GPU benchmarks should track the frequency of impl-thread vs. main-thread scrolls

content/renderer/gpu/gpu_benchmarking_extension.cc

Index: content/renderer/gpu/gpu_benchmarking_extension.cc
diff --git a/content/renderer/gpu/gpu_benchmarking_extension.cc b/content/renderer/gpu/gpu_benchmarking_extension.cc
index 495812531afe45ae585f25cf7b6d572833583cde..cb562a7cd9fb6d8c4a947d784eb792719d414cfd 100644
--- a/content/renderer/gpu/gpu_benchmarking_extension.cc
+++ b/content/renderer/gpu/gpu_benchmarking_extension.cc
@@ -170,6 +170,8 @@ class GpuBenchmarkingWrapper : public v8::Extension {
                       v8::Number::New(stats.totalCommitTimeInSeconds));
     stats_object->Set(v8::String::New("totalCommitCount"),
                       v8::Integer::New(stats.totalCommitCount));
+    stats_object->Set(v8::String::New("didImplThreadScroll"),
+                      v8::Boolean::New(stats.didImplThreadScroll));
 
     stats_object->Set(v8::String::New("globalTextureUploadCount"),
                       v8::Number::New(gpu_stats.global_texture_upload_count));