variability_measure => 'mad',
# no. of "sigma"s for the outlier rejection
- outlier_rejection => 2,
+ outlier_rejection => 1,
# automatically determined at runtime to not run
# longer than $max_bench_duration seconds
# our local addition to Dumbbench
max_bench_duration => 20,
gettime_clock_id => Time::HiRes::CLOCK_PROCESS_CPUTIME_ID(),
- code_subiterations => 250,
+ code_subiterations => 200,
};
my $acc_name = 'accessor';
|,
},
- CAG_S_XS => {
- provider => 'Class::Accessor::Grouped',
- add_isa => 1,
- env => {
- '$Class::Accessor::Grouped::USE_XS' => 1,
- },
- install => qq|
- __PACKAGE__->mk_group_accessors (simple => $q_acc_name);
- |,
- },
-
CAG_INH => {
provider => 'Class::Accessor::Grouped',
add_isa => 1,
type => 'mk_accessors'
},
- CAF_XSAC => {
- provider => 'Class::XSAccessor::Compat',
- type => 'mk_accessors'
- },
-
- XSA => {
- provider => 'Class::XSAccessor',
- install => qq|
- Class::XSAccessor->import({
- accessors => [ $q_acc_name ]
- }),
- |,
- },
-
HANDMADE => {
install => qq|
sub $acc_name {
type => 'mooselike',
},
- moo_XS => {
- provider => 'Moo',
- env => {
- '$Method::Generate::Accessor::CAN_HAZ_XS' => 1,
- },
- type => 'mooselike',
- },
-
moo => {
provider => 'Moo',
env => {
provider => 'Mo',
type => 'mooselike',
},
+
+ XSA => {
+ provider => 'Class::XSAccessor',
+ install => qq|
+ Class::XSAccessor->import({
+ accessors => [ $q_acc_name ]
+ }),
+ |,
+ },
+
+# # all of the things below should be identical
+# # to XSA, as they are essentially the same code
+# # yet I can't get the nu,bers to agree <sratchhead>
+# CAF_XSAC => {
+# provider => 'Class::XSAccessor::Compat',
+# type => 'mk_accessors'
+# },
+#
+# CAG_S_XS => {
+# provider => 'Class::Accessor::Grouped',
+# add_isa => 1,
+# env => {
+# '$Class::Accessor::Grouped::USE_XS' => 1,
+# },
+# install => qq|
+# __PACKAGE__->mk_group_accessors (simple => $q_acc_name);
+# |,
+# },
+#
+# moo_XS => {
+# provider => 'Moo',
+# env => {
+# '$Method::Generate::Accessor::CAN_HAZ_XS' => 1,
+# },
+# type => 'mooselike',
+# },
};
=head2 Benchmark
-This is the result of a set/get/set loop benchmark on perl 5.12.1 with
-thread support, showcasing most popular accessor builders: L<Moose>, L<Mouse>,
-L<Moo>, L<CAF|Class::Accessor::Fast>, L<CAF_XS|Class::Accessor::Fast::XS>,
-L<XSA|Class::XSAccessor>, and L<CAF_XSA|Class::XSAccessor::Compat>:
-
- Rate CAG moOse CAF moUse moo HANDMADE CAF_XS moUse_XS moo_XS CAF_XSA XSA CAG_XS
- CAG 169/s -- -21% -24% -32% -32% -34% -59% -63% -67% -67% -67% -67%
- moOse 215/s 27% -- -3% -13% -13% -15% -48% -53% -58% -58% -58% -58%
- CAF 222/s 31% 3% -- -10% -10% -13% -46% -52% -57% -57% -57% -57%
- moUse 248/s 46% 15% 11% -- -0% -3% -40% -46% -52% -52% -52% -52%
- moo 248/s 46% 15% 11% 0% -- -3% -40% -46% -52% -52% -52% -52%
- HANDMADE 255/s 50% 18% 14% 3% 3% -- -38% -45% -50% -51% -51% -51%
- CAF_XS 411/s 143% 91% 85% 66% 66% 61% -- -11% -20% -20% -21% -21%
- moUse_XS 461/s 172% 114% 107% 86% 86% 81% 12% -- -10% -11% -11% -11%
- moo_XS 514/s 204% 139% 131% 107% 107% 102% 25% 12% -- -0% -1% -1%
- CAF_XSA 516/s 205% 140% 132% 108% 108% 103% 26% 12% 0% -- -0% -0%
- XSA 519/s 206% 141% 133% 109% 109% 104% 26% 13% 1% 0% -- -0%
- CAG_XS 519/s 206% 141% 133% 109% 109% 104% 26% 13% 1% 0% 0% --
-
-Benchmark program is available in the root of the
+This is the benchmark of 200 get/get/set/get/set cycles on perl 5.16.2 with
+thread support, showcasing how this modules L<simple (CAG_S)|/get_simple>,
+L<inherited (CAG_INH)|/get_inherited> and L<inherited with parent-class data
+(CAG_INHP)|/get_inherited> accessors stack up against most popular accessor
+builders: L<Moose>, L<Moo>, L<Mo>, L<Mouse> (both pure-perl and XS variant),
+L<Object::Tiny::RW (OTRW)|Object::Tiny::RW>,
+L<Class::Accessor (CA)|Class::Accessor>,
+L<Class::Accessor::Lite (CAL)|Class::Accessor::Lite>,
+L<Class::Accessor::Fast (CAF)|Class::Accessor::Fast>,
+L<Class::Accessor::Fast::XS (CAF_XS)|Class::Accessor::Fast::XS>
+and L<Class::XSAccessor (XSA)|Class::XSAccessor>
+
+ Rate CAG_INHP CAG_INH CA CAG_S CAF moOse OTRW CAL mo moUse HANDMADE moo CAF_XS moUse_XS XSA
+
+ CAG_INHP 287.021+-0.02/s -- -0.3% -10.0% -37.1% -53.1% -53.6% -53.7% -54.1% -56.9% -59.0% -59.6% -59.8% -78.7% -81.9% -83.5%
+
+ CAG_INH 288.025+-0.031/s 0.3% -- -9.7% -36.9% -52.9% -53.5% -53.5% -53.9% -56.7% -58.8% -59.5% -59.7% -78.6% -81.9% -83.5%
+
+ CA 318.967+-0.047/s 11.1% 10.7% -- -30.1% -47.9% -48.5% -48.5% -49.0% -52.1% -54.4% -55.1% -55.3% -76.3% -79.9% -81.7%
+
+ CAG_S 456.107+-0.054/s 58.9% 58.4% 43.0% -- -25.4% -26.3% -26.4% -27.0% -31.5% -34.8% -35.8% -36.1% -66.1% -71.3% -73.9%
+
+ CAF 611.745+-0.099/s 113.1% 112.4% 91.8% 34.1% -- -1.2% -1.2% -2.1% -8.1% -12.6% -14.0% -14.3% -54.5% -61.5% -64.9%
+
+ moOse 619.051+-0.059/s 115.7% 114.9% 94.1% 35.7% 1.2% -- -0.1% -1.0% -7.0% -11.6% -12.9% -13.3% -54.0% -61.0% -64.5%
+
+ OTRW 619.475+-0.1/s 115.8% 115.1% 94.2% 35.8% 1.3% 0.1% -- -0.9% -6.9% -11.5% -12.9% -13.2% -54.0% -61.0% -64.5%
+
+ CAL 625.106+-0.085/s 117.8% 117.0% 96.0% 37.1% 2.2% 1.0% 0.9% -- -6.1% -10.7% -12.1% -12.5% -53.5% -60.6% -64.2%
+
+ mo 665.44+-0.12/s 131.8% 131.0% 108.6% 45.9% 8.8% 7.5% 7.4% 6.5% -- -4.9% -6.4% -6.8% -50.5% -58.1% -61.9%
+
+ moUse 699.9+-0.15/s 143.9% 143.0% 119.4% 53.5% 14.4% 13.1% 13.0% 12.0% 5.2% -- -1.6% -2.0% -48.0% -55.9% -59.9%
+
+ HANDMADE 710.98+-0.16/s 147.7% 146.8% 122.9% 55.9% 16.2% 14.9% 14.8% 13.7% 6.8% 1.6% -- -0.4% -47.2% -55.2% -59.2%
+
+ moo 714.04+-0.13/s 148.8% 147.9% 123.9% 56.6% 16.7% 15.3% 15.3% 14.2% 7.3% 2.0% 0.4% -- -46.9% -55.0% -59.1%
+
+ CAF_XS 1345.55+-0.051/s 368.8% 367.2% 321.8% 195.0% 120.0% 117.4% 117.2% 115.3% 102.2% 92.2% 89.3% 88.4% -- -15.3% -22.9%
+
+ moUse_XS 1588+-0.036/s 453.3% 451.3% 397.9% 248.2% 159.6% 156.5% 156.3% 154.0% 138.6% 126.9% 123.4% 122.4% 18.0% -- -9.0%
+
+ XSA 1744.67+-0.052/s 507.9% 505.7% 447.0% 282.5% 185.2% 181.8% 181.6% 179.1% 162.2% 149.3% 145.4% 144.3% 29.7% 9.9% --
+
+Benchmarking program is available in the root of the
L<repository|http://search.cpan.org/dist/Class-Accessor-Grouped/>:
=head2 Notes on Class::XSAccessor
projects / p5sagit/Class-Accessor-Grouped.git / commitdiff