Always inline rb_to_integer to prevent a method call penalty for integer types

[methodmissing]

On going through a hardware event trace of optcarrot in Intel VTune Amplifier I noticed rb_to_int showing up in the Core Bound category:

I looked through the source in object.c and noticed it calls rb_to_integer which is also invoked from only one other callsite and noticed the early return on RB_INTEGER_TYPE_P for integer types (bignum specifically). On re-running optcarrot CPI (cycles per instruction) went down significantly and I got similar values out of various runs:

Inlining for this case does not bloat code size by much as the function is small and has only 2 callsites but does seem to improve CPI rate quite a bit.

optcarrot

lourens@CarbonX1:~/src/optcarrot$ benchmark-driver -e "inline-rb-to-integer::~/src/ruby/ruby/ruby -I~/src/ruby/ruby/lib -I~/src/ruby/ruby/. -I~/src/ruby/ruby/.ext/x86_64-linux" -e "trunk::~/src/ruby/trunk/ruby -I~/src/ruby/trunk/lib -I~/src/ruby/trunk/. -I~/src/ruby/trunk/.ext/x86_64-linux" --repeat-count 10 benchmark.yml
Calculating -------------------------------------
                     inline-rb-to-integer       trunk 
           optcarrot               49.232      45.364 fps

Comparison:
                        optcarrot
inline-rb-to-integer:        49.2 fps 
               trunk:        45.4 fps - 1.09x  slower

lourens@CarbonX1:~/src/optcarrot$ benchmark-driver -e "inline-rb-to-integer::~/src/ruby/ruby/ruby -I~/src/ruby/ruby/lib -I~/src/ruby/ruby/. -I~/src/ruby/ruby/.ext/x86_64-linux" -e "trunk::~/src/ruby/trunk/ruby -I~/src/ruby/trunk/lib -I~/src/ruby/trunk/. -I~/src/ruby/trunk/.ext/x86_64-linux" --repeat-count 10 benchmark.yml
Calculating -------------------------------------
                     inline-rb-to-integer       trunk 
           optcarrot               47.584      43.686 fps

Comparison:
                        optcarrot
inline-rb-to-integer:        47.6 fps 
               trunk:        43.7 fps - 1.09x  slower

lourens@CarbonX1:~/src/optcarrot$ benchmark-driver -e "inline-rb-to-integer::~/src/ruby/ruby/ruby -I~/src/ruby/ruby/lib -I~/src/ruby/ruby/. -I~/src/ruby/ruby/.ext/x86_64-linux" -e "trunk::~/src/ruby/trunk/ruby -I~/src/ruby/trunk/lib -I~/src/ruby/trunk/. -I~/src/ruby/trunk/.ext/x86_64-linux" --repeat-count 10 benchmark.yml
Calculating -------------------------------------
                     inline-rb-to-integer       trunk 
           optcarrot               45.710      42.930 fps

Comparison:
                        optcarrot
inline-rb-to-integer:        45.7 fps 
               trunk:        42.9 fps - 1.06x  slower

benchmarks

Some noise but I think they're expected for some.

lourens@CarbonX1:~/src/ruby/ruby$ make benchmark COMPARE_RUBY="~/src/ruby/trunk/ruby -I~/src/ruby/trunk/lib -I~/src/ruby/trunk"
/usr/bin/ruby --disable=gems -rrubygems -I./benchmark/lib ./benchmark/benchmark-driver/exe/benchmark-driver \
            --executables="compare-ruby::~/src/ruby/trunk/ruby -I~/src/ruby/trunk/lib -I~/src/ruby/trunk -I.ext/common --disable-gem" \
            --executables="built-ruby::./miniruby -I./lib -I. -I.ext/common  -r./prelude --disable-gem" \
            $(find ./benchmark -maxdepth 1 -name '**.yml' -o -name '**.rb' | sort) 
Calculating -------------------------------------
                               compare-ruby  built-ruby 
                    app_answer       45.809      47.424 i/s -       1.000 times in 0.021830s 0.021086s
                   app_aobench        0.030       0.015 i/s -       1.000 times in 33.401929s 65.442848s
                       app_erb       8.694k      8.992k i/s -     15.000k times in 1.725275s 1.668058s
                 app_factorial        1.525       1.605 i/s -       1.000 times in 0.655634s 0.623192s
                       app_fib        3.038       3.093 i/s -       1.000 times in 0.329200s 0.323330s
               app_lc_fizzbuzz        0.044       0.046 i/s -       1.000 times in 22.536389s 21.760657s
                app_mandelbrot        1.015       0.981 i/s -       1.000 times in 0.985123s 1.019837s
                 app_pentomino        0.094       0.094 i/s -       1.000 times in 10.692741s 10.587383s
                     app_raise        7.117       7.887 i/s -       1.000 times in 0.140503s 0.126787s
                 app_strconcat        2.770       2.750 i/s -       1.000 times in 0.360997s 0.363610s
                       app_tak        2.354       2.369 i/s -       1.000 times in 0.424736s 0.422176s
                     app_tarai        2.992       2.951 i/s -       1.000 times in 0.334267s 0.338848s
                       app_uri        2.659       2.675 i/s -       1.000 times in 0.376047s 0.373877s
        array_sample_100k__100       16.996      21.458 i/s -       1.000 times in 0.058838s 0.046602s
       array_sample_100k___10k        0.415       0.421 i/s -       1.000 times in 2.410055s 2.373961s
          array_sample_100k_10      150.050     127.263 i/s -       1.000 times in 0.006664s 0.007858s
          array_sample_100k_11       73.871      80.807 i/s -       1.000 times in 0.013537s 0.012375s
         array_sample_100k__1k        2.454       2.444 i/s -       1.000 times in 0.407573s 0.409243s
       array_sample_100k___50k        0.130       0.120 i/s -       1.000 times in 7.684671s 8.365602s
         array_sample_100k__6k        0.571       0.594 i/s -       1.000 times in 1.752663s 1.682788s
                   array_shift        0.567       0.551 i/s -       1.000 times in 1.762512s 1.814166s
               array_small_and       78.755      68.984 i/s -       1.000 times in 0.012698s 0.014496s
              array_small_diff       70.944      70.082 i/s -       1.000 times in 0.014096s 0.014269s
                array_small_or       53.359      51.713 i/s -       1.000 times in 0.018741s 0.019338s
              array_sort_block        0.202       0.194 i/s -       1.000 times in 4.949996s 5.163504s
              array_sort_float        0.672       0.603 i/s -       1.000 times in 1.487680s 1.659480s
           array_values_at_int      129.999     138.713 i/s -       1.000 times in 0.007692s 0.007209s
         array_values_at_range        4.602       4.557 i/s -       1.000 times in 0.217277s 0.219426s
                       bighash        0.572       0.595 i/s -       1.000 times in 1.746733s 1.680554s
                   dir_empty_p        2.987       3.216 i/s -       1.000 times in 0.334739s 0.310965s
          enum_lazy_grep_v_100        3.489       3.933 i/s -       1.000 times in 0.286580s 0.254274s
           enum_lazy_grep_v_20        6.701       6.839 i/s -       1.000 times in 0.149221s 0.146225s
           enum_lazy_grep_v_50        5.004       5.502 i/s -       1.000 times in 0.199854s 0.181759s
            enum_lazy_uniq_100        2.905       2.975 i/s -       1.000 times in 0.344248s 0.336167s
             enum_lazy_uniq_20        6.005       5.677 i/s -       1.000 times in 0.166531s 0.176153s
             enum_lazy_uniq_50        4.013       3.981 i/s -       1.000 times in 0.249200s 0.251221s
                    erb_render       1.134M      1.124M i/s -      1.500M times in 1.323141s 1.334440s
                    file_chmod        3.153       3.174 i/s -       1.000 times in 0.317199s 0.315099s
                   file_rename        0.885       0.890 i/s -       1.000 times in 1.130221s 1.123085s
           hash_aref_dsym_long        0.299       0.304 i/s -       1.000 times in 3.341323s 3.287371s
                hash_aref_dsym        3.645       3.503 i/s -       1.000 times in 0.274383s 0.285430s
                 hash_aref_fix        3.885       3.842 i/s -       1.000 times in 0.257382s 0.260273s
                 hash_aref_flo       28.924      34.869 i/s -       1.000 times in 0.034574s 0.028679s
                hash_aref_miss        2.799       2.587 i/s -       1.000 times in 0.357325s 0.386502s
                 hash_aref_str        3.012       3.038 i/s -       1.000 times in 0.332015s 0.329192s
            hash_aref_sym_long        2.522       2.317 i/s -       1.000 times in 0.396574s 0.431504s
                 hash_aref_sym        3.360       3.567 i/s -       1.000 times in 0.297597s 0.280347s
                  hash_flatten        5.986       5.841 i/s -       1.000 times in 0.167066s 0.171212s
                hash_ident_flo       31.692      28.871 i/s -       1.000 times in 0.031554s 0.034637s
                hash_ident_num        3.848       3.873 i/s -       1.000 times in 0.259888s 0.258215s
                hash_ident_obj        3.637       3.874 i/s -       1.000 times in 0.274946s 0.258124s
                hash_ident_str        3.889       3.747 i/s -       1.000 times in 0.257120s 0.266857s
                hash_ident_sym        3.646       3.359 i/s -       1.000 times in 0.274290s 0.297705s
                     hash_keys       10.735      11.148 i/s -       1.000 times in 0.093152s 0.089704s
                     hash_long        1.658       1.655 i/s -       1.000 times in 0.603284s 0.604170s
                    hash_shift      118.862     116.037 i/s -       1.000 times in 0.008413s 0.008618s
                hash_shift_u16       15.637      17.641 i/s -       1.000 times in 0.063949s 0.056687s
                hash_shift_u24       15.741      17.693 i/s -       1.000 times in 0.063529s 0.056520s
                hash_shift_u32       15.765      17.494 i/s -       1.000 times in 0.063432s 0.057163s
                   hash_small2        1.859       1.981 i/s -       1.000 times in 0.538045s 0.504818s
                   hash_small4        1.467       1.480 i/s -       1.000 times in 0.681605s 0.675663s
                   hash_small8        0.982       0.999 i/s -       1.000 times in 1.018450s 1.000523s
                  hash_to_proc      372.122     331.097 i/s -       1.000 times in 0.002687s 0.003020s
                   hash_values       10.653      12.222 i/s -       1.000 times in 0.093874s 0.081818s
                       int_quo        1.172       1.152 i/s -       1.000 times in 0.852893s 0.867791s
          io_copy_stream_write        5.732       6.194 i/s -       1.000 times in 0.174461s 0.161444s
   io_copy_stream_write_socket      571.512     382.024 i/s -       1.000 times in 0.001750s 0.002618s
                io_file_create        0.879       0.880 i/s -       1.000 times in 1.137430s 1.136432s
                  io_file_read        0.938       1.026 i/s -       1.000 times in 1.065788s 0.975030s
                 io_file_write        1.456       1.496 i/s -       1.000 times in 0.686973s 0.668469s
             io_nonblock_noex2        0.708       0.705 i/s -       1.000 times in 1.411532s 1.417809s
              io_nonblock_noex        0.615       0.610 i/s -       1.000 times in 1.626382s 1.639540s
                    io_pipe_rw        0.964       0.940 i/s -       1.000 times in 1.037290s 1.063496s
                    io_select2        0.508       0.549 i/s -       1.000 times in 1.970211s 1.822986s
                    io_select3       86.375      75.231 i/s -       1.000 times in 0.011577s 0.013292s
                     io_select        0.594       0.596 i/s -       1.000 times in 1.684734s 1.678993s
                      loop_for        0.986       0.951 i/s -       1.000 times in 1.014360s 1.051968s
                loop_generator        2.214       2.215 i/s -       1.000 times in 0.451772s 0.451390s
                    loop_times        1.067       1.065 i/s -       1.000 times in 0.937584s 0.939224s
               loop_whileloop2       11.961      12.075 i/s -       1.000 times in 0.083602s 0.082813s
                loop_whileloop        2.403       2.400 i/s -       1.000 times in 0.416113s 0.416651s
              marshal_dump_flo        4.774       4.830 i/s -       1.000 times in 0.209468s 0.207049s
       marshal_dump_load_geniv        3.311       3.330 i/s -       1.000 times in 0.302007s 0.300276s
        marshal_dump_load_time        1.515       1.556 i/s -       1.000 times in 0.660234s 0.642545s
                require_thread        0.020       0.022 i/s -       1.000 times in 48.951655s 44.677606s
                       require        1.081       1.425 i/s -       1.000 times in 0.925329s 0.701762s
                  securerandom        4.361       4.455 i/s -       1.000 times in 0.229288s 0.224489s
                  so_ackermann        1.886       1.958 i/s -       1.000 times in 0.530219s 0.510715s
                      so_array        1.076       1.079 i/s -       1.000 times in 0.929349s 0.926620s
               so_binary_trees        0.176       0.190 i/s -       1.000 times in 5.684107s 5.275014s
                so_concatenate        0.293       0.298 i/s -       1.000 times in 3.407739s 3.350192s
                so_count_words        5.840       5.873 i/s -       1.000 times in 0.171237s 0.170261s
                  so_exception        4.356       4.264 i/s -       1.000 times in 0.229543s 0.234545s
                   so_fannkuch        1.607       1.510 i/s -       1.000 times in 0.622432s 0.662223s
                      so_fasta        0.572       0.584 i/s -       1.000 times in 1.749299s 1.711525s
               so_k_nucleotidepreparing /tmp/fasta.output.100000
        1.011       1.051 i/s -       1.000 times in 0.988858s 0.951214s
                      so_lists        2.523       2.375 i/s -       1.000 times in 0.396295s 0.420972s
                 so_mandelbrot        0.515       0.536 i/s -       1.000 times in 1.940491s 1.865239s
                     so_matrix        2.037       2.228 i/s -       1.000 times in 0.490852s 0.448922s
             so_meteor_contest        0.380       0.397 i/s -       1.000 times in 2.633571s 2.519249s
                      so_nbody        0.935       0.920 i/s -       1.000 times in 1.069357s 1.086940s
                so_nested_loop        1.246       1.235 i/s -       1.000 times in 0.802465s 0.809854s
                so_nsieve_bits        0.630       0.632 i/s -       1.000 times in 1.588024s 1.583345s
                     so_nsieve        0.815       0.855 i/s -       1.000 times in 1.227633s 1.170055s
                     so_object        1.917       2.059 i/s -       1.000 times in 0.521755s 0.485698s
               so_partial_sums        0.735       0.733 i/s -       1.000 times in 1.359887s 1.364217s
                   so_pidigits        1.366       1.376 i/s -       1.000 times in 0.732164s 0.726735s
                     so_random        2.481       2.422 i/s -       1.000 times in 0.403125s 0.412854s
         so_reverse_complementpreparing /tmp/fasta.output.2500000
        0.930       0.906 i/s -       1.000 times in 1.075443s 1.103970s
                      so_sieve        2.781       2.753 i/s -       1.000 times in 0.359644s 0.363285s
               so_spectralnorm        0.750       0.739 i/s -       1.000 times in 1.333871s 1.354055s
                  string_index        3.279       2.944 i/s -       1.000 times in 0.304966s 0.339690s
                string_scan_re        6.190       5.883 i/s -       1.000 times in 0.161540s 0.169975s
               string_scan_str        9.493       9.042 i/s -       1.000 times in 0.105339s 0.110599s
                   time_subsec        1.047       1.022 i/s -       1.000 times in 0.955096s 0.978530s
             vm1_attr_ivar_set      42.215M     43.834M i/s -     30.000M times in 0.710645s 0.684393s
                 vm1_attr_ivar      59.357M     56.732M i/s -     30.000M times in 0.505419s 0.528803s
           vm1_blockparam_call      20.041M     19.999M i/s -     30.000M times in 1.496923s 1.500071s
           vm1_blockparam_pass      15.626M     15.740M i/s -     30.000M times in 1.919817s 1.906016s
          vm1_blockparam_yield      23.030M     22.101M i/s -     30.000M times in 1.302659s 1.357430s
                vm1_blockparam      30.343M     31.247M i/s -     30.000M times in 0.988711s 0.960085s
                     vm1_block      30.384M     32.123M i/s -     30.000M times in 0.987352s 0.933916s
                     vm1_const     157.440M    157.258M i/s -     30.000M times in 0.190549s 0.190769s
                    vm1_ensure        2.477       2.439 i/s -       1.000 times in 0.403774s 0.410039s
              vm1_float_simple      15.046M     14.547M i/s -     30.000M times in 1.993948s 2.062275s
            vm1_gc_short_lived       7.772M      7.771M i/s -     30.000M times in 3.859791s 3.860575s
vm1_gc_short_with_complex_long       1.024M      5.381M i/s -     30.000M times in 29.302920s 5.575261s
        vm1_gc_short_with_long       5.587M      5.706M i/s -     30.000M times in 5.369505s 5.258017s
      vm1_gc_short_with_symbol       6.651M      7.002M i/s -     30.000M times in 4.510498s 4.284321s
        vm1_gc_wb_ary_promoted      54.314M     56.756M i/s -     30.000M times in 0.552347s 0.528580s
                 vm1_gc_wb_ary      58.077M     56.551M i/s -     30.000M times in 0.516559s 0.530499s
        vm1_gc_wb_obj_promoted      70.633M     78.933M i/s -     30.000M times in 0.424731s 0.380069s
                 vm1_gc_wb_obj      76.639M     85.005M i/s -     30.000M times in 0.391447s 0.352920s
                  vm1_ivar_set     152.398M    150.701M i/s -     30.000M times in 0.196853s 0.199069s
                      vm1_ivar     145.601M    131.962M i/s -     30.000M times in 0.206043s 0.227338s
                    vm1_length      96.001M    103.800M i/s -     30.000M times in 0.312496s 0.289018s
                 vm1_lvar_init        0.745       0.756 i/s -       1.000 times in 1.342239s 1.322386s
                  vm1_lvar_set      17.827M     17.086M i/s -     30.000M times in 1.682847s 1.755825s
                       vm1_neq      69.489M     72.555M i/s -     30.000M times in 0.431721s 0.413480s
                       vm1_not     192.969M    192.476M i/s -     30.000M times in 0.155465s 0.155864s
                    vm1_rescue     704.467M    702.235M i/s -     30.000M times in 0.042585s 0.042721s
              vm1_simplereturn      79.282M     78.489M i/s -     30.000M times in 0.378398s 0.382221s
                      vm1_swap     152.454M    148.295M i/s -     30.000M times in 0.196781s 0.202300s
                     vm1_yield        1.127       1.109 i/s -       1.000 times in 0.887165s 0.901839s
                     vm2_array      29.059M     28.809M i/s -      6.000M times in 0.206476s 0.208265s
                  vm2_bigarray       1.887M      1.715M i/s -      6.000M times in 3.179650s 3.498251s
                   vm2_bighash     131.382k    121.989k i/s -     60.000k times in 0.456683s 0.491847s
                  vm2_case_lit        2.092       2.137 i/s -       1.000 times in 0.477961s 0.467936s
                      vm2_case      77.211M     85.770M i/s -      6.000M times in 0.077709s 0.069954s
            vm2_defined_method       2.647M      2.729M i/s -      6.000M times in 2.266994s 2.198555s
                      vm2_dstr       7.568M      7.507M i/s -      6.000M times in 0.792848s 0.799207s
                      vm2_eval     382.740k    130.762k i/s -      6.000M times in 15.676439s 45.884974s
              vm2_fiber_switch     988.682k      1.063M i/s -      6.000M times in 6.068686s 5.642898s
              vm2_freezestring       8.092M      7.980M i/s -      6.000M times in 0.741498s 0.751835s
            vm2_method_missing       2.438M      2.518M i/s -      6.000M times in 2.460998s 2.382646s
         vm2_method_with_block       6.234M      6.485M i/s -      6.000M times in 0.962435s 0.925207s
                    vm2_method       7.435M      7.665M i/s -      6.000M times in 0.807036s 0.782742s
      vm2_module_ann_const_set       1.306M      1.396M i/s -      6.000M times in 4.593104s 4.297203s
          vm2_module_const_set       1.386M      1.397M i/s -      6.000M times in 4.327854s 4.296172s
                     vm2_mutex      12.286M     10.523M i/s -      6.000M times in 0.488341s 0.570205s
                 vm2_newlambda      10.858M     11.448M i/s -      6.000M times in 0.552579s 0.524098s
            vm2_poly_method_ov        3.839       3.847 i/s -       1.000 times in 0.260510s 0.259929s
               vm2_poly_method        0.464       0.466 i/s -       1.000 times in 2.155434s 2.147868s
            vm2_poly_singleton        1.032       1.020 i/s -       1.000 times in 0.968624s 0.980597s
                      vm2_proc      35.662M     36.738M i/s -      6.000M times in 0.168244s 0.163317s
                    vm2_raise1       1.879M      1.872M i/s -      6.000M times in 3.192521s 3.205435s
                    vm2_raise2       1.140M      1.201M i/s -      6.000M times in 5.262587s 4.994649s
                    vm2_regexp       7.523M      7.526M i/s -      6.000M times in 0.797524s 0.797240s
                      vm2_send      23.996M     23.441M i/s -      6.000M times in 0.250039s 0.255966s
            vm2_string_literal      47.660M     47.091M i/s -      6.000M times in 0.125893s 0.127413s
        vm2_struct_big_aref_hi      43.924M     52.838M i/s -      6.000M times in 0.136598s 0.113555s
        vm2_struct_big_aref_lo      53.784M     47.241M i/s -      6.000M times in 0.111558s 0.127009s
           vm2_struct_big_aset        4.495       4.560 i/s -       1.000 times in 0.222490s 0.219316s
        vm2_struct_big_href_hi      29.434M     30.270M i/s -      6.000M times in 0.203849s 0.198217s
        vm2_struct_big_href_lo      29.089M     30.707M i/s -      6.000M times in 0.206263s 0.195398s
           vm2_struct_big_hset        3.203       3.086 i/s -       1.000 times in 0.312198s 0.324082s
         vm2_struct_small_aref      80.776M     71.286M i/s -      6.000M times in 0.074279s 0.084168s
         vm2_struct_small_aset        4.863       4.735 i/s -       1.000 times in 0.205629s 0.211180s
         vm2_struct_small_href      36.659M     37.172M i/s -      6.000M times in 0.163671s 0.161411s
         vm2_struct_small_hset      33.299M     32.893M i/s -      6.000M times in 0.180188s 0.182412s
                     vm2_super      20.390M     20.468M i/s -      6.000M times in 0.294265s 0.293140s
                     vm2_unif1      66.830M     64.008M i/s -      6.000M times in 0.089780s 0.093738s
                    vm2_zsuper      19.626M     19.127M i/s -      6.000M times in 0.305721s 0.313694s
                 vm3_backtrace        9.753       9.894 i/s -       1.000 times in 0.102536s 0.101075s
          vm3_clearmethodcache        4.985       5.586 i/s -       1.000 times in 0.200602s 0.179004s
               vm3_gc_old_full        0.373       0.367 i/s -       1.000 times in 2.682735s 2.723101s
          vm3_gc_old_immediate        0.560       0.496 i/s -       1.000 times in 1.786821s 2.015963s
               vm3_gc_old_lazy        0.397       0.391 i/s -       1.000 times in 2.519232s 2.556574s
                        vm3_gc        0.812       0.818 i/s -       1.000 times in 1.231078s 1.223168s
          vm_symbol_block_pass        1.260       1.283 i/s -       1.000 times in 0.793460s 0.779474s
        vm_thread_alive_check1       19.222      22.667 i/s -       1.000 times in 0.052022s 0.044117s
               vm_thread_close        1.881       1.911 i/s -       1.000 times in 0.531724s 0.523305s
            vm_thread_condvar1        1.472       1.440 i/s -       1.000 times in 0.679352s 0.694618s
            vm_thread_condvar2        1.569       1.865 i/s -       1.000 times in 0.637233s 0.536216s
         vm_thread_create_join        1.049       1.123 i/s -       1.000 times in 0.953475s 0.890818s
              vm_thread_mutex1        2.547       2.584 i/s -       1.000 times in 0.392638s 0.386976s
              vm_thread_mutex2        2.629       2.495 i/s -       1.000 times in 0.380407s 0.400727s
              vm_thread_mutex3        1.055       0.874 i/s -       1.000 times in 0.947615s 1.144158s
          vm_thread_pass_flood       19.225      20.829 i/s -       1.000 times in 0.052015s 0.048009s
                vm_thread_pass        2.994       3.211 i/s -       1.000 times in 0.333969s 0.311461s
                vm_thread_pipe        4.506       6.539 i/s -       1.000 times in 0.221919s 0.152933s
               vm_thread_queue       12.701      12.844 i/s -       1.000 times in 0.078732s 0.077859s
        vm_thread_sized_queue2        1.838       1.692 i/s -       1.000 times in 0.544110s 0.591028s
        vm_thread_sized_queue3        1.727       1.774 i/s -       1.000 times in 0.578993s 0.563790s
        vm_thread_sized_queue4        2.817       3.240 i/s -       1.000 times in 0.354992s 0.308627s
         vm_thread_sized_queue        6.463       5.946 i/s -       1.000 times in 0.154718s 0.168182s

[k0kubun]

Cool!

optcarrot

What's the architecture of your machine (I recommend to always use -v of benchmark-driver command when you share results)? I failed to reproduce such a big difference. Compiler's version would be also helpful (I'm using gcc 7.3.0).

$ benchmark-driver benchmark.yml --rbenv 'before::before --disable-gems;after::after --disable-gems' -v --repeat-count 24
before: ruby 2.6.0dev (2018-11-03 trunk 65510) [x86_64-linux]
after: ruby 2.6.0dev (2018-11-03 trunk 65510) [x86_64-linux]
last_commit=Always inline rb_to_integer to prevent a method call penalty
Calculating -------------------------------------
                             before       after
Optcarrot Lan_Master.nes     54.515      54.851 fps

Comparison:
             Optcarrot Lan_Master.nes
                   after:        54.9 fps
                  before:        54.5 fps - 1.01x  slower

[methodmissing]

Thanks for the feedback (and benchmark-driver :-) ) - forgot about --disable-gems as well

cpu arch is 4 core i7-8650U:

lourens@CarbonX1:~/asan$ cat /proc/cpuinfo
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 142
model name	: Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz
stepping	: 10
microcode	: 0x96
cpu MHz		: 3200.015
cache size	: 8192 KB

gcc 7.3, with CFLAGS -g -O3:

lourens@CarbonX1:~/src/ruby/trunk$ gcc --version
gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0

Results less skewed and inline with yours, thanks:

lourens@CarbonX1:~/src/optcarrot$ benchmark-driver -e "inline-rb-to-integer::~/src/ruby/ruby/ruby -I~/src/ruby/ruby/lib -I~/src/ruby/ruby/. -I~/src/ruby/ruby/.ext/x86_64-linux --disable-gems" -e "trunk::~/src/ruby/trunk/ruby -I~/src/ruby/trunk/lib -I~/src/ruby/trunk/. -I~/src/ruby/trunk/.ext/x86_64-linux --disable-gems" -v --repeat-count 24 benchmark.yml
inline-rb-to-integer: ruby 2.6.0dev (2018-11-03 inline-rb-to-i.. 65513) [x86_64-linux]
last_commit=Always inline rb_to_integer to prevent a method call penalty for integer types
trunk: ruby 2.6.0dev (2018-11-03 trunk 65513) [x86_64-linux]
Calculating -------------------------------------
                     inline-rb-to-integer       trunk 
           optcarrot               49.394      48.343 fps

Comparison:
                        optcarrot
inline-rb-to-integer:        49.4 fps 
               trunk:        48.3 fps - 1.02x  slower

[k0kubun]

Now that's what I see on my environment, and it does improve the performance as such. As long as you see that in the Core Bound category, I trust that this is worth merging. 38caab2

Thanks!