diff --git a/CMakeLists.txt b/CMakeLists.txt
index 67d96eb2..ffd682b3 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -6,7 +6,7 @@ endif()
 project(PIP)
 set(PIP_MAJOR    4)
 set(PIP_MINOR    7)
-set(PIP_REVISION 1)
+set(PIP_REVISION 2)
 set(PIP_SUFFIX   )
 set(PIP_COMPANY  SHS)
 set(PIP_DOMAIN   org.SHS)
diff --git a/libs/main/thread/pithreadpoolloop.cpp b/libs/main/thread/pithreadpoolloop.cpp
index 1e364e4a..b8e1494f 100644
--- a/libs/main/thread/pithreadpoolloop.cpp
+++ b/libs/main/thread/pithreadpoolloop.cpp
@@ -106,16 +106,29 @@
 PIThreadPoolLoop::PIThreadPoolLoop(int thread_cnt) {
 	if (thread_cnt <= 0) thread_cnt = piMaxi(1, PISystemInfo::instance()->processorsCount);
 	piForTimes(thread_cnt) {
-		auto * t = new PIThread();
+		auto * t = new PIThread([this]() {
+			while (true) {
+				sem_exec.acquire();
+				if (is_destroy) return;
+				int cc = counter.fetch_add(1);
+				func(cc);
+				sem_done.release();
+			}
+		});
 		threads << t;
 	}
+	for (auto * t: threads)
+		t->start();
 	// piCout << "PIThreadPoolLoop" << proc_cnt << "threads";
 }
 
 
 PIThreadPoolLoop::~PIThreadPoolLoop() {
-	for (auto * t: threads) {
+	is_destroy = true;
+	for (auto * t: threads)
 		t->stop();
+	sem_exec.release(threads.size());
+	for (auto * t: threads) {
 		if (!t->waitForFinish(100_ms)) t->terminate();
 		delete t;
 	}
@@ -127,20 +140,19 @@ void PIThreadPoolLoop::setFunction(std::function<void(int)> f) {
 }
 
 
+void PIThreadPoolLoop::wait() {
+	// piCout << "wait" << wait_count;
+	if (wait_count <= 0) return;
+	sem_done.acquire(wait_count);
+	wait_count = 0;
+	// piCout << "wait done";
+}
+
+
 void PIThreadPoolLoop::start(int index_start, int index_count) {
-	counter = index_start;
-	int end = index_start + index_count;
-	for (auto * t: threads)
-		t->start([this, end, t]() {
-			while (1) {
-				int cc = counter.fetch_add(1);
-				if (cc >= end) {
-					t->stop();
-					return;
-				}
-				func(cc);
-			}
-		});
+	counter    = index_start;
+	wait_count = index_count;
+	sem_exec.release(index_count);
 }
 
 
@@ -154,9 +166,3 @@ void PIThreadPoolLoop::exec(int index_start, int index_count, std::function<void
 	setFunction(f);
 	exec(index_start, index_count);
 }
-
-
-void PIThreadPoolLoop::wait() {
-	for (auto * t: threads)
-		t->waitForFinish();
-}
diff --git a/libs/main/thread/pithreadpoolloop.h b/libs/main/thread/pithreadpoolloop.h
index 0dbbb606..d79af806 100644
--- a/libs/main/thread/pithreadpoolloop.h
+++ b/libs/main/thread/pithreadpoolloop.h
@@ -26,6 +26,7 @@
 #ifndef PITHREADPOOLLOOP_H
 #define PITHREADPOOLLOOP_H
 
+#include "pisemaphore.h"
 #include "pivector.h"
 
 class PIThread;
@@ -83,7 +84,9 @@ public:
 private:
 	PIVector<PIThread *> threads;
 	std::function<void(int)> func;
-	std::atomic_int counter = {0};
+	PISemaphore sem_exec, sem_done;
+	std::atomic_bool is_destroy = {false};
+	std::atomic_int counter = {0}, wait_count = {0};
 };