* New ATermMap, seems more-or-less finished.

This commit is contained in:
Eelco Dolstra 2006-05-03 23:07:38 +00:00
parent 052cefe1bd
commit db0d865ec4
2 changed files with 154 additions and 50 deletions

View file

@ -48,32 +48,53 @@ public:
ATerm get(ATerm key) const; ATerm get(ATerm key) const;
void remove(ATerm key); void remove(ATerm key);
void remove(const string & key);
private: private:
void init(unsigned int expectedCount); void init(unsigned int expectedCount);
void free();
void resizeTable(unsigned int expectedCount); void resizeTable(unsigned int expectedCount);
unsigned int hash1(ATerm key) const; void copy(KeyValue * elements, unsigned int size);
unsigned int hash2(ATerm key) const;
inline unsigned int hash1(ATerm key) const;
inline unsigned int hash2(ATerm key) const;
}; };
ATermMap::ATermMap(unsigned int expectedCount) ATermMap::ATermMap(unsigned int expectedCount)
{ {
init(expectedCount); init(expectedCount * 10 / 9); /* slight adjustment */
}
ATermMap::ATermMap(const ATermMap & map)
{
init(map.maxCount);
copy(map.hashTable, map.size);
}
ATermMap & ATermMap::operator = (const ATermMap & map)
{
if (this == &map) return *this;
free();
init(map.maxCount);
copy(map.hashTable, map.size);
return *this;
} }
ATermMap::~ATermMap() ATermMap::~ATermMap()
{ {
if (hashTable) free(hashTable); free();
} }
void ATermMap::init(unsigned int expectedCount) void ATermMap::init(unsigned int expectedCount)
{ {
assert(sizeof(ATerm) * 2 == sizeof(KeyValue));
size = 0; size = 0;
count = 0; count = 0;
maxCount = 0; maxCount = 0;
@ -82,18 +103,67 @@ void ATermMap::init(unsigned int expectedCount)
} }
void ATermMap::free()
{
if (hashTable) {
ATunprotectArray((ATerm *) hashTable);
::free(hashTable);
hashTable = 0;
}
}
static unsigned int roundToPowerOf2(unsigned int x)
{
x--;
x |= x >> 1; x |= x >> 2; x |= x >> 4; x |= x >> 8; x |= x >> 16;
x++;
return x;
}
static const unsigned int maxLoadFactor = /* 1 / */ 3;
static unsigned int nrResizes = 0;
void ATermMap::resizeTable(unsigned int expectedCount) void ATermMap::resizeTable(unsigned int expectedCount)
{ {
assert(size == 0); if (expectedCount == 0) expectedCount = 1;
// cout << maxCount << " -> " << expectedCount << endl;
// cout << maxCount << " " << size << endl;
// cout << (double) size / maxCount << endl;
this->maxCount = expectedCount; unsigned int oldSize = size;
KeyValue * oldHashTable = hashTable;
unsigned int newSize = 128; maxCount = expectedCount;
size = roundToPowerOf2(maxCount * maxLoadFactor);
hashTable = (KeyValue *) calloc(sizeof(KeyValue), size);
ATprotectArray((ATerm *) hashTable, size * 2);
hashTable = (KeyValue *) calloc(sizeof(KeyValue), newSize); // cout << size << endl;
size = newSize; /* Re-hash the elements in the old table. */
if (oldSize != 0) {
count = 0;
copy(oldHashTable, oldSize);
ATunprotectArray((ATerm *) oldHashTable);
::free(oldHashTable);
nrResizes++;
} }
}
void ATermMap::copy(KeyValue * elements, unsigned int size)
{
for (unsigned int i = 0; i < size; ++i)
if (elements[i].value) /* i.e., non-empty, non-deleted element */
set(elements[i].key, elements[i].value);
}
static const unsigned int shift = 16;
static const unsigned int knuth = (unsigned int) (0.6180339887 * (1 << shift));
unsigned int ATermMap::hash1(ATerm key) const unsigned int ATermMap::hash1(ATerm key) const
@ -102,68 +172,58 @@ unsigned int ATermMap::hash1(ATerm key) const
pointer since they're always 0. */ pointer since they're always 0. */
unsigned int key2 = ((unsigned int) key) >> 2; unsigned int key2 = ((unsigned int) key) >> 2;
#if 0 /* Approximately equal to:
double d1 = key2 * 0.6180339887; double d = key2 * 0.6180339887;
unsigned int h1 = (int) (size * (d1 - floor(d1))); unsigned int h = (int) (size * (d - floor(d)));
#endif */
#if 0 unsigned int h = (size * ((key2 * knuth) & ((1 << shift) - 1))) >> shift;
unsigned int h1 = size * (key2 * 61803 % 100000);
#endif
unsigned int h1 = (size * ((key2 * 40503) & 0xffff)) >> 16; return h;
// cout << key2 << " " << h1 << endl;
// unsigned int h1 = (key2 * 134217689) & (size - 1);
return h1 % size;
} }
unsigned int ATermMap::hash2(ATerm key) const unsigned int ATermMap::hash2(ATerm key) const
{ {
unsigned int key2 = ((unsigned int) key) >> 2; unsigned int key2 = ((unsigned int) key) >> 2;
/* Note: the result must be relatively prime to `size' (which is a
#if 0 power of 2), so we make sure that the result is always odd. */
double d2 = key2 * 0.6180339887; unsigned int h = ((key2 * 134217689) & (size - 1)) | 1;
unsigned int h2 = 1 | (int) (size * (d2 - floor(d2))); return h;
#endif
unsigned int h3 = ((key2 * 134217689) & (size - 1)) | 1;
return h3;
} }
unsigned int nrItemsSet = 0; static unsigned int nrItemsSet = 0;
unsigned int nrSetProbes = 0; static unsigned int nrSetProbes = 0;
unsigned int nrMaxProbes = 0;
void ATermMap::set(ATerm key, ATerm value) void ATermMap::set(ATerm key, ATerm value)
{ {
unsigned int probes = 0; if (count == maxCount) resizeTable(size * 2 / maxLoadFactor);
nrItemsSet++; nrItemsSet++;
for (unsigned int i = 0, h = hash1(key); i < size; for (unsigned int i = 0, h = hash1(key); i < size;
++i, h = (h + hash2(key)) & (size - 1)) ++i, h = (h + hash2(key)) & (size - 1))
{ {
assert(h < size); // assert(h < size);
probes++;
nrSetProbes++; nrSetProbes++;
if (hashTable[h].key == 0) { /* Note: to see whether a slot is free, we check
if (probes > nrMaxProbes) nrMaxProbes = probes; hashTable[h].value, not hashTable[h].key, since we use
value == 0 to mark deleted slots. */
if (hashTable[h].value == 0 || hashTable[h].key == key) {
hashTable[h].key = key; hashTable[h].key = key;
hashTable[h].value = value; hashTable[h].value = value;
count++; count++;
return; return;
} }
} }
abort(); abort();
} }
unsigned int nrItemsGet = 0; static unsigned int nrItemsGet = 0;
unsigned int nrGetProbes = 0; static unsigned int nrGetProbes = 0;
ATerm ATermMap::get(ATerm key) const ATerm ATermMap::get(ATerm key) const
@ -180,6 +240,20 @@ ATerm ATermMap::get(ATerm key) const
} }
void ATermMap::remove(ATerm key)
{
for (unsigned int i = 0, h = hash1(key); i < size;
++i, h = (h + hash2(key)) & (size - 1))
{
if (hashTable[h].key == 0) return;
if (hashTable[h].key == key) {
hashTable[h].value = 0;
return;
}
}
}
int main(int argc, char * * argv) int main(int argc, char * * argv)
{ {
ATerm bottomOfStack; ATerm bottomOfStack;
@ -211,18 +285,37 @@ int main(int argc, char * * argv)
for (int test = 0; test < 100000; ++test) { for (int test = 0; test < 100000; ++test) {
ATermMap map(100); // cerr << test << endl;
for (int i = 0; i < 30; ++i) unsigned int n = 300;
map.set(someTerm(), someTerm()); ATermMap map(300);
for (int i = 0; i < 100; ++i) ATerm keys[n], values[n];
for (unsigned int i = 0; i < n; ++i) {
keys[i] = someTerm();
values[i] = someTerm();
map.set(keys[i], values[i]);
// cerr << "INSERT: " << keys[i] << " " << values[i] << endl;
}
values[n - 1] = 0;
map.remove(keys[n - 1]);
for (unsigned int i = 0; i < n; ++i) {
if (map.get(keys[i]) != values[i]) {
for (unsigned int j = i + 1; j < n; ++j)
if (keys[i] == keys[j]) goto x;
cerr << "MISMATCH: " << keys[i] << " " << values[i] << " " << map.get(keys[i]) << " " << i << endl;
abort();
x: ;
}
}
for (unsigned int i = 0; i < 100; ++i)
map.get(someTerm()); map.get(someTerm());
} }
cout << "RESIZES: " << nrResizes << endl;
cout << "SET: " cout << "SET: "
<< nrItemsSet << " " << nrItemsSet << " "
<< nrSetProbes << " " << nrSetProbes << " "
<< (double) nrSetProbes / nrItemsSet << " " << (double) nrSetProbes / nrItemsSet << endl;
<< nrMaxProbes << endl;
cout << "GET: " cout << "GET: "
<< nrItemsGet << " " << nrItemsGet << " "

View file

@ -17,3 +17,14 @@ GET: 10000000 13032311 1.30323
real 0m0.956s real 0m0.956s
user 0m0.928s user 0m0.928s
sys 0m0.024s sys 0m0.024s
[eelco@tyros:~/Dev/nix/table]$ time ./a.out
testing...
RESIZES: 0
SET: 30000000 35420570 1.18069
GET: 40000000 49507966 1.2377
real 0m3.796s
user 0m3.700s
sys 0m0.016s