From 47a4c95f3e35cc6e85b5182e3557cd19102502eb Mon Sep 17 00:00:00 2001
From: dtookey <hocestpotest@gmail.com>
Date: Thu, 1 Aug 2024 08:46:19 -0400
Subject: [PATCH] ~5.5s performance checkin

---
 1brc.go | 67 +++++++++++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 60 insertions(+), 7 deletions(-)

diff --git a/1brc.go b/1brc.go
index 35542f8..b9307e1 100644
--- a/1brc.go
+++ b/1brc.go
@@ -19,7 +19,7 @@ const testFile = "dummy.txt"
 const resultFile = "my_results.txt"
 const benchFile = "results.txt"
 
-const profile = true
+const profile = false
 
 var nGoRoutine = 64
 
@@ -32,6 +32,17 @@ type partitionRange struct {
 
 type resultSet map[uint64]result
 
+type fRS struct {
+	v    []result
+	keys []uint64
+	size int
+}
+
+func newFRS() *fRS {
+	const size = 50000
+	return &fRS{v: make([]result, size), keys: make([]uint64, size)}
+}
+
 type result struct {
 	name  []byte
 	count float64
@@ -69,7 +80,7 @@ func oneBRC() {
 
 	parts := createPartitions(bits, '\n', nGoRoutine)
 	wg := &sync.WaitGroup{}
-	results := make([]resultSet, len(parts))
+	results := make([]*fRS, len(parts))
 
 	for i, part := range parts {
 		wg.Add(1)
@@ -80,7 +91,7 @@ func oneBRC() {
 	finalRS := make(resultSet)
 
 	for _, rs := range results {
-		finalRS.merge(rs)
+		finalRS.merge(rs.toResultSet())
 	}
 
 	os.WriteFile(resultFile, []byte(finalRS.String()), 0666)
@@ -100,13 +111,17 @@ func validate() {
 	meLines := bytes.Split(me, []byte{'\r', '\n'})
 	refLines := bytes.Split(ref, []byte{'\r', '\n'})
 
+	errs := 0
 	for i, line := range meLines {
 		refLine := string(refLines[i])
 		meLine := string(line)
 		if refLine != meLine {
+			errs++
 			fmt.Printf("Validation tripped: reference[%s]\tme[%s]\n", refLine, meLine)
 		}
 	}
+
+	fmt.Printf("Validation passed: %v\n", errs == 0)
 }
 
 func startCpuProfile() *os.File {
@@ -127,11 +142,11 @@ func stopProfiling(f *os.File) {
 	f.Close()
 }
 
-func workerComputePartition(aData []byte, wPart partitionRange, workerNumber int, container []resultSet, wg *sync.WaitGroup) {
+func workerComputePartition(aData []byte, wPart partitionRange, workerNumber int, container []*fRS, wg *sync.WaitGroup) {
 	defer wg.Done()
 	const delimiter = byte(';')
 
-	rs := make(resultSet)
+	rs := newFRS()
 
 	alloc := make([]result, 500)
 	aCnt := 0
@@ -161,7 +176,7 @@ func workerComputePartition(aData []byte, wPart partitionRange, workerNumber int
 		hasher.Write(line[:di])
 		key := hasher.Sum64()
 		hasher.Reset()
-		r, ok := rs[key]
+		r, ok := rs.get(key)
 		if !ok {
 			r = alloc[aCnt]
 			r.name = line[:di]
@@ -181,7 +196,7 @@ func workerComputePartition(aData []byte, wPart partitionRange, workerNumber int
 			r.min = temp
 		}
 
-		rs[key] = r
+		rs.put(key, r)
 		start = end
 		if end == wPart.end {
 			break
@@ -352,6 +367,7 @@ func (rs resultSet) String() string {
 	sort.Strings(keys)
 	b := &strings.Builder{}
 	hasher := fnv.New64()
+
 	for i, key := range keys {
 		hasher.Write([]byte(key))
 		r := rs[hasher.Sum64()]
@@ -370,3 +386,40 @@ func (rs resultSet) String() string {
 	}
 	return b.String()
 }
+
+func (f *fRS) get(key uint64) (result, bool) {
+	p := f.v[f.keyMap(key)]
+	return p, p.name != nil
+}
+
+func (f *fRS) put(key uint64, v result) {
+	idx := f.keyMap(key)
+	f.v[idx] = v
+	f.keys[idx] = key
+	f.size += 1
+}
+
+func (f *fRS) keyMap(key uint64) uint64 {
+	//a := key & math.MaxUint32
+	//b := (key >> 32) & math.MaxUint32
+	return (key) % uint64(len(f.v))
+}
+
+func (f *fRS) getKeys() []uint64 {
+	keys := make([]uint64, 0, f.size)
+	for _, k := range f.keys {
+		if k > 0 {
+			keys = append(keys, k)
+		}
+	}
+	return keys
+}
+
+func (f *fRS) toResultSet() resultSet {
+	rs := make(resultSet)
+	for _, key := range f.getKeys() {
+		v, _ := f.get(key)
+		rs[key] = v
+	}
+	return rs
+}