hashmap: new load_factor and optimizations
ka-weihe
Alexander Medvednikov
parent
bf6b206fca
commit
31899eac2a
|
|
@ -6,29 +6,18 @@ module hashmap
|
|||
const (
|
||||
initial_size = 2 << 4
|
||||
initial_cap = initial_size - 1
|
||||
load_factor = 0.5
|
||||
probe_offset = u16(256)
|
||||
load_factor = 0.8
|
||||
)
|
||||
|
||||
// hash-function should not be in this file
|
||||
const (
|
||||
fnv64_prime = 1099511628211
|
||||
fnv64_offset_basis = 14695981039346656037
|
||||
fnv32_offset_basis = u32(2166136261)
|
||||
fnv32_prime = u32(16777619)
|
||||
)
|
||||
|
||||
pub struct Hashmap {
|
||||
mut:
|
||||
info &u16
|
||||
key_values &KeyValue
|
||||
cap int
|
||||
pub mut:
|
||||
size int
|
||||
}
|
||||
|
||||
struct KeyValue {
|
||||
key string
|
||||
mut:
|
||||
value int
|
||||
}
|
||||
|
||||
[inline]
|
||||
fn fnv1a64(data string) u64 {
|
||||
mut hash := fnv64_offset_basis
|
||||
|
|
@ -38,11 +27,28 @@ fn fnv1a64(data string) u64 {
|
|||
return hash
|
||||
}
|
||||
|
||||
pub struct Hashmap {
|
||||
mut:
|
||||
info &u16
|
||||
key_values &KeyValue
|
||||
cap int
|
||||
pub mut:
|
||||
load_factor f32
|
||||
size int
|
||||
}
|
||||
|
||||
struct KeyValue {
|
||||
key string
|
||||
mut:
|
||||
value int
|
||||
}
|
||||
|
||||
pub fn new_hashmap() Hashmap {
|
||||
return Hashmap{
|
||||
info: &u16(calloc(sizeof(u16) * initial_size))
|
||||
key_values: &KeyValue(calloc(sizeof(KeyValue) * initial_size))
|
||||
cap: initial_cap
|
||||
load_factor: 0.8
|
||||
size: 0
|
||||
}
|
||||
}
|
||||
|
|
@ -51,7 +57,7 @@ pub fn (h mut Hashmap) set(key string, value int) {
|
|||
// The load factor is 0.5.
|
||||
// It will be adjustable in the future and with
|
||||
// a higher default settings to lower memory usage.
|
||||
if (h.size<<1) == (h.cap - 1) {
|
||||
if (f32(h.size) / f32(h.cap)) > h.load_factor {
|
||||
h.rehash()
|
||||
}
|
||||
// Hash-function will be swapped for wyhash
|
||||
|
|
@ -75,18 +81,17 @@ pub fn (h mut Hashmap) set(key string, value int) {
|
|||
// Match is not possible anymore.
|
||||
// Probe until an empty index is found.
|
||||
// Swap when probe count is higher/richer (Robin Hood).
|
||||
mut current_key := key
|
||||
mut current_value := value
|
||||
mut current_kv := KeyValue{key, value}
|
||||
for h.info[index] != 0 {
|
||||
if info > h.info[index] {
|
||||
tmp_kv := h.key_values[index]
|
||||
// Swap info word
|
||||
tmp_info := h.info[index]
|
||||
h.key_values[index] = KeyValue{
|
||||
current_key,current_value}
|
||||
h.info[index] = info
|
||||
current_key = tmp_kv.key
|
||||
current_value = tmp_kv.value
|
||||
info = tmp_info
|
||||
// Swap KeyValue
|
||||
tmp_kv := h.key_values[index]
|
||||
h.key_values[index] = current_kv
|
||||
current_kv = tmp_kv
|
||||
}
|
||||
index = (index + 1) & h.cap
|
||||
info += probe_offset
|
||||
|
|
@ -94,12 +99,11 @@ pub fn (h mut Hashmap) set(key string, value int) {
|
|||
// Should almost never happen
|
||||
if (info & 0xFF00) == 0xFF00 {
|
||||
h.rehash()
|
||||
h.set(current_key, current_value)
|
||||
h.set(current_kv.key, current_kv.value)
|
||||
return
|
||||
}
|
||||
h.info[index] = info
|
||||
h.key_values[index] = KeyValue{
|
||||
current_key,current_value}
|
||||
h.key_values[index] = current_kv
|
||||
h.size++
|
||||
}
|
||||
|
||||
|
|
@ -110,9 +114,8 @@ fn (h mut Hashmap) rehash() {
|
|||
mut new_info := &u16(calloc(sizeof(u16) * (h.cap + 1)))
|
||||
for i in 0 .. (old_cap + 1) {
|
||||
if h.info[i] != 0 {
|
||||
key := h.key_values[i].key
|
||||
value := h.key_values[i].value
|
||||
hash := fnv1a64(key)
|
||||
mut kv := h.key_values[i]
|
||||
hash := fnv1a64(kv.key)
|
||||
mut info := u16((hash >> 56) | probe_offset)
|
||||
mut index := hash & h.cap
|
||||
// While probe count is less
|
||||
|
|
@ -120,30 +123,18 @@ fn (h mut Hashmap) rehash() {
|
|||
index = (index + 1) & h.cap
|
||||
info += probe_offset
|
||||
}
|
||||
// While we might have a match
|
||||
for info == new_info[index] {
|
||||
if key == new_key_values[index].key {
|
||||
new_key_values[index].value = value
|
||||
return
|
||||
}
|
||||
index = (index + 1) & h.cap
|
||||
info += probe_offset
|
||||
}
|
||||
// Match is not possible anymore.
|
||||
// Probe until an empty index is found.
|
||||
// Swap when probe count is higher/richer (Robin Hood).
|
||||
mut current_key := key
|
||||
mut current_value := value
|
||||
for new_info[index] != 0 {
|
||||
if info > new_info[index] {
|
||||
tmp_kv := new_key_values[index]
|
||||
// Swap info word
|
||||
tmp_info := new_info[index]
|
||||
new_key_values[index] = KeyValue{
|
||||
current_key,current_value}
|
||||
new_info[index] = info
|
||||
current_key = tmp_kv.key
|
||||
current_value = tmp_kv.value
|
||||
info = tmp_info
|
||||
// Swap KeyValue
|
||||
tmp_kv := new_key_values[index]
|
||||
new_key_values[index] = kv
|
||||
kv = tmp_kv
|
||||
}
|
||||
index = (index + 1) & h.cap
|
||||
info += probe_offset
|
||||
|
|
@ -151,12 +142,11 @@ fn (h mut Hashmap) rehash() {
|
|||
// Should almost never happen
|
||||
if (info & 0xFF00) == 0xFF00 {
|
||||
h.rehash()
|
||||
h.set(current_key, current_value)
|
||||
h.set(kv.key, kv.value)
|
||||
return
|
||||
}
|
||||
new_info[index] = info
|
||||
new_key_values[index] = KeyValue{
|
||||
current_key,current_value}
|
||||
new_key_values[index] = kv
|
||||
}
|
||||
}
|
||||
h.key_values = new_key_values
|
||||
|
|
@ -230,8 +220,7 @@ pub fn (h Hashmap) exists(key string) bool {
|
|||
}
|
||||
|
||||
pub fn (h Hashmap) keys() []string {
|
||||
size := h.size
|
||||
mut keys := [''].repeat(size)
|
||||
mut keys := [''].repeat(h.size)
|
||||
mut j := 0
|
||||
for i in 0 .. (h.cap + 1) {
|
||||
if h.info[i] != 0 {
|
||||
|
|
|
|||
Loading…
Reference in New Issue