Playground/1brc
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

parse value before going to map (#548)

Browse Source 
parse value before going to map
This commit is contained in:
Artsiom Korzun
2024-01-23 19:38:32 +01:00
committed by GitHub
parent 46d2058bd4
commit d858959097
2 changed files with 50 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
prepare_artsiomkorzun.sh
View File

@@ -19,6 +19,6 @@ source "$HOME/.sdkman/bin/sdkman-init.sh"
sdk use java 21.0.2-graal 1>&2
if [ ! -f target/CalculateAverage_artsiomkorzun_image ]; then
NATIVE_IMAGE_OPTS="--gc=epsilon -O3 -march=native -R:MaxHeapSize=64m --enable-preview --initialize-at-build-time=dev.morling.onebrc.CalculateAverage_artsiomkorzun"
NATIVE_IMAGE_OPTS="--gc=epsilon -O3 -march=native -R:MaxHeapSize=64m -H:-GenLoopSafepoints --enable-preview --initialize-at-build-time=dev.morling.onebrc.CalculateAverage_artsiomkorzun"
native-image $NATIVE_IMAGE_OPTS -cp target/average-1.0.0-SNAPSHOT.jar -o target/CalculateAverage_artsiomkorzun_image dev.morling.onebrc.CalculateAverage_artsiomkorzun
fi

76
src/main/java/dev/morling/onebrc/CalculateAverage_artsiomkorzun.java
View File

@@ -383,6 +383,13 @@ public class CalculateAverage_artsiomkorzun {
}
}
private static long next(long position) {
while (UNSAFE.getByte(position++) != '\n') {
// continue
}
return position;
}
private static void aggregate(Aggregates aggregates, long position, long limit) {
// this parsing can produce seg fault at page boundaries
// e.g. file size is 4096 and the last entry is X=0.0, which is less than 8 bytes
@@ -392,18 +399,27 @@ public class CalculateAverage_artsiomkorzun {
while (position <= limit) { // branchy version, credit: thomaswue
int length;
int hash;
int value;
long word = word(position);
long separator = separator(word);
long end = position;
if (separator != 0) {
length = length(separator);
word = mask(word, separator);
hash = mix(word);
end += length;
long num = word(end);
int dot = dot(num);
value = value(num, dot);
end += (dot >> 3) + 3;
long ptr = aggregates.find(word, hash);
if (ptr != 0) {
position = update(ptr, position + length);
Aggregates.update(ptr, value);
position = end;
continue;
}
}
@@ -416,10 +432,17 @@ public class CalculateAverage_artsiomkorzun {
length = length(separator) + 8;
word = mask(word, separator);
hash = mix(word ^ word0);
end += length;
long num = word(end);
int dot = dot(num);
value = value(num, dot);
end += (dot >> 3) + 3;
long ptr = aggregates.find(word0, word, hash);
if (ptr != 0) {
position = update(ptr, position + length);
Aggregates.update(ptr, value);
position = end;
continue;
}
}
@@ -440,31 +463,23 @@ public class CalculateAverage_artsiomkorzun {
length += length(separator);
word = mask(word, separator);
hash = mix(h ^ word);
end += length;
long num = word(end);
int dot = dot(num);
value = value(num, dot);
end += (dot >> 3) + 3;
break;
}
}
}
long ptr = aggregates.put(position, word, length, hash);
position = update(ptr, position + length);
Aggregates.update(ptr, value);
position = end;
}
}
private static long update(long ptr, long position) {
// idea: merykitty
long word = word(position);
long inverted = ~word;
int dot = Long.numberOfTrailingZeros(inverted & DOT_BITS);
long signed = (inverted << 59) >> 63;
long mask = ~(signed & 0xFF);
long digits = ((word & mask) << (28 - dot)) & 0x0F000F0F00L;
long abs = ((digits * MAGIC_MULTIPLIER) >>> 32) & 0x3FF;
int value = (int) ((abs ^ signed) - signed);
Aggregates.update(ptr, value);
return position + (dot >> 3) + 3;
}
private static long separator(long word) {
long match = word ^ COMMA_PATTERN;
return (match - 0x0101010101010101L) & (~match & 0x8080808080808080L);
@@ -479,17 +494,24 @@ public class CalculateAverage_artsiomkorzun {
return (Long.numberOfTrailingZeros(separator) >>> 3) + 1;
}
private static long next(long position) {
while (UNSAFE.getByte(position++) != '\n') {
// continue
}
return position;
}
private static int mix(long x) {
long h = x * -7046029254386353131L;
h ^= h >>> 32;
return (int) (h ^ h >>> 16);
h ^= h >>> 35;
return (int) h;
// h ^= h >>> 32;
// return (int) (h ^ h >>> 16);
}
private static int dot(long num) {
return Long.numberOfTrailingZeros(~num & DOT_BITS);
}
private static int value(long w, int dot) {
long signed = (~w << 59) >> 63;
long mask = ~(signed & 0xFF);
long digits = ((w & mask) << (28 - dot)) & 0x0F000F0F00L;
long abs = ((digits * MAGIC_MULTIPLIER) >>> 32) & 0x3FF;
return (int) ((abs ^ signed) - signed);
}
}
}