Skip to content

Commit 65d7bf6

Browse files
AskAlexSharovAskAlexSharov
committed
save
1 parent de9398b commit 65d7bf6

File tree

2 files changed

+72
-81
lines changed

2 files changed

+72
-81
lines changed

‎db/recsplit/eliasfano16/elias_fano.go‎

Lines changed: 66 additions & 68 deletions
Original file line numberDiff line numberDiff line change
@@ -25,6 +25,7 @@ import (
2525
"unsafe"
2626

2727
"github.com/c2h5oh/datasize"
28+
2829
"github.com/erigontech/erigon/common/bitutil"
2930
)
3031

@@ -118,36 +119,35 @@ func (ef *EliasFano) deriveFields() int {
118119
// Build construct Elias Fano index for a given sequences
119120
func (ef *EliasFano) Build() {
120121
for i, c, lastSuperQ := uint64(0), uint64(0), uint64(0); i < uint64(ef.wordsUpperBits); i++ {
121-
for b := uint64(0); b < 64; b++ {
122-
if ef.upperBits[i]&(uint64(1)<<b) != 0 {
123-
if (c & superQMask) == 0 {
124-
// When c is multiple of 2^14 (4096)
125-
lastSuperQ = i*64 + b
126-
ef.jump[(c/superQ)*superQSize] = lastSuperQ
127-
}
128-
if (c & qMask) == 0 {
129-
// When c is multiple of 2^8 (256)
130-
var offset = i*64 + b - lastSuperQ // offset can be either 0, 256, 512, 768, ..., up to 4096-256
131-
// offset needs to be encoded as 16-bit integer, therefore the following check
132-
if offset >= (1 << 16) {
133-
fmt.Printf("ef.l=%x,ef.u=%x\n", ef.l, ef.u)
134-
fmt.Printf("offset=%x,lastSuperQ=%x,i=%x,b=%x,c=%x\n", offset, lastSuperQ, i, b, c)
135-
fmt.Printf("ef.minDelta=%x\n", ef.minDelta)
136-
//fmt.Printf("ef.upperBits=%x\n", ef.upperBits)
137-
//fmt.Printf("ef.lowerBits=%x\n", ef.lowerBits)
138-
//fmt.Printf("ef.wordsUpperBits=%b\n", ef.wordsUpperBits)
139-
panic("")
140-
}
141-
// c % superQ is the bit index inside the group of 4096 bits
142-
jumpSuperQ := (c / superQ) * superQSize
143-
jumpInsideSuperQ := (c % superQ) / q
144-
idx64 := jumpSuperQ + 1 + (jumpInsideSuperQ >> 2)
145-
shift := 16 * (jumpInsideSuperQ % 4)
146-
mask := uint64(0xffff) << shift
147-
ef.jump[idx64] = (ef.jump[idx64] &^ mask) | (offset << shift)
122+
for word := ef.upperBits[i]; word != 0; word &= word - 1 { // iterate over set bits only; word &= word-1 clears the lowest set bit
123+
b := uint64(bits.TrailingZeros64(word))
124+
if (c & superQMask) == 0 {
125+
// When c is multiple of 2^14 (4096)
126+
lastSuperQ = i*64 + b
127+
ef.jump[(c/superQ)*superQSize] = lastSuperQ
128+
}
129+
if (c & qMask) == 0 {
130+
// When c is multiple of 2^8 (256)
131+
var offset = i*64 + b - lastSuperQ // offset can be either 0, 256, 512, 768, ..., up to 4096-256
132+
// offset needs to be encoded as 16-bit integer, therefore the following check
133+
if offset >= (1 << 16) {
134+
fmt.Printf("ef.l=%x,ef.u=%x\n", ef.l, ef.u)
135+
fmt.Printf("offset=%x,lastSuperQ=%x,i=%x,b=%x,c=%x\n", offset, lastSuperQ, i, b, c)
136+
fmt.Printf("ef.minDelta=%x\n", ef.minDelta)
137+
//fmt.Printf("ef.upperBits=%x\n", ef.upperBits)
138+
//fmt.Printf("ef.lowerBits=%x\n", ef.lowerBits)
139+
//fmt.Printf("ef.wordsUpperBits=%b\n", ef.wordsUpperBits)
140+
panic("")
148141
}
149-
c++
142+
// c % superQ is the bit index inside the group of 4096 bits
143+
jumpSuperQ := (c / superQ) * superQSize
144+
jumpInsideSuperQ := (c % superQ) / q
145+
idx64 := jumpSuperQ + 1 + (jumpInsideSuperQ >> 2)
146+
shift := 16 * (jumpInsideSuperQ % 4)
147+
mask := uint64(0xffff) << shift
148+
ef.jump[idx64] = (ef.jump[idx64] &^ mask) | (offset << shift)
150149
}
150+
c++
151151
}
152152
}
153153
}
@@ -355,54 +355,52 @@ func (ef *DoubleEliasFano) Build(cumKeys []uint64, position []uint64) {
355355
// c/superQ is the index of the current 4096 block of bits
356356
// superQSize is how many words is required to encode one block of 4096 bits. It is 17 words which is 1088 bits
357357
for i, c, lastSuperQ := uint64(0), uint64(0), uint64(0); i < uint64(wordsCumKeys); i++ {
358-
for b := uint64(0); b < 64; b++ {
359-
if ef.upperBitsCumKeys[i]&(uint64(1)<<b) != 0 {
360-
if (c & superQMask) == 0 {
361-
// When c is multiple of 2^14 (4096)
362-
lastSuperQ = i*64 + b
363-
ef.jump[(c/superQ)*(superQSize*2)] = lastSuperQ
364-
}
365-
if (c & qMask) == 0 {
366-
// When c is multiple of 2^8 (256)
367-
var offset = i*64 + b - lastSuperQ // offset can be either 0, 256, 512, 768, ..., up to 4096-256
368-
// offset needs to be encoded as 16-bit integer, therefore the following check
369-
if offset >= (1 << 16) {
370-
panic("")
371-
}
372-
// c % superQ is the bit index inside the group of 4096 bits
373-
jumpSuperQ := (c / superQ) * (superQSize * 2)
374-
jumpInsideSuperQ := 2 * (c % superQ) / q
375-
idx64 := jumpSuperQ + 2 + (jumpInsideSuperQ >> 2)
376-
shift := 16 * (jumpInsideSuperQ % 4)
377-
mask := uint64(0xffff) << shift
378-
ef.jump[idx64] = (ef.jump[idx64] &^ mask) | (offset << shift)
358+
for word := ef.upperBitsCumKeys[i]; word != 0; word &= word - 1 { // iterate over set bits only; word &= word-1 clears the lowest set bit
359+
b := uint64(bits.TrailingZeros64(word))
360+
if (c & superQMask) == 0 {
361+
// When c is multiple of 2^14 (4096)
362+
lastSuperQ = i*64 + b
363+
ef.jump[(c/superQ)*(superQSize*2)] = lastSuperQ
364+
}
365+
if (c & qMask) == 0 {
366+
// When c is multiple of 2^8 (256)
367+
var offset = i*64 + b - lastSuperQ // offset can be either 0, 256, 512, 768, ..., up to 4096-256
368+
// offset needs to be encoded as 16-bit integer, therefore the following check
369+
if offset >= (1 << 16) {
370+
panic("")
379371
}
380-
c++
372+
// c % superQ is the bit index inside the group of 4096 bits
373+
jumpSuperQ := (c / superQ) * (superQSize * 2)
374+
jumpInsideSuperQ := 2 * (c % superQ) / q
375+
idx64 := jumpSuperQ + 2 + (jumpInsideSuperQ >> 2)
376+
shift := 16 * (jumpInsideSuperQ % 4)
377+
mask := uint64(0xffff) << shift
378+
ef.jump[idx64] = (ef.jump[idx64] &^ mask) | (offset << shift)
381379
}
380+
c++
382381
}
383382
}
384383

385384
for i, c, lastSuperQ := uint64(0), uint64(0), uint64(0); i < uint64(wordsPosition); i++ {
386-
for b := uint64(0); b < 64; b++ {
387-
if ef.upperBitsPosition[i]&(uint64(1)<<b) != 0 {
388-
if (c & superQMask) == 0 {
389-
lastSuperQ = i*64 + b
390-
ef.jump[(c/superQ)*(superQSize*2)+1] = lastSuperQ
391-
}
392-
if (c & qMask) == 0 {
393-
var offset = i*64 + b - lastSuperQ
394-
if offset >= (1 << 16) {
395-
panic("")
396-
}
397-
jumpSuperQ := (c / superQ) * (superQSize * 2)
398-
jumpInsideSuperQ := 2*((c%superQ)/q) + 1
399-
idx64 := jumpSuperQ + 2 + (jumpInsideSuperQ >> 2)
400-
shift := 16 * (jumpInsideSuperQ % 4)
401-
mask := uint64(0xffff) << shift
402-
ef.jump[idx64] = (ef.jump[idx64] &^ mask) | (offset << shift)
385+
for word := ef.upperBitsPosition[i]; word != 0; word &= word - 1 { // iterate over set bits only; word &= word-1 clears the lowest set bit
386+
b := uint64(bits.TrailingZeros64(word))
387+
if (c & superQMask) == 0 {
388+
lastSuperQ = i*64 + b
389+
ef.jump[(c/superQ)*(superQSize*2)+1] = lastSuperQ
390+
}
391+
if (c & qMask) == 0 {
392+
var offset = i*64 + b - lastSuperQ
393+
if offset >= (1 << 16) {
394+
panic("")
403395
}
404-
c++
396+
jumpSuperQ := (c / superQ) * (superQSize * 2)
397+
jumpInsideSuperQ := 2*((c%superQ)/q) + 1
398+
idx64 := jumpSuperQ + 2 + (jumpInsideSuperQ >> 2)
399+
shift := 16 * (jumpInsideSuperQ % 4)
400+
mask := uint64(0xffff) << shift
401+
ef.jump[idx64] = (ef.jump[idx64] &^ mask) | (offset << shift)
405402
}
403+
c++
406404
}
407405
}
408406
//fmt.Printf("jump: %x\n", ef.jump)

‎db/recsplit/eliasfano32/elias_fano.go‎

Lines changed: 6 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -123,10 +123,8 @@ func (ef *EliasFano) deriveFields() int {
123123
// Build construct Elias Fano index for a given sequences
124124
func (ef *EliasFano) Build() {
125125
for i, c, lastSuperQ := uint64(0), uint64(0), uint64(0); i < uint64(ef.wordsUpperBits); i++ {
126-
for b := uint64(0); b < 64; b++ {
127-
if ef.upperBits[i]&(uint64(1)<<b) == 0 {
128-
continue
129-
}
126+
for word := ef.upperBits[i]; word != 0; word &= word - 1 { // iterate over set bits only; word &= word-1 clears the lowest set bit
127+
b := uint64(bits.TrailingZeros64(word))
130128
if (c & superQMask) == 0 {
131129
// When c is multiple of 2^14 (4096)
132130
lastSuperQ = i*64 + b
@@ -739,10 +737,8 @@ func (ef *DoubleEliasFano) Build(cumKeys []uint64, position []uint64) {
739737
// c/superQ is the index of the current 4096 block of bits
740738
// superQSize is how many words is required to encode one block of 4096 bits. It is 17 words which is 1088 bits
741739
for i, c, lastSuperQ := uint64(0), uint64(0), uint64(0); i < uint64(wordsCumKeys); i++ {
742-
for b := uint64(0); b < 64; b++ {
743-
if ef.upperBitsCumKeys[i]&(uint64(1)<<b) == 0 {
744-
continue
745-
}
740+
for word := ef.upperBitsCumKeys[i]; word != 0; word &= word - 1 { // iterate over set bits only; word &= word-1 clears the lowest set bit
741+
b := uint64(bits.TrailingZeros64(word))
746742
if (c & superQMask) == 0 {
747743
// When c is multiple of 2^14 (4096)
748744
lastSuperQ = i*64 + b
@@ -768,11 +764,8 @@ func (ef *DoubleEliasFano) Build(cumKeys []uint64, position []uint64) {
768764
}
769765

770766
for i, c, lastSuperQ := uint64(0), uint64(0), uint64(0); i < uint64(wordsPosition); i++ {
771-
for b := uint64(0); b < 64; b++ {
772-
if ef.upperBitsPosition[i]&(uint64(1)<<b) == 0 {
773-
continue
774-
}
775-
767+
for word := ef.upperBitsPosition[i]; word != 0; word &= word - 1 { // iterate over set bits only; word &= word-1 clears the lowest set bit
768+
b := uint64(bits.TrailingZeros64(word))
776769
if (c & superQMask) == 0 {
777770
lastSuperQ = i*64 + b
778771
ef.jump[(c/superQ)*(superQSize*2)+1] = lastSuperQ

0 commit comments

Comments
 (0)