Merge "Openmp blocktime 1->20ms to keep cores enabled" into pi-dev

3 files changed, 12 insertions, 10 deletions

diff --git a/nn/common/CpuExecutor.cpp b/nn/common/CpuExecutor.cpp
index 0c6219308..8f8706776 100644
--- a/nn/common/CpuExecutor.cpp
+++ b/nn/common/CpuExecutor.cpp
@@ -1536,7 +1536,7 @@ int CpuExecutor::executeOperation(const Operation& operation) {
 
 ScopedOpenmpSettings::ScopedOpenmpSettings() {
     mBlocktimeInitial = kmp_get_blocktime();
-    kmp_set_blocktime(1);  // ms
+    kmp_set_blocktime(20);  // ms, see b/109645291
 
 #if NNAPI_LIMIT_CPU_THREADS
     // Code not yet enabled. Choosing the number of threads to be based on
diff --git a/nn/common/include/CpuExecutor.h b/nn/common/include/CpuExecutor.h
index 64a46b65f..78b8910ea 100644
--- a/nn/common/include/CpuExecutor.h
+++ b/nn/common/include/CpuExecutor.h
@@ -142,9 +142,9 @@ private:
 //
 // Currently sets a low blocktime: the time OpenMP threads busy-wait for more
 // work before going to sleep. See b/79159165, https://reviews.llvm.org/D18577.
-// The default is 200ms, we set to 1ms here. This should allow for the threads
-// to not sleep before the next operation, but release CPU to other work
-// quickly.
+// The default is 200ms, we set to 20ms here, see b/109645291. This keeps the
+// cores enabled throughout inference computation without too much extra power
+// consumption afterwards.
 //
 // The OpenMP settings are thread-local (applying only to worker threads formed
 // from that thread), see https://software.intel.com/en-us/node/522688 and
diff --git a/nn/runtime/test/TestOpenmpSettings.cpp b/nn/runtime/test/TestOpenmpSettings.cpp
index 549473b46..59a794224 100644
--- a/nn/runtime/test/TestOpenmpSettings.cpp
+++ b/nn/runtime/test/TestOpenmpSettings.cpp
@@ -38,26 +38,28 @@ protected:
         ASSERT_EQ(blocktimeRestored, kOpenmpDefaultBlockTime);
     }
     static const int kOpenmpDefaultBlockTime;
+    static const int kPreferredBlockTime;
 };
 
 const int OpenmpSettingsTest::kOpenmpDefaultBlockTime = 200;
+const int OpenmpSettingsTest::kPreferredBlockTime = 20;
 
 using ::android::nn::ScopedOpenmpSettings;
 
-TEST_F(OpenmpSettingsTest, Test1) {
+TEST_F(OpenmpSettingsTest, TestkPreferredBlockTime) {
     ScopedOpenmpSettings s;
     const int blocktimeSet = kmp_get_blocktime();
-    ASSERT_EQ(blocktimeSet, 1);
+    ASSERT_EQ(blocktimeSet, kPreferredBlockTime);
 }
 
 TEST_F(OpenmpSettingsTest, Test2) {
     ScopedOpenmpSettings s1;
     const int blocktimeSet1 = kmp_get_blocktime();
-    ASSERT_EQ(blocktimeSet1, 1);
+    ASSERT_EQ(blocktimeSet1, kPreferredBlockTime);
 
     ScopedOpenmpSettings s2;
     const int blocktimeSet2 = kmp_get_blocktime();
-    ASSERT_EQ(blocktimeSet2, 1);
+    ASSERT_EQ(blocktimeSet2, kPreferredBlockTime);
 }
 
 TEST_F(OpenmpSettingsTest, TestThreaded) {
@@ -74,12 +76,12 @@ TEST_F(OpenmpSettingsTest, TestThreaded) {
             ScopedOpenmpSettings s;
 
             const int blocktimeSet2 = kmp_get_blocktime();
-            ASSERT_EQ(blocktimeSet2, 1);
+            ASSERT_EQ(blocktimeSet2, kPreferredBlockTime);
 
             usleep(sleepFor);
 
             const int blocktimeSet3 = kmp_get_blocktime();
-            ASSERT_EQ(blocktimeSet3, 1);
+            ASSERT_EQ(blocktimeSet3, kPreferredBlockTime);
         }));
     }
     std::for_each(threads.begin(), threads.end(), [](std::thread& t) {