#第二个RDS IPC只有第三个的30%多点，这就是为什么CPU高这么多，但是QPS差不多
perf stat -e branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-store-misses,L1-dcache-stores,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,iTLB-load-misses  -a -p 61238
^C
 Performance counter stats for process id '61238':

        86,491,052      branch-misses                                                 (58.55%)
    98,481,418,793      bus-cycles                                                    (55.64%)
       113,095,618      cache-misses              #    6.169 % of all cache refs      (53.20%)
     1,833,344,484      cache-references                                              (52.00%)
   101,516,165,898      cpu-cycles                                                    (57.09%)
     4,229,190,014      instructions              #    0.04  insns per cycle          (55.91%)
       111,780,025      L1-dcache-load-misses     #    6.34% of all L1-dcache hits    (55.40%)
     1,764,421,570      L1-dcache-loads                                               (52.62%)
       112,261,128      L1-dcache-store-misses                                        (49.34%)
     1,814,998,338      L1-dcache-stores                                              (48.51%)
       219,372,119      L1-icache-load-misses                                         (49.56%)
     2,816,279,627      L1-icache-loads                                               (49.15%)
        85,321,093      branch-load-misses                                            (50.38%)
     1,038,572,653      branch-loads                                                  (50.65%)
        45,166,831      dTLB-load-misses                                              (51.98%)
        29,892,473      iTLB-load-misses                                              (52.56%)

       1.163750756 seconds time elapsed

#第三个RDS
perf stat -e branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-store-misses,L1-dcache-stores,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,iTLB-load-misses  -a -p 53400
^C
 Performance counter stats for process id '53400':

       295,575,513      branch-misses                                                 (40.51%)
   110,934,600,206      bus-cycles                                                    (39.30%)
       537,938,496      cache-misses              #    8.310 % of all cache refs      (38.99%)
     6,473,688,885      cache-references                                              (39.80%)
   110,540,950,757      cpu-cycles                                                    (46.10%)
    14,766,013,708      instructions              #    0.14  insns per cycle          (46.85%)
       538,521,226      L1-dcache-load-misses     #    8.36% of all L1-dcache hits    (48.00%)
     6,440,728,959      L1-dcache-loads                                               (46.69%)
       533,693,357      L1-dcache-store-misses                                        (45.91%)
     6,413,111,024      L1-dcache-stores                                              (44.92%)
       673,725,952      L1-icache-load-misses                                         (42.76%)
     9,216,663,639      L1-icache-loads                                               (38.27%)
       299,202,001      branch-load-misses                                            (37.62%)
     3,285,957,082      branch-loads                                                  (36.10%)
       149,348,740      dTLB-load-misses                                              (35.20%)
       102,444,469      iTLB-load-misses                                              (34.78%)

       8.080841166 seconds time elapsed