Intel® Quartus® Prime Software
Intel® Quartus® Prime Design Software, Design Entry, Synthesis, Simulation, Verification, Timing Analysis, System Design (Platform Designer, formerly Qsys)
Announcements
Intel Support hours are Monday-Fridays, 8am-5pm PST, except Holidays. Thanks to our community members who provide support during our down time or before we get to your questions. We appreciate you!

Need Forum Guidance? Click here
Search our FPGA Knowledge Articles here.
15553 Discussions

clfinish is taking more time than kernel execution time

Altera_Forum
Honored Contributor II
800 Views

Hi All, 

I am working on a application where every millisec counts. Please refer to the below data.  

 

 

Time clFinish 40889 micro seconds 

OpenCl Execution time is: 32.301 milliseconds 

Time ReadBuffer 234 

 

 

 

 

Time clFinish 41386 micro seconds 

OpenCl Execution time is: 31.889 milliseconds 

Time ReadBuffer 331 

 

 

The kernel execution above is taking ~32msec, I have printed this using clGetEventProfilingInfo. But when I check the clfinish time it is close to ~40msec. 

As you can see above here there is a gap of 8msec from clfinish and kernel execution time. Even in profile.mon file the kernel execution status is close to 32msec.  

Can you please help me understand where I am loosing 8msec, as per my understanding the clfinish time should equivalent to that of the Kernel execution time as I dont have any pending Queues in commandQueue.  

 

 

 

Thanks in advance.
0 Kudos
1 Reply
Altera_Forum
Honored Contributor II
96 Views

Is this with a kernel that has profiling enabled? Profiling has extra overhead since it needs to dump the profiling info onto the disk.

Reply