Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Speed up writeVInt #62345

Merged
merged 11 commits into from
Sep 15, 2020
Merged
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion benchmarks/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -78,7 +78,6 @@ cd fcml*
make
cd example/hsdis
make
cp .libs/libhsdis.so.0.0.0
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This was just wrong.

sudo cp .libs/libhsdis.so.0.0.0 /usr/lib/jvm/java-14-adoptopenjdk/lib/hsdis-amd64.so
```

Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,77 @@
package org.elasticsearch.benchmark.search.aggregations.bucket.terms;

import org.apache.lucene.util.BytesRef;
import org.elasticsearch.Version;
import org.elasticsearch.common.bytes.BytesReference;
import org.elasticsearch.common.io.stream.BytesStreamOutput;
import org.elasticsearch.common.io.stream.DelayableWriteable;
import org.elasticsearch.common.io.stream.NamedWriteableRegistry;
import org.elasticsearch.search.DocValueFormat;
import org.elasticsearch.search.aggregations.BucketOrder;
import org.elasticsearch.search.aggregations.InternalAggregation;
import org.elasticsearch.search.aggregations.InternalAggregations;
import org.elasticsearch.search.aggregations.bucket.terms.StringTerms;
import org.openjdk.jmh.annotations.Benchmark;
import org.openjdk.jmh.annotations.BenchmarkMode;
import org.openjdk.jmh.annotations.Fork;
import org.openjdk.jmh.annotations.Measurement;
import org.openjdk.jmh.annotations.Mode;
import org.openjdk.jmh.annotations.OutputTimeUnit;
import org.openjdk.jmh.annotations.Param;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.Setup;
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.annotations.Warmup;

import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.concurrent.TimeUnit;

@Fork(2)
@Warmup(iterations = 10)
@Measurement(iterations = 5)
@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MILLISECONDS)
@State(Scope.Benchmark)
public class StringTermsSerializationBenchmark {
private static final NamedWriteableRegistry REGISTRY = new NamedWriteableRegistry(
List.of(new NamedWriteableRegistry.Entry(InternalAggregation.class, StringTerms.NAME, StringTerms::new))
);
@Param(value = { "1000" })
private int buckets;

private DelayableWriteable<InternalAggregations> results;

@Setup
public void initResults() {
results = DelayableWriteable.referencing(InternalAggregations.from(List.of(newTerms(true))));
}

private StringTerms newTerms(boolean withNested) {
List<StringTerms.Bucket> resultBuckets = new ArrayList<>(buckets);
for (int i = 0; i < buckets; i++) {
InternalAggregations inner = withNested ? InternalAggregations.from(List.of(newTerms(false))) : InternalAggregations.EMPTY;
resultBuckets.add(new StringTerms.Bucket(new BytesRef("test" + i), i, inner, false, 0, DocValueFormat.RAW));
}
return new StringTerms(
"test",
BucketOrder.key(true),
BucketOrder.key(true),
buckets,
1,
null,
DocValueFormat.RAW,
buckets,
false,
100000,
resultBuckets,
0
);
}

@Benchmark
public DelayableWriteable<InternalAggregations> serialize() {
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm unsure if we actually want this benchmark, especially compared to the one that @jimczi showed me. But it is fairly targeted which can be useful.

return results.asSerialized(InternalAggregations::readFrom, REGISTRY);
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -218,14 +218,61 @@ public void writeInt(int i) throws IOException {
* using {@link #writeInt}
*/
public void writeVInt(int i) throws IOException {
final byte[] buffer = scratch.get();
int index = 0;
while ((i & ~0x7F) != 0) {
buffer[index++] = ((byte) ((i & 0x7f) | 0x80));
i >>>= 7;
/*
* Pick the number of bytes that we need based on the value and then
* encode the int, unrolling the loops by hand. This allows writing
* small numbers to use `writeByte` which is simple and fast. The
* unrolling saves a few comparisons and bitwise operations. All
* together this saves quite a bit of time compared to a naive
* implementation.
*/
if (i < 0x7f) {
if (i >= 0) {
writeByte((byte) i);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I love this until here :) The fact that we can special case number of leading zeros > 24 is pretty significant and I can see the ~30% performance gain as well.

Hard coding all possible offsets below and doing all the buffer getting and writeBytes inline with those hard coded offsets I don't think is a good idea. This blows up the method size significantly for a tiny saving in CPU when it comes to evaluating the loop.

I benchmarked both this version and:

    public void writeVInt(int i) throws IOException {
        if (Integer.numberOfLeadingZeros(i) > 24) {
            writeByte((byte) i);
        } else {
            final byte[] buffer = scratch.get();
            int index = 0;
            do {
                buffer[index++] = ((byte) ((i & 0x7f) | 0x80));
                i >>>= 7;
            } while ((i & ~0x7F) != 0);
            buffer[index++] = ((byte) i);
            writeBytes(buffer, 0, index);
        }
    }

and I can't see statistically significant difference so that's not worth the complication IMO.

I would in fact expect the above version with the loop to be faster than what is in this PR in the real world because the smaller method size has a better better chance of getting inlined in some places (73 vs 507 bytes on JDK14/Linux for me).

I suppose you could work around the code bloat by doing this:

        final int leadingZeros = Integer.numberOfLeadingZeros(i);
        if (Integer.numberOfLeadingZeros(i) > 24) {
            writeByte((byte) i);
        } else {
            final byte[] buffer = scratch.get();
            final int length;
            switch (leadingZeros) {
                case 24:
                case 23:
                case 22:
                case 21:
                case 20:
                case 19:
                case 18:
                    buffer[0] = (byte) (i & 0x7f | 0x80);
                    buffer[1] = (byte) (i >>> 7);
                    assert buffer[1] <= 0x7f;
                    length = 2;
                    break;
                case 17:
                case 16:
                case 15:
                case 14:
                case 13:
                case 12:
                case 11:
                    buffer[0] = (byte) (i & 0x7f | 0x80);
                    buffer[1] = (byte) ((i >>> 7) & 0x7f | 0x80);
                    buffer[2] = (byte) (i >>> 14);
                    assert buffer[2] <= 0x7f;
                    length = 3;
                    break;
                case 10:
                case 9:
                case 8:
                case 7:
                case 6:
                case 5:
                case 4:
                    buffer[0] = (byte) (i & 0x7f | 0x80);
                    buffer[1] = (byte) ((i >>> 7) & 0x7f | 0x80);
                    buffer[2] = (byte) ((i >>> 14) & 0x7f | 0x80);
                    buffer[3] = (byte) (i >>> 21);
                    assert buffer[3] <= 0x7f;
                    length = 4;
                    break;
                case 3:
                case 2:
                case 1:
                case 0:
                    buffer[0] = (byte) (i & 0x7f | 0x80);
                    buffer[1] = (byte) ((i >>> 7) & 0x7f | 0x80);
                    buffer[2] = (byte) ((i >>> 14) & 0x7f | 0x80);
                    buffer[3] = (byte) ((i >>> 21) & 0x7f | 0x80);
                    buffer[4] = (byte) (i >>> 28);
                    assert buffer[4] <= 0x7f;
                    length = 5;
                    break;
                default:
                    throw new UnsupportedOperationException(
                            "Can't encode [" + i + "]. Missing case for [" + Integer.numberOfLeadingZeros(i) + "]?"
                    );
            }
            writeBytes(buffer, 0, length);
        }

but I can't measure a performance difference to the loop at all so personally I'd go for the shorter loop just for simplicity's sake.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think your right about your implementation being faster in practice. I put together a quick and dirty benchmark for writeVInt directly and my hand unrolled thing is faster there. By a pretty wide margin. But the benchmark for serializing the agg result is slower. I can see in the decompiled output that my method results in writeVInt not being inlined because it is too big like you say. And yours gets it inlined.

Optimize for size
Benchmark                                        (buckets)  Mode  Cnt   Score   Error  Units
StringTermsSerializationBenchmark.serialize           1000  avgt   10  59.064 ± 0.360  ms/op
StringTermsSerializationBenchmark.serializeVint       1000  avgt   10  17.211 ± 0.088  ms/op

Unroll loops
Benchmark                                        (buckets)  Mode  Cnt   Score   Error  Units
StringTermsSerializationBenchmark.serialize           1000  avgt   10  61.560 ± 0.124  ms/op
StringTermsSerializationBenchmark.serializeVint       1000  avgt   10  11.775 ± 0.048  ms/op

Unroll loops with if instead of switch
Benchmark                                        (buckets)  Mode  Cnt   Score   Error  Units
StringTermsSerializationBenchmark.serialize           1000  avgt   10  60.794 ± 1.069  ms/op
StringTermsSerializationBenchmark.serializeVint       1000  avgt   10  17.703 ± 0.075  ms/op

Compromise
Benchmark                                        (buckets)  Mode  Cnt   Score   Error  Units
StringTermsSerializationBenchmark.serialize           1000  avgt   10  62.106 ± 0.173  ms/op
StringTermsSerializationBenchmark.serializeVint       1000  avgt   10  16.425 ± 0.033  ms/op

The compromise solution doesn't seem to shrink the method enough.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for testing this!

return;
}
byte[] buffer = scratch.get();
buffer[0] = (byte) (i & 0x7f | 0x80);
buffer[1] = (byte) ((i >>> 7) & 0x7f | 0x80);
buffer[2] = (byte) ((i >>> 14) & 0x7f | 0x80);
buffer[3] = (byte) ((i >>> 21) & 0x7f | 0x80);
buffer[4] = (byte) (i >>> 28);
assert buffer[4] <= 0x7f;
writeBytes(buffer, 0, 5);
return;
}
buffer[index++] = ((byte) i);
writeBytes(buffer, 0, index);
byte[] buffer = scratch.get();
if (i < 0x3fff) {
buffer[0] = (byte) (i & 0x7f | 0x80);
buffer[1] = (byte) (i >>> 7);
assert buffer[1] <= 0x7f;
writeBytes(buffer, 0, 2);
return;
}
if (i < 0x1f_ffff) {
buffer[0] = (byte) (i & 0x7f | 0x80);
buffer[1] = (byte) ((i >>> 7) & 0x7f | 0x80);
buffer[2] = (byte) (i >>> 14);
assert buffer[2] <= 0x7f;
writeBytes(buffer, 0, 3);
return;
}
if (i < 0x0fff_ffff) {
buffer[0] = (byte) (i & 0x7f | 0x80);
buffer[1] = (byte) ((i >>> 7) & 0x7f | 0x80);
buffer[2] = (byte) ((i >>> 14) & 0x7f | 0x80);
buffer[3] = (byte) (i >>> 21);
assert buffer[3] <= 0x7f;
writeBytes(buffer, 0, 4);
return;
}
buffer[0] = (byte) ((i & 0x7f) | 0x80);
buffer[1] = (byte) ((i >>> 7) & 0x7f | 0x80);
buffer[2] = (byte) ((i >>> 14) & 0x7f | 0x80);
buffer[3] = (byte) ((i >>> 21) & 0x7f | 0x80);
buffer[4] = (byte) (i >>> 28);
assert buffer[4] <= 0x7f;
writeBytes(buffer, 0, 5);
}

/**
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -61,7 +61,7 @@
import static org.hamcrest.Matchers.sameInstance;

/**
* Tests for {@link BytesStreamOutput} paging behaviour.
* Tests for {@link StreamOutput}.
*/
public class BytesStreamsTests extends ESTestCase {
public void testEmpty() throws Exception {
Expand Down Expand Up @@ -829,6 +829,15 @@ public void testVInt() throws IOException {
output.writeVInt(value);
StreamInput input = output.bytes().streamInput();
assertEquals(value, input.readVInt());

BytesStreamOutput simple = new BytesStreamOutput();
int i = value;
while ((i & ~0x7F) != 0) {
simple.writeByte(((byte) ((i & 0x7f) | 0x80)));
i >>>= 7;
}
simple.writeByte((byte) i);
assertEquals(simple.bytes().toBytesRef().toString(), output.bytes().toBytesRef().toString());
}

public void testVLong() throws IOException {
Expand Down