mycpp Code Generation

Measure the speedup from mycpp, and the resource usage.

Source code: oil/mycpp/examples

User Time (milliseconds)

Lower ratios are better.

example name gen C++ Python C++ : Python
classes gen 0 22 0.000
fib_iter gen 4 899 0.004
modules gen 2 186 0.010
fib_recursive gen 11 945 0.011
loops gen 4 292 0.012
asdl_generated gen 7 388 0.019
parse gen 25 761 0.033
scoped_resource gen 47 1,048 0.045
containers gen 8 109 0.071
files gen 7 72 0.102
tuple_return_value gen 20 185 0.108
length gen 45 204 0.220
cartesian gen 88 324 0.272
escape gen 99 344 0.288
varargs gen 8 25 0.308
gc_stack_roots gen 2 4 0.451
cgi gen 261 528 0.494
control_flow gen 208 109 1.908

Max Resident Set Size (MB)

Lower ratios are better. We use MB (powers of 10), not MiB (powers of 2).

example name gen C++ Python C++ : Python
classes gen 4.5 10.7 0.41
parse gen 3.7 7.6 0.48
gc_stack_roots gen 3.5 6.9 0.51
cgi gen 3.7 7.1 0.52
asdl_generated gen 3.7 6.9 0.53
cartesian gen 3.7 6.9 0.53
fib_recursive gen 3.7 6.9 0.53
scoped_resource gen 3.7 6.9 0.53
escape gen 3.8 7.1 0.54
loops gen 3.8 7.1 0.54
tuple_return_value gen 3.8 7.1 0.54
fib_iter gen 3.8 6.9 0.55
files gen 3.8 6.9 0.55
modules gen 3.8 6.9 0.55
control_flow gen 3.9 7.1 0.56
length gen 3.9 6.9 0.57
containers gen 28.7 48.0 0.60
varargs gen 5.5 6.9 0.79

System Time (milliseconds)

Lower ratios are better.

example name gen C++ Python C++ : Python
control_flow gen 0 4 0.000
files gen 0 4 0.000
gc_stack_roots gen 0 8 0.000
length gen 0 4 0.000
loops gen 0 4 0.000
modules gen 0 4 0.000
scoped_resource gen 0 4 0.000
tuple_return_value gen 0 8 0.000
asdl_generated gen 4 12 0.299
containers gen 8 24 0.318
classes gen 3 9 0.369
cartesian gen 4 8 0.501
parse gen 4 8 0.522
fib_iter gen 4 4 0.939
varargs gen 58 46 1.262
cgi gen 8 4 1.962
escape gen 8 4 1.980
fib_recursive gen 0 0 NA

raw benchmark files