[Mesa-dev] [PATCH] glsl: optimize list handling in opt_dead_code

[Jan Ziak]

Perf stat results for shader-db:

This is measured on an AMD Kaveri CPU.

gcc-6.2.0 -fno-omit-frame-pointer -g -O2

**** Unpatched:

$ cd shader-db
$ ../run-upstream perfstat-u --repeat=5 -- ./run -1 shaders >/dev/null

 Performance counter stats for './run -1 shaders' (5 runs):

      13689.962374      task-clock (msec)         #    1.000 CPUs utilized
           ( +-  0.29% )
               138      context-switches          #    0.010 K/sec
           ( +- 17.82% )
                 6      cpu-migrations            #    0.000 K/sec
           ( +- 13.36% )
            78,559      page-faults               #    0.006 M/sec
           ( +-  0.24% )
    53,578,642,861      cycles:u                  #    3.914 GHz
           ( +-  0.29% )
    44,813,859,985      instructions:u            #    0.84  insn per cycle
          ( +-  0.01% )
     1,069,586,875      cache-references:u        #   78.129 M/sec
           ( +-  0.65% )
        51,295,256      cache-misses:u            #    4.796 % of all cache
refs      ( +-  0.56% )
     9,508,996,305      branches:u                #  694.596 M/sec
           ( +-  0.01% )
       453,237,236      branch-misses:u           #    4.77% of all
branches          ( +-  0.84% )

      13.692494394 seconds time elapsed
     ( +-  0.29% )

**** Patched:

$ cd shader-db
$ ../run-upstream-patched perfstat-u --repeat=5 -- ./run -1 shaders
>/dev/null

 Performance counter stats for './run -1 shaders' (5 runs):

      13602.106171      task-clock (msec)         #    1.000 CPUs utilized
           ( +-  0.14% )
                86      context-switches          #    0.006 K/sec
           ( +- 13.95% )
                 6      cpu-migrations            #    0.000 K/sec
           ( +- 26.35% )
            78,271      page-faults               #    0.006 M/sec
           ( +-  0.82% )
    53,299,046,681      cycles:u                  #    3.918 GHz
           ( +-  0.13% )
    44,577,707,063      instructions:u            #    0.84  insn per cycle
          ( +-  0.01% )
     1,078,158,307      cache-references:u        #   79.264 M/sec
           ( +-  0.70% )
        51,521,287      cache-misses:u            #    4.779 % of all cache
refs      ( +-  1.03% )
     9,459,962,609      branches:u                #  695.478 M/sec
           ( +-  0.01% )
       456,593,871      branch-misses:u           #    4.83% of all
branches          ( +-  0.27% )

      13.603795247 seconds time elapsed
     ( +-  0.14% )
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20161018/70d8bb73/attachment.html>