Some more optimizations. Also I realized that print uses stderr instead of stdout, fixed that.

main.go

@@ -30,9 +30,9 @@ func main() {
 
 	// Make a channel and a thread for each CPU core
 	var cores = runtime.NumCPU()
-	var linechan = make(chan []byte)
+	var workchannel = make(chan []byte)
 	for i := 0; i < cores; i++ {
-		go scannerThread(linechan, []byte(needle), &wg)
+		go scannerThread(workchannel, []byte(needle), &wg)
 		wg.Add(1)
 	}
 
@@ -54,37 +54,36 @@ func main() {
 			}
 		}
 
-		linechan <- append(buf[:nread], remainder...)
+		workchannel <- append(buf[:nread], remainder...)
 
 		if err == io.EOF {
 			break
 		}
 	}
 
-	close(linechan)
-
+	close(workchannel)
 	wg.Wait()
 }
 
-func scannerThread(linechannel chan []byte, needle []byte, wg *sync.WaitGroup) {
-	var line []byte
+func scannerThread(workchannel chan []byte, needle []byte, wg *sync.WaitGroup) {
+	var batch []byte
 	var ok bool
 	var start, end int
 	var linelen int
 	var i int
 	defer wg.Done()
 	for {
-		line, ok = <-linechannel
+		batch, ok = <-workchannel
 		if !ok {
 			return // channel closed. we're done
 		}
 
 		// This loop is for every result found, when there are no more results it stops
-		for i = bytes.Index(line, needle); i != -1; i = bytes.Index(line, needle) {
+		for i = bytes.Index(batch, needle); i != -1; i = bytes.Index(batch, needle) {
 			start, end = i, i
 			// needle was found, but where?
 			for { // find the start
-				if line[start] == lf {
+				if batch[start] == lf {
 					start++ // the line feed is from the previous line, so skip it
 					break
 				} else if start == 0 {
@@ -94,18 +93,22 @@ func scannerThread(linechannel chan []byte, needle []byte, wg *sync.WaitGroup) {
 			}
 
 			for { // find the end
-				if line[end] == lf || end == linelen-1 {
+				if batch[end] == lf {
+					end++ // include the line feed in the line
+					// https://stackoverflow.com/questions/26857582/in-a-go-slice-why-does-slohi-end-at-element-hi-1
+					break
+				} else if end == linelen {
 					break
 				}
 				end++
 			}
 
-			print(string(line[start:end]) + "\n")
+			os.Stdout.Write(batch[start:end])
 
 			// Chop all of the bytes before the result off so it doesn't get
 			// searched again
-			line = line[end:]
-			linelen = len(line)
+			batch = batch[end:]
+			linelen = len(batch)
 		}
 	}
 }