* Following Perf stats were collected ------------------------------------------- 0 - Memory Reservation statistics: CPU_COMMITTED_INST CPU_CYCLE_COUNT L2_LWARX_COMPLETE L2_STWCX_SUCCESS L2_READ_AFTER_WRITE L2_WRITE_AFTER_WRITE 1 - L1 Cache performance L2_MISS_I_FETCH L2_HIT_I_FETCH CPU_ICACHE_HIT L2_MISS_D_FETCH L2_MISS_STORE L2_HIT_STORE L2_HIT_D_FETCH CPU_DCACHE_HIT 2 - L2 Cache performance L2_MISS_D_FETCH L2_MISS_I_FETCH L2_MISS_STORE L2_HIT_D_FETCH L2_HIT_I_FETCH L2_HIT_STORE 3 - Shadow TLB misses - L2 Miss Eviction CPU_COMMITTED_INST CPU_CYCLE_COUNT CPU_DTLB_RELOAD CPU_ITLB_RELOAD L2_MISS_EVICTION L2_MISS_D_FETCH L2_MISS_I_FETCH L2_MISS_STORE 4 - PLB6 Commands PLB_MASTER_COMMAND PLB_MASTER_READ PLB_MASTER_RWITM PLB_MASTER_DCLAIM PLB_MASTER_WRITE PLB_MASTER_INTVN_M PLB_MASTER_INTVN_S PLB_MASTER_MEM_DATA ======================================================= A loop with an active tracepoint inside was executed a million times. Time was taken before entering the loop and after exiting it. Average time was then calculated. At very start, application sleeps 2 seconds to wait that everything is initialized. * LTTng 2.1.1 stats ------------------------------------------- root@du1:~# ./rbs-perf-test -t 0 -s -c ./ltt-test-2.1.1 -l 1000000 -t 1 --- Memory Reservation statistics - exec cmd - ./ltt-test-2.1.1 --- Benchmark test starts now... Time it took to loop 1000000 times : 3 s 586888706 ns | [ 3586 ns per tp ] Stats for pid 2826 Counters enabled in user and in kernel space ------------------------------------------------------------------ 2735477556 Instruction count # 0.48 Instructions per Cycle 5690106496 CPU cycles # 1.596 GHz 3306491 Completed lwarx instructions 3306180 Successful stwcx instructions 2062689 L2 Read after Write 99173776 L2 Write after Write 1245 Context Switches # 0.000 M/sec 2 CPU-migrations # 0.000 M/sec 359 Page Faults # 0.000 M/sec 3566.016640 Task Clock # 0.635 CPUs utilized 5.615581770 seconds time elapsed root@du1:~# ./rbs-perf-test -t 1 -s -c ./ltt-test-2.1.1 -l 1000000 -t 1 --- L1 Cache performance - exec cmd - ./ltt-test-2.1.1 --- Benchmark test starts now... Time it took to loop 1000000 times : 3 s 618456855 ns | [ 3618 ns per tp ] Stats for pid 2846 Counters enabled in user and in kernel space ------------------------------------------------------------------ 1382879363 L1-icache-loads # 387.507 M/sec 1376632814 L1-icache-load-hits # 99.55 of all L1-icache hits 838963771 L1-dcache-loads # 235.092 M/sec 705560597 L1-dcache-load-hits # 84.10 of all L1-dcache hits 1157 Context Switches # 0.000 M/sec 2 CPU-migrations # 0.000 M/sec 358 Page Faults # 0.000 M/sec 3568.655296 Task Clock # 0.632 CPUs utilized 5.645967350 seconds time elapsed root@du1:~# ./rbs-perf-test -t 3 -s -c ./ltt-test-2.1.1 -l 1000000 -t 1 --- Shadow TLB misses - L2 Miss Eviction - exec cmd - ./ltt-test-2.1.1 --- Benchmark test starts now... Time it took to loop 1000000 times : 3 s 492358390 ns | [ 3492 ns per tp ] Stats for pid 2863 Counters enabled in user and in kernel space ------------------------------------------------------------------ 2733834879 Instruction count # 0.50 Instructions per Cycle 5508422280 CPU cycles # 1.595 GHz 91880488 dcache shadow TLB misses # 26.607 M/sec 37082980 icache shadow TLB misses # 10.739 M/sec 66490 L2 Miss Eviction # 56.39 of all L2 cache misses 1046 Context Switches # 0.000 M/sec 2 CPU-migrations # 0.000 M/sec 358 Page Faults # 0.000 M/sec 3453.270784 Task Clock # 0.626 CPUs utilized 5.518885690 seconds time elapsed root@du1:~# ./rbs-perf-test -t 4 -s -c ./ltt-test-2.1.1 -l 1000000 -t 1 --- PLB6 Commands - exec cmd - ./ltt-test-2.1.1 --- Benchmark test starts now... Time it took to loop 1000000 times : 3 s 562387238 ns | [ 3562 ns per tp ] Stats for pid 2877 Counters enabled in user and in kernel space ------------------------------------------------------------------ 6588193 PLB6 Commands # 1.894 M/sec 106741 READ/READ-ATOMIC commands # 0.031 M/sec 11681 RWITM commands # 0.003 M/sec 90533 DCLAIM commands # 0.026 M/sec 13808 Cache line WRITE commands # 0.004 M/sec 8133 Cache line fill IntvnM # 0.002 M/sec 48460 Cache line fill IntvnS # 0.014 M/sec 59714 Cache line fill memory # 0.017 M/sec 2985 Context Switches # 0.001 M/sec 2 CPU-migrations # 0.000 M/sec 359 Page Faults # 0.000 M/sec 3477.678672 Task Clock # 0.622 CPUs utilized 5.589642530 seconds time elapsed ======================================================= * LTTng 2.2.0-rc2 stats ------------------------------------------- root@du1:~# ./rbs-perf-test -t 0 -s -c ./ltt-test-2.2.0 -l 1000000 -t 1 --- Memory Reservation statistics - exec cmd - ./ltt-test-2.2.0 --- Benchmark test starts now... Time it took to loop 1000000 times : 8 s 666169809 ns | [8666 ns per tp] Stats for pid 2737 Counters enabled in user and in kernel space ------------------------------------------------------------------ 6682138400 Instruction count # 0.50 Instructions per Cycle 13486937352 CPU cycles # 1.593 GHz 3378637 Completed lwarx instructions 3378115 Successful stwcx instructions 1116115 L2 Read after Write 419986854 L2 Write after Write 4674 Context Switches # 0.001 M/sec 8 CPU-migrations # 0.000 M/sec 375 Page Faults # 0.000 M/sec 8465.781152 Task Clock # 0.790 CPUs utilized 10.710508740 seconds time elapsed root@du1:~# ./rbs-perf-test -t 1 -s -c ./ltt-test-2.2.0 -l 1000000 -t 1 --- L1 Cache performance - exec cmd - ./ltt-test-2.2.0 --- Benchmark test starts now... Time it took to loop 1000000 times : 8 s 750318129 ns | [8750 ns per tp] Stats for pid 2752 Counters enabled in user and in kernel space ------------------------------------------------------------------ 3076810638 L1-icache-loads # 361.077 M/sec 3042256266 L1-icache-load-hits # 98.88 of all L1-icache hits 2320374878 L1-dcache-loads # 272.306 M/sec 1913574867 L1-dcache-load-hits # 82.47 of all L1-dcache hits 5575 Context Switches # 0.001 M/sec 15 CPU-migrations # 0.000 M/sec 373 Page Faults # 0.000 M/sec 8521.200256 Task Clock # 0.790 CPUs utilized 10.788846760 seconds time elapsed root@du1:~# ./rbs-perf-test -t 2 -s -c ./ltt-test-2.2.0 -l 1000000 -t 1 --- L2 Cache performance - exec cmd - ./ltt-test-2.2.0 --- Benchmark test starts now... Time it took to loop 1000000 times : 8 s 769646969 ns | [8769 ns per tp] Stats for pid 2767 Counters enabled in user and in kernel space ------------------------------------------------------------------ 422920080 L2-loads # 49.870 M/sec 422522434 L2-load-hits # 99.91 of all L2 cache hits 6906 Context Switches # 0.001 M/sec 15 CPU-migrations # 0.000 M/sec 372 Page Faults # 0.000 M/sec 8480.512512 Task Clock # 0.784 CPUs utilized 10.812143770 seconds time elapsed root@du1:~# ./rbs-perf-test -t 3 -s -c ./ltt-test-2.2.0 -l 1000000 -t 1 --- Shadow TLB misses - L2 Miss Eviction - exec cmd - ./ltt-test-2.2.0 --- Benchmark test starts now... Time it took to loop 1000000 times : 8 s 621587970 ns | [8621 ns per tp] Stats for pid 2782 Counters enabled in user and in kernel space ------------------------------------------------------------------ 6682376594 Instruction count # 0.50 Instructions per Cycle 13404210872 CPU cycles # 1.593 GHz 117040641 dcache shadow TLB misses # 13.906 M/sec 60244653 icache shadow TLB misses # 7.158 M/sec 222672 L2 Miss Eviction # 58.50 of all L2 cache misses 3934 Context Switches # 0.000 M/sec 12 CPU-migrations # 0.000 M/sec 370 Page Faults # 0.000 M/sec 8416.596896 Task Clock # 0.789 CPUs utilized 10.661324180 seconds time elapsed root@du1:~# ./rbs-perf-test -t 4 -s -c ./ltt-test-2.2.0 -l 1000000 -t 1 --- PLB6 Commands - exec cmd - ./ltt-test-2.2.0 --- Benchmark test starts now... Time it took to loop 1000000 times : 8 s 768384390 ns | [8768 ns per tp] Stats for pid 2797 Counters enabled in user and in kernel space ------------------------------------------------------------------ 11252534 PLB6 Commands # 1.335 M/sec 363519 READ/READ-ATOMIC commands # 0.043 M/sec 30251 RWITM commands # 0.004 M/sec 248485 DCLAIM commands # 0.029 M/sec 32418 Cache line WRITE commands # 0.004 M/sec 30128 Cache line fill IntvnM # 0.004 M/sec 159751 Cache line fill IntvnS # 0.019 M/sec 191845 Cache line fill memory # 0.023 M/sec 8081 Context Switches # 0.001 M/sec 6 CPU-migrations # 0.000 M/sec 372 Page Faults # 0.000 M/sec 8427.167536 Task Clock # 0.639 CPUs utilized 13.190474860 seconds time elapsed