Software Tuning, Performance Optimization & Platform Monitoring
Discussion around monitoring and software tuning methodologies, Performance Monitoring Unit (PMU) of Intel microprocessors, and platform monitoring
Announcements
This community is designed for sharing of public information. Please do not share Intel or third-party confidential information here.

Performance monitoring through intel VTune amplifier

Saumya_B_
Beginner
130 Views

Hie all,

I am new to VTune. I wanted to analyze program on intel xeon-phi coprocessor, for which I executed the following command:

amplxe-cl -collect-with runsa -knob event config=CPU_CLK_UNHALTED.THREAD,INST_RETIRED.ANY,MEM_LOAD_UOPS_RETIRED.L1_HIT,MEM_LOAD_UOPS_RETIRED.L2_HIT,MEM_LOAD_UOPS_RETIRED.L1_MISS,MEM_LOAD_UOPS_RETIRED.L2_MISS,MEM_UOPS_RETIRED.ALL_LOADS,MEM_UOPS_RETIRED.ALL_STORES,MOVE_ELIMINATION.SIMD_ELIMINATED,CYCLE_ACTIVITY.CYCLES_L1D_PENDING,L2_RQSTS.L2_PF_MISS,L2_RQSTS.L2_PF_HIT -- ssh mic0 "export OMP_NUM_THREADS=8; ./a.out"

But I get the following results :

INST_RETIRED.ANY                                    60000090                                30               2000003          
CPU_CLK_UNHALTED.THREAD                             40000060                  20               2000003          
CPU_CLK_UNHALTED.REF_TSC                            40000060                 20               2000003          
CYCLE_ACTIVITY.CYCLES_L1D_PENDING                          0                   0               2000003          
MEM_LOAD_UOPS_RETIRED.L1_HIT                               0                        0               2000003          
MEM_LOAD_UOPS_RETIRED.L2_HIT                               0                        0               100003           
MEM_LOAD_UOPS_RETIRED.L1_MISS                              0                      0               100003           
MEM_LOAD_UOPS_RETIRED.L2_MISS                              0                      0               50021            
MEM_UOPS_RETIRED.ALL_LOADS                          42000063                 7               2000003          
MEM_UOPS_RETIRED.ALL_STORES                         18000027               3              2000003          
MOVE_ELIMINATION.SIMD_ELIMINATED                           0                      0             1000003          
L2_RQSTS.L2_PF_MISS                                        0                                       0             200003           
L2_RQSTS.L2_PF_HIT                                         0                                          0            200003           

Any reason why so many values are zero?

 

0 Kudos
2 Replies
Dmitry_R_Intel1
Employee
130 Views

You can't profile application on MIC in such way. Your command actually profiled the ssh process running on the host which did almost nothing except i/o. Thus very small or zero event counts.

Please use '-target-system mic-host-launch:mic0' option, e.g.:

amplxe-cl -target-system mic-host-launch:mic0 -collect-with runsa -knob event-config=CPU_CLK_UNHALTED:sa=2000000,INSTRUCTIONS_EXECUTED:sa=2000000 -- a.out

More info on this in VTune OLH:

https://software.intel.com/en-us/node/596763

https://software.intel.com/en-us/node/596531

 

Saumya_B_
Beginner
130 Views

Thanks a lot

Reply