optimized intsets to not allocate for the common cases

This commit is contained in:
Andreas Rumpf 2017-07-25 09:50:49 +02:00
commit 1539d9d95b

View file

@ -31,16 +31,18 @@ const
type type
PTrunk = ref Trunk PTrunk = ref Trunk
Trunk {.final.} = object Trunk = object
next: PTrunk # all nodes are connected with this pointer next: PTrunk # all nodes are connected with this pointer
key: int # start address at bit 0 key: int # start address at bit 0
bits: array[0..IntsPerTrunk - 1, BitScalar] # a bit vector bits: array[0..IntsPerTrunk - 1, BitScalar] # a bit vector
TrunkSeq = seq[PTrunk] TrunkSeq = seq[PTrunk]
IntSet* = object ## an efficient set of 'int' implemented as a sparse bit set IntSet* = object ## an efficient set of 'int' implemented as a sparse bit set
elems: int # only valid for small numbers
counter, max: int counter, max: int
head: PTrunk head: PTrunk
data: TrunkSeq data: TrunkSeq
a: array[0..33, int] # profiling shows that 34 elements are enough
{.deprecated: [TIntSet: IntSet, TTrunk: Trunk, TTrunkSeq: TrunkSeq].} {.deprecated: [TIntSet: IntSet, TTrunk: Trunk, TTrunkSeq: TrunkSeq].}
@ -95,6 +97,10 @@ proc intSetPut(t: var IntSet, key: int): PTrunk =
proc contains*(s: IntSet, key: int): bool = proc contains*(s: IntSet, key: int): bool =
## returns true iff `key` is in `s`. ## returns true iff `key` is in `s`.
if s.elems <= s.a.len:
for i in 0..<s.elems:
if s.a[i] == key: return true
else:
var t = intSetGet(s, `shr`(key, TrunkShift)) var t = intSetGet(s, `shr`(key, TrunkShift))
if t != nil: if t != nil:
var u = key and TrunkMask var u = key and TrunkMask
@ -102,15 +108,38 @@ proc contains*(s: IntSet, key: int): bool =
else: else:
result = false result = false
proc incl*(s: var IntSet, key: int) = proc bitincl(s: var IntSet, key: int) {.inline.} =
## includes an element `key` in `s`.
var t = intSetPut(s, `shr`(key, TrunkShift)) var t = intSetPut(s, `shr`(key, TrunkShift))
var u = key and TrunkMask var u = key and TrunkMask
t.bits[`shr`(u, IntShift)] = t.bits[`shr`(u, IntShift)] or t.bits[`shr`(u, IntShift)] = t.bits[`shr`(u, IntShift)] or
`shl`(1, u and IntMask) `shl`(1, u and IntMask)
proc incl*(s: var IntSet, key: int) =
## includes an element `key` in `s`.
if s.elems <= s.a.len:
for i in 0..<s.elems:
if s.a[i] == key: return
if s.elems < s.a.len:
s.a[s.elems] = key
inc s.elems
return
newSeq(s.data, InitIntSetSize)
s.max = InitIntSetSize-1
for i in 0..<s.elems:
bitincl(s, s.a[i])
s.elems = s.a.len + 1
# fall through:
bitincl(s, key)
proc excl*(s: var IntSet, key: int) = proc excl*(s: var IntSet, key: int) =
## excludes `key` from the set `s`. ## excludes `key` from the set `s`.
if s.elems <= s.a.len:
for i in 0..<s.elems:
if s.a[i] == key:
s.a[i] = s.a[s.elems-1]
dec s.elems
return
else:
var t = intSetGet(s, `shr`(key, TrunkShift)) var t = intSetGet(s, `shr`(key, TrunkShift))
if t != nil: if t != nil:
var u = key and TrunkMask var u = key and TrunkMask
@ -120,6 +149,13 @@ proc excl*(s: var IntSet, key: int) =
proc containsOrIncl*(s: var IntSet, key: int): bool = proc containsOrIncl*(s: var IntSet, key: int): bool =
## returns true if `s` contains `key`, otherwise `key` is included in `s` ## returns true if `s` contains `key`, otherwise `key` is included in `s`
## and false is returned. ## and false is returned.
if s.elems <= s.a.len:
for i in 0..<s.elems:
if s.a[i] == key:
return true
incl(s, key)
result = false
else:
var t = intSetGet(s, `shr`(key, TrunkShift)) var t = intSetGet(s, `shr`(key, TrunkShift))
if t != nil: if t != nil:
var u = key and TrunkMask var u = key and TrunkMask
@ -133,23 +169,38 @@ proc containsOrIncl*(s: var IntSet, key: int): bool =
proc initIntSet*: IntSet = proc initIntSet*: IntSet =
## creates a new int set that is empty. ## creates a new int set that is empty.
newSeq(result.data, InitIntSetSize)
result.max = InitIntSetSize-1 #newSeq(result.data, InitIntSetSize)
#result.max = InitIntSetSize-1
result.data = nil
result.max = 0
result.counter = 0 result.counter = 0
result.head = nil result.head = nil
result.elems = 0
proc clear*(result: var IntSet) = proc clear*(result: var IntSet) =
setLen(result.data, InitIntSetSize) #setLen(result.data, InitIntSetSize)
for i in 0..InitIntSetSize-1: result.data[i] = nil #for i in 0..InitIntSetSize-1: result.data[i] = nil
result.max = InitIntSetSize-1 #result.max = InitIntSetSize-1
result.data = nil
result.max = 0
result.counter = 0 result.counter = 0
result.head = nil result.head = nil
result.elems = 0
proc isNil*(x: IntSet): bool {.inline.} = x.head.isNil proc isNil*(x: IntSet): bool {.inline.} = x.head.isNil and x.elems == 0
proc assign*(dest: var IntSet, src: IntSet) = proc assign*(dest: var IntSet, src: IntSet) =
## copies `src` to `dest`. `dest` does not need to be initialized by ## copies `src` to `dest`. `dest` does not need to be initialized by
## `initIntSet`. ## `initIntSet`.
if src.elems <= src.a.len:
dest.data = nil
dest.max = 0
dest.counter = src.counter
dest.head = nil
dest.elems = src.elems
dest.a = src.a
else:
dest.counter = src.counter dest.counter = src.counter
dest.max = src.max dest.max = src.max
newSeq(dest.data, src.data.len) newSeq(dest.data, src.data.len)
@ -173,6 +224,10 @@ proc assign*(dest: var IntSet, src: IntSet) =
iterator items*(s: IntSet): int {.inline.} = iterator items*(s: IntSet): int {.inline.} =
## iterates over any included element of `s`. ## iterates over any included element of `s`.
if s.elems <= s.a.len:
for i in 0..<s.elems:
yield s.a[i]
else:
var r = s.head var r = s.head
while r != nil: while r != nil:
var i = 0 var i = 0
@ -225,3 +280,9 @@ when isMainModule:
ys.sort(cmp[int]) ys.sort(cmp[int])
assert ys == @[1, 2, 7, 1056] assert ys == @[1, 2, 7, 1056]
var z: IntSet
for i in 0..1000:
incl z, i
for i in 0..1000:
assert i in z