
if d.i.BitLen() > maxDecBitLen {
panic("Int overflow")
}
return Dec{chopped}
return d
}
// quotient
func (d Dec) QuoInt(i Int) Dec {
mul := new(big.Int).Quo(d.i, i.i)
return Dec{mul}
return d.ImmutOpInt(Dec.QuoIntMut, i)
}
func (d Dec) QuoIntMut(i Int) Dec {
d.i.Quo(d.i, i.i)
return d
}
// QuoInt64 - quotient with int64
func (d Dec) QuoInt64(i int64) Dec {
mul := new(big.Int).Quo(d.i, big.NewInt(i))
return Dec{mul}
return d.ImmutOpInt64(Dec.QuoInt64Mut, i)
}
func (d Dec) QuoInt64Mut(i int64) Dec {
d.i.Quo(d.i, big.NewInt(i))
return d
}
// ApproxRoot returns an approximate estimation of a Dec's positive real nth root
@ -357,8 +429,8 @@ func (d Dec) ApproxRoot(root uint64) (guess Dec, err error) {
}()
if d.IsNegative() {
absRoot, err := d.MulInt64(-1).ApproxRoot(root)
return absRoot.MulInt64(-1), err
absRoot, err := d.Neg().ApproxRoot(root)
return absRoot.NegMut(), err
}
if root == 1 || d.IsZero() || d.Equal(OneDec()) {
@ -369,7 +441,6 @@ func (d Dec) ApproxRoot(root uint64) (guess Dec, err error) {
return OneDec(), nil
}
rootInt := NewIntFromUint64(root)
guess, delta := OneDec(), OneDec()
for iter := 0; delta.Abs().GT(SmallestDec()) && iter < maxApproxRootIterations; iter++ {
@ -377,11 +448,11 @@ func (d Dec) ApproxRoot(root uint64) (guess Dec, err error) {
if prev.IsZero() {
prev = SmallestDec()
}
delta = d.Quo(prev)
delta = delta.Sub(guess)
delta = delta.QuoInt(rootInt)
delta.Set(d).QuoMut(prev)
delta.SubMut(guess)
delta.QuoInt64Mut(int64(root))
guess = guess.Add(delta)
guess.AddMut(delta)
}
return guess, nil
@ -389,20 +460,26 @@ func (d Dec) ApproxRoot(root uint64) (guess Dec, err error) {
// Power returns a the result of raising to a positive integer power
func (d Dec) Power(power uint64) Dec {
res := Dec{new(big.Int).Set(d.i)}
return res.PowerMut(power)
}
func (d Dec) PowerMut(power uint64) Dec {
if power == 0 {
return OneDec()
d.SetInt64(1)
return d
}
tmp := OneDec()
for i := power; i > 1; {
if i%2 != 0 {
tmp = tmp.Mul(d)
tmp.MulMut(d)
}
i /= 2
d = d.Mul(d)
d.MulMut(d)
}
return d.Mul(tmp)
return d.MulMut(tmp)
}
// ApproxSqrt is a wrapper around ApproxRoot for the common special case
@ -543,7 +620,7 @@ func chopPrecisionAndRoundUp(d *big.Int) *big.Int {
// make d positive, compute chopped value, and then un-mutate d
d = d.Neg(d)
// truncate since d is negative...
d = chopPrecisionAndTruncate(d)
chopPrecisionAndTruncate(d)
d = d.Neg(d)
return d
}
@ -580,13 +657,19 @@ func (d Dec) RoundInt() Int {
// chopPrecisionAndTruncate is similar to chopPrecisionAndRound,
// but always rounds down. It does not mutate the input.
func chopPrecisionAndTruncate(d *big.Int) *big.Int {
return new(big.Int).Quo(d, precisionReuse)
func chopPrecisionAndTruncate(d *big.Int) {
d.Quo(d, precisionReuse)
}
func chopPrecisionAndTruncateNonMutative(d *big.Int) *big.Int {
tmp := new(big.Int).Set(d)
chopPrecisionAndTruncate(tmp)
return tmp
}
// TruncateInt64 truncates the decimals from the number and returns an int64
func (d Dec) TruncateInt64() int64 {
chopped := chopPrecisionAndTruncate(d.i)
chopped := chopPrecisionAndTruncateNonMutative(d.i)
if !chopped.IsInt64() {
panic("Int64() out of bound")
}
@ -595,12 +678,12 @@ func (d Dec) TruncateInt64() int64 {
// TruncateInt truncates the decimals from the number and returns an Int
func (d Dec) TruncateInt() Int {
return NewIntFromBigInt(chopPrecisionAndTruncate(d.i))
return NewIntFromBigInt(chopPrecisionAndTruncateNonMutative(d.i))
}
// TruncateDec truncates the decimals from the number and returns a Dec
func (d Dec) TruncateDec() Dec {
return NewDecFromBigInt(chopPrecisionAndTruncate(d.i))
return NewDecFromBigInt(chopPrecisionAndTruncateNonMutative(d.i))
}
// Ceil returns the smallest interger value (as a decimal) that is greater than
@ -625,7 +708,11 @@ func (d Dec) Ceil() Dec {
// MaxSortableDec is the largest Dec that can be passed into SortableDecBytes()
// Its negative form is the least Dec that can be passed in.
var MaxSortableDec = OneDec().Quo(SmallestDec())
var MaxSortableDec Dec
func init() {
MaxSortableDec = OneDec().Quo(SmallestDec())
}
// ValidSortableDec ensures that a Dec is within the sortable bounds,
// a Dec can't have a precision of less than 10^-18.

View File

@ -376,6 +376,7 @@ func (s *decimalTestSuite) TestPower() {
power uint64
expected sdk.Dec
}{
{sdk.NewDec(100), 0, sdk.OneDec()}, // 10 ^ (0) => 1.0
{sdk.OneDec(), 10, sdk.OneDec()}, // 1.0 ^ (10) => 1.0
{sdk.NewDecWithPrec(5, 1), 2, sdk.NewDecWithPrec(25, 2)}, // 0.5 ^ 2 => 0.25
{sdk.NewDecWithPrec(2, 1), 2, sdk.NewDecWithPrec(4, 2)}, // 0.2 ^ 2 => 0.04
@ -386,7 +387,13 @@ func (s *decimalTestSuite) TestPower() {
for i, tc := range testCases {
res := tc.input.Power(tc.power)
s.Require().True(tc.expected.Sub(res).Abs().LTE(sdk.SmallestDec()), "unexpected result for test case %d, input: %v", i, tc.input)
s.Require().True(tc.expected.Sub(res).Abs().LTE(sdk.SmallestDec()), "unexpected result for test case %d, normal power, input: %v", i, tc.input)
mutableInput := tc.input
mutableInput.PowerMut(tc.power)
s.Require().True(tc.expected.Sub(mutableInput).Abs().LTE(sdk.SmallestDec()),
"unexpected result for test case %d, input %v", i, tc.input)
s.Require().True(res.Equal(tc.input), "unexpected result for test case %d, mutable power, input: %v", i, tc.input)
}
}