perf(query): Improve Intersection with UID pack

algo/uidlist.go

@@ -105,34 +105,45 @@ func IntersectCompressedWithBin(dec *codec.Decoder, q []uint64, o *[]uint64) {
 
 	// Pick the shorter list and do binary search
 	if ld < lq {
-		uids := dec.Uids()
-		for len(uids) > 0 {
-			for _, u := range uids {
-				qidx := sort.Search(len(q), func(idx int) bool {
-					return q[idx] >= u
-				})
-				if qidx >= len(q) {
-					return
-				}
-				if q[qidx] == u {
-					*o = append(*o, u)
-					qidx++
-				}
-				q = q[qidx:]
+		for {
+			block_uids := dec.Uids()
+			if len(block_uids) == 0 {
+				break
 			}
-			uids = dec.Next()
+			IntersectWithBin(block_uids, q, o)
+			last_uid := block_uids[len(block_uids)-1]
+			qidx := sort.Search(len(q), func(idx int) bool {
+				return q[idx] >= last_uid
+			})
+			if qidx >= len(q) {
+				return
+			}
+			q = q[qidx:]
+			dec.Next()
 		}
 		return
 	}
 
+	uids := dec.Seek(q[0], codec.SeekStart)
 	for _, u := range q {
-		uids := dec.Seek(u, codec.SeekStart)
-		if len(uids) == 0 {
-			return
+		if len(uids) == 0 || u > uids[len(uids)-1] {
+			uids = dec.Seek(u, codec.SeekStart)
+			if len(uids) == 0 {
+				return
+			}
+		}
+		uidIdx := sort.Search(len(uids), func(idx int) bool {
+			return uids[idx] >= u
+		})
+		if uidIdx >= len(uids) {
+			// We know that u < max(uids). If we didn't find it here, it's not here.
+			continue
 		}
-		if uids[0] == u {
+		if uids[uidIdx] == u {
 			*o = append(*o, u)
+			uidIdx++
 		}
+		uids = uids[uidIdx:]
 	}
 }
 

algo/uidlist_test.go

@@ -367,6 +367,52 @@ func BenchmarkListIntersectRandom(b *testing.B) {
 	randomTests(1024000, 0.01)
 }
 
+func BenchmarkListIntersectCompressBin(b *testing.B) {
+	randomTests := func(sz int, overlap float64) {
+		rs := []float64{0.01, 0.1, 1, 10, 100}
+		for _, r := range rs {
+			sz1 := sz
+			sz2 := int(float64(sz) * r)
+			if sz2 > 1000000 || sz2 == 0 {
+				break
+			}
+
+			u1, v1 := make([]uint64, sz1), make([]uint64, sz2)
+			limit := int64(float64(sz) / overlap)
+			for i := 0; i < sz1; i++ {
+				u1[i] = uint64(rand.Int63n(limit))
+			}
+			for i := 0; i < sz2; i++ {
+				v1[i] = uint64(rand.Int63n(limit))
+			}
+			sort.Slice(u1, func(i, j int) bool { return u1[i] < u1[j] })
+			sort.Slice(v1, func(i, j int) bool { return v1[i] < v1[j] })
+
+			dst2 := &pb.List{}
+			compressedUids := codec.Encode(v1, 256)
+
+			b.Run(fmt.Sprintf("compressed:IntersectWith:ratio=%v:size=%d:overlap=%.2f:", r, sz, overlap),
+				func(b *testing.B) {
+					for k := 0; k < b.N; k++ {
+						dec := codec.Decoder{Pack: compressedUids}
+						dec.Seek(0, codec.SeekStart)
+						IntersectCompressedWithBin(&dec, u1, &dst2.Uids)
+					}
+				})
+			fmt.Println()
+
+			codec.FreePack(compressedUids)
+		}
+	}
+
+	randomTests(10, 0.01)
+	randomTests(100, 0.01)
+	randomTests(1000, 0.01)
+	randomTests(10000, 0.01)
+	randomTests(100000, 0.01)
+	randomTests(1000000, 0.01)
+}
+
 func BenchmarkListIntersectRatio(b *testing.B) {
 	randomTests := func(sz int, overlap float64) {
 		rs := []int{1, 10, 50, 100, 500, 1000, 10000, 100000, 1000000}