Reorganize the code: use SimpleMixSplit instead of TwoIndependentMultiplyShift #18

Author: thomasmueller

src/gqf/gqf_cpp.h

@@ -25,7 +25,7 @@ enum Status {
 };
 
 template <typename ItemType, size_t bits_per_item,
-    typename HashFamily = TwoIndependentMultiplyShift>
+    typename HashFamily = SimpleMixSplit>
 class GQFilter {
 
   QF qf;
@@ -49,17 +49,11 @@ class GQFilter {
     uint64_t nhashbits = qbits + 8;
     mask = (1ULL << nhashbits) - 1;
 
-// std::cout << "(CQF: nslots " << nslots << " nhashbits " << nhashbits << " n " << n << " bitsPerItem " << bitsPerItem << ")\n";
-
-// if (!qf_malloc(&qf, nslots, nhashbits, 0, QF_HASH_INVERTIBLE, 0)) {
-//    if (!qf_malloc(&qf, nslots, nhashbits, 0, QF_HASH_DEFAULT, 0)) {
     if (!qf_malloc(&qf, nslots, nhashbits, 0, QF_HASH_NONE, 0)) {
         std::cout << "Can't allocate CQF.\n";
         abort();
     }
 
-    // qf_set_auto_resize(&qf, true);
-
     bytesUsed = qf.metadata->total_size_in_bytes;
     bitsPerItem = (double) bytesUsed / n;
 

src/gqf/gqf_hashutil.c

@@ -46,12 +46,12 @@ uint64_t MurmurHash64A ( const void * key, int len, unsigned int seed )
 
 	switch(len & 7)
 	{
-		case 7: h ^= (uint64_t)data2[6] << 48;
-		case 6: h ^= (uint64_t)data2[5] << 40;
-		case 5: h ^= (uint64_t)data2[4] << 32;
-		case 4: h ^= (uint64_t)data2[3] << 24;
-		case 3: h ^= (uint64_t)data2[2] << 16;
-		case 2: h ^= (uint64_t)data2[1] << 8;
+		case 7: h ^= (uint64_t)data2[6] << 48; do {} while (0);  /* fallthrough */
+		case 6: h ^= (uint64_t)data2[5] << 40; do {} while (0);  /* fallthrough */
+		case 5: h ^= (uint64_t)data2[4] << 32; do {} while (0);  /* fallthrough */
+		case 4: h ^= (uint64_t)data2[3] << 24; do {} while (0);  /* fallthrough */
+		case 3: h ^= (uint64_t)data2[2] << 16; do {} while (0);  /* fallthrough */
+		case 2: h ^= (uint64_t)data2[1] << 8; do {} while (0); /* fallthrough */
 		case 1: h ^= (uint64_t)data2[0];
 						h *= m;
 	};
@@ -99,10 +99,10 @@ uint64_t MurmurHash64B ( const void * key, int len, unsigned int seed )
 
 	switch(len)
 	{
-		case 3: h2 ^= ((unsigned char*)data)[2] << 16;
-		case 2: h2 ^= ((unsigned char*)data)[1] << 8;
+		case 3: h2 ^= ((unsigned char*)data)[2] << 16;  do {} while (0);  /* fallthrough */
+		case 2: h2 ^= ((unsigned char*)data)[1] << 8;  do {} while (0);  /* fallthrough */
 		case 1: h2 ^= ((unsigned char*)data)[0];
-						h2 *= m;
+						h2 *= m; do {} while (0);  /* fallthrough */
 	};
 
 	h1 ^= h2 >> 18; h1 *= m;

src/morton/block.h

@@ -165,7 +165,6 @@ struct Block{ // Assuming block size is a multiple of atom_t's size in bytes
       // How many bits spilled over into atom1
       uint16_t spillover = (global_index + field_width_bits) % atom_size_bits;
       spillover = (atom1 - atom0) ? spillover : 0; // Compare and swap ins?
-      //std::cout << "Spillover is " << spillover << std::endl;
 
       // May be different than the default read mask if field_width_bits 
       // differs from what is stored in the struct
@@ -402,7 +401,6 @@ struct Block{ // Assuming block size is a multiple of atom_t's size in bytes
       memcpy(source_address, &item, field_width_bytes);
     }
 
-    __attribute__((optimize("unroll-loops")))
     INLINE void add_cross_left_displace(uint64_t raw_offset_bits, 
       uint64_t field_width_bits, uint64_t index, atom_t item){
       uint64_t global_index = index * field_width_bits + raw_offset_bits;
@@ -420,8 +418,6 @@ struct Block{ // Assuming block size is a multiple of atom_t's size in bytes
       for(uint64_t atom = 1; atom < atoms_per_block; atom++){
         shifted_block._block_storage[atom] |= _block_storage[atom - 1] >> (atom_size_bits - field_width_bits);
       } 
-      //std::cout << "Block before shift: " << block_storage_as_bit_string(entry_size_bits)  << std::endl;
-      //std::cout << "Block after shift: " << shifted_block.block_storage_as_bit_string(entry_size_bits) << std::endl;
 
       // Clear the entire atom
       constexpr atom_t full_clear_mask = static_cast<atom_t>(0);
@@ -482,8 +478,6 @@ struct Block{ // Assuming block size is a multiple of atom_t's size in bytes
       for(uint64_t atom = 1; atom < atoms_per_block; atom++){
         shifted_block._block_storage[atom - 1] |= _block_storage[atom] << (atom_size_bits - field_width_bits);
       } 
-      //std::cout << "Block before shift: " << block_storage_as_bit_string(entry_size_bits)  << std::endl;
-      //std::cout << "Block after shift: " << shifted_block.block_storage_as_bit_string(entry_size_bits) << std::endl;
 
       // What follows is a merge operation that is exactly the same as 
       // what appears in add_cross_left_displace