verilator_performance.txt

base commit 6882f8c55e:

➜  XiangShan git:(master) ✗ numactl --physcpubind=0-7,16-23 perf stat -e L1-dcache-load-misses -e L1-icache-load-misses -e branch-misses -e branch-load-misses -e cpu-cycles -e instructions -e stalled-cycles-frontend -e stalled-cycles-backend -e branch-instructions sh -c "./build/emu --no-diff -i /nfs/home/chenyangyu/xs-env/XiangShan/ready-to-run/coremark-2-iteration.bin 2>/dev/null"
emu compiled at Jul 11 2024, 18:09:34
Using simulated 32768B flash
Using simulated 8192MB RAM
The image is /nfs/home/chenyangyu/xs-env/XiangShan/ready-to-run/coremark-2-iteration.bin
Running CoreMark for 2 iterations
2K performance run parameters for coremark.
CoreMark Size    : 666
Total time (ms)  : 3161
Iterations       : 2
Compiler version : GCC10.2.0
seedcrc          : 0xe9f5
[0]crclist       : 0xe714
[0]crcmatrix     : 0x1fd7
[0]crcstate      : 0x8e3a
[0]crcfinal      : 0x72be
Finised in 3161 ms.
==================================================
CoreMark Iterations/Sec 632
Core 0: HIT GOOD TRAP at pc = 0x800026ba
Core-0 instrCnt = 626,872, cycleCnt = 367,843, IPC = 1.704184
Seed=0 Guest cycle spent: 367,847 (this will be different from cycleCnt if emu loads a snapshot)
Host time spent: 538,357ms

 Performance counter stats for 'sh -c ./build/emu --no-diff -i /nfs/home/chenyangyu/xs-env/XiangShan/ready-to-run/coremark-2-iteration.bin 2>/dev/null':

    68,711,043,143      L1-dcache-load-misses                                                   (55.56%)
     8,428,518,084      L1-icache-load-misses                                                   (55.56%)
    76,144,179,278      branch-misses                    #   58.21% of all branches             (55.56%)
    76,141,639,930      branch-load-misses                                                      (55.56%)
 2,781,783,140,530      cpu-cycles                                                              (55.56%)
 1,639,902,151,274      instructions                     #    0.59  insn per cycle
                                                  #    0.15  stalled cycles per insn     (55.55%)
   149,185,033,052      stalled-cycles-frontend          #    5.36% frontend cycles idle        (55.56%)
   247,403,041,208      stalled-cycles-backend           #    8.89% backend cycles idle         (55.56%)
   130,799,759,571      branch-instructions                                                     (55.56%)

     538.449886606 seconds time elapsed

     537.950798000 seconds user
       0.095995000 seconds sys

Using PR #5254:

➜  XiangShan git:(master) ✗ numactl --physcpubind=0-7,16-23 perf stat -e L1-dcache-load-misses -e L1-icache-load-misses -e branch-misses -e branch-load-misses -e cpu-cycles -e instructions -e stalled-cycles-frontend -e stalled-cycles-backend -e branch-instructions sh -c "./build/emu --no-diff -i /nfs/home/chenyangyu/xs-env/XiangShan/ready-to-run/coremark-2-iteration.bin 2>/dev/null"
emu compiled at Jul 11 2024, 18:09:34
Using simulated 32768B flash
Using simulated 8192MB RAM
The image is /nfs/home/chenyangyu/xs-env/XiangShan/ready-to-run/coremark-2-iteration.bin
Running CoreMark for 2 iterations
2K performance run parameters for coremark.
CoreMark Size    : 666
Total time (ms)  : 3161
Iterations       : 2
Compiler version : GCC10.2.0
seedcrc          : 0xe9f5
[0]crclist       : 0xe714
[0]crcmatrix     : 0x1fd7
[0]crcstate      : 0x8e3a
[0]crcfinal      : 0x72be
Finised in 3161 ms.
==================================================
CoreMark Iterations/Sec 632
Core 0: HIT GOOD TRAP at pc = 0x800026ba
Core-0 instrCnt = 626,872, cycleCnt = 367,843, IPC = 1.704184
Seed=0 Guest cycle spent: 367,847 (this will be different from cycleCnt if emu loads a snapshot)
Host time spent: 522,615ms

 Performance counter stats for 'sh -c ./build/emu --no-diff -i /nfs/home/chenyangyu/xs-env/XiangShan/ready-to-run/coremark-2-iteration.bin 2>/dev/null':

    68,974,376,546      L1-dcache-load-misses                                                   (55.56%)
     8,139,884,116      L1-icache-load-misses                                                   (55.56%)
    75,480,376,632      branch-misses                    #   65.96% of all branches             (55.56%)
    75,479,910,127      branch-load-misses                                                      (55.56%)
 2,710,532,253,408      cpu-cycles                                                              (55.55%)
 1,652,333,385,051      instructions                     #    0.61  insn per cycle
                                                  #    0.15  stalled cycles per insn     (55.55%)
   147,980,754,624      stalled-cycles-frontend          #    5.46% frontend cycles idle        (55.55%)
   252,459,154,223      stalled-cycles-backend           #    9.31% backend cycles idle         (55.56%)
   114,434,414,249      branch-instructions                                                     (55.56%)

     522.700185759 seconds time elapsed

     522.262359000 seconds user
       0.099995000 seconds sys