1
0
mirror of https://github.com/opencv/opencv.git synced 2026-07-30 07:43:03 +04:00

--perf_instrument parameter now has int type and 0, 1, 2 modes (1 - simple trees, 2 - expanded trees for functions with same name but different calling address);

Maximum depth limit var was added to the instrumentation structure;

Trace names output console output fix: improper tree formatting could happen;
Output in case of error was added;

Custom regions improvements;

Improved timing and weight calculation for parallel regions; New TC (threads counter) value to indicate how many different threads accessed particular node;

parallel_for, warnings fixes and ReturnAddress code from Alexander Alekhin;
This commit is contained in:
Pavel Vlasov
2016-11-07 12:15:51 +03:00
parent 442380bfac
commit 349d5ba012
13 changed files with 211 additions and 161 deletions
+2 -2
View File
@@ -3450,7 +3450,7 @@ int Kernel::set(int i, const KernelArg& arg)
bool Kernel::run(int dims, size_t _globalsize[], size_t _localsize[],
bool sync, const Queue& q)
{
CV_INSTRUMENT_REGION_META(p->name.c_str(), instr::TYPE_FUN, instr::IMPL_OPENCL);
CV_INSTRUMENT_REGION_OPENCL_RUN(p->name.c_str());
if(!p || !p->handle || p->e != 0)
return false;
@@ -3563,7 +3563,7 @@ struct Program::Impl
Impl(const ProgramSource& _src,
const String& _buildflags, String& errmsg)
{
CV_INSTRUMENT_REGION_OPENCL_(cv::format("Compile: %" PRIx64 " options: %s", _src.hash(), _buildflags.c_str()).c_str());
CV_INSTRUMENT_REGION_OPENCL_COMPILE(cv::format("Compile: %" PRIx64 " options: %s", _src.hash(), _buildflags.c_str()).c_str());
refcount = 1;
const Context& ctx = Context::getDefault();
src = _src;