diff --git a/benchmark/worker/benchmark_client.go b/benchmark/worker/benchmark_client.go
index 1a05ec0d..0067d5eb 100644
--- a/benchmark/worker/benchmark_client.go
+++ b/benchmark/worker/benchmark_client.go
@@ -69,6 +69,7 @@ func printClientConfig(config *testpb.ClientConfig) {
 	// - core list
 	grpclog.Printf(" * client type: %v (ignored, always creates sync client)", config.ClientType)
 	grpclog.Printf(" * async client threads: %v (ignored)", config.AsyncClientThreads)
+	// TODO: use cores specified by CoreList when setting list of cores is supported in go.
 	grpclog.Printf(" * core list: %v (ignored)", config.CoreList)
 
 	grpclog.Printf(" - security params: %v", config.SecurityParams)
@@ -87,6 +88,8 @@ func setupClientEnv(config *testpb.ClientConfig) {
 	// TODO: Revisit this for the optimal default setup.
 	if config.CoreLimit > 0 {
 		runtime.GOMAXPROCS(int(config.CoreLimit))
+	} else {
+		runtime.GOMAXPROCS(runtime.NumCPU())
 	}
 }
 
diff --git a/benchmark/worker/benchmark_server.go b/benchmark/worker/benchmark_server.go
index 5f4da0e6..0a928a11 100644
--- a/benchmark/worker/benchmark_server.go
+++ b/benchmark/worker/benchmark_server.go
@@ -70,6 +70,7 @@ func printServerConfig(config *testpb.ServerConfig) {
 	// - core list
 	grpclog.Printf(" * server type: %v (ignored, always starts sync server)", config.ServerType)
 	grpclog.Printf(" * async server threads: %v (ignored)", config.AsyncServerThreads)
+	// TODO: use cores specified by CoreList when setting list of cores is supported in go.
 	grpclog.Printf(" * core list: %v (ignored)", config.CoreList)
 
 	grpclog.Printf(" - security params: %v", config.SecurityParams)
@@ -86,8 +87,8 @@ func startBenchmarkServer(config *testpb.ServerConfig, serverPort int) (*benchma
 	numOfCores := runtime.NumCPU()
 	if config.CoreLimit > 0 {
 		numOfCores = int(config.CoreLimit)
-		runtime.GOMAXPROCS(numOfCores)
 	}
+	runtime.GOMAXPROCS(numOfCores)
 
 	var opts []grpc.ServerOption