VTune what analysis produces the DFLOP count?

VTune what analysis produces the DFLOP count?


What VTune analysis produces the DFLOP count for a given application? this would be extremely useful for efortlessly computing the GFlops,



Best regards,


6 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

There is no associated events to measure flops in predefined analysis. However you can find/use events in supported events, it depends on your using processor. For example, if you working on SandBridge processor, use:

# amplxe-runss -event-list | grep FP_COMP_OPS_EXE

So, MFLOPS Formula = FP_COMP_OPS_EXE.X87 / 1x10e-6 / Elapsed Time

Using event FP_COMP_OPS_EXE.X87 directly in command line without VTune's predefined analysis, refer this article


For information about using FLOPS related events for your processor, read under VTune_amplifier_xe_2013/documentation/help/reference

"What VTune analysis produces the DFLOP count for a given application?"

I presume that DFLOP stands for double precision FLOPs.

Correct I am interested in Double Precision FLoating Point Operations therefore FP_COMP_OPS_EXE.SSE_SCALAR_SINGLE and FP_COMP_OPS_EXE.SSE_PACKED_SINGLE are of no interest to me. Furthermore, this PC seems to give the total FP_COMP_OPS_EXE.X87 that includes single precision, which is also not interesting to me. I only only want to measure Double Precision events.

How can I do that using VTune?

OK I found out how to do this. The solution is to create a new Analysis or custom analysis type and add those counters.




Best Reply

I am late with the answer,but anyway it is this counter FP_COMP_OPS_EXE.SSE_PACKED_DOUBLE.

Leave a Comment

Please sign in to add a comment. Not a member? Join today